repository_name
stringlengths
5
67
func_path_in_repository
stringlengths
4
234
func_name
stringlengths
0
314
whole_func_string
stringlengths
52
3.87M
language
stringclasses
6 values
func_code_string
stringlengths
39
1.84M
func_code_tokens
sequencelengths
15
672k
func_documentation_string
stringlengths
1
47.2k
func_documentation_tokens
sequencelengths
1
3.92k
split_name
stringclasses
1 value
func_code_url
stringlengths
85
339
eyecatchup/SEOstats
SEOstats/Services/SemRush.php
SemRush.getDomainRank
public static function getDomainRank($url = false, $db = false) { $data = self::getBackendData($url, $db, 'domain_rank'); return is_array($data) ? $data['rank']['data'][0] : $data; }
php
public static function getDomainRank($url = false, $db = false) { $data = self::getBackendData($url, $db, 'domain_rank'); return is_array($data) ? $data['rank']['data'][0] : $data; }
[ "public", "static", "function", "getDomainRank", "(", "$", "url", "=", "false", ",", "$", "db", "=", "false", ")", "{", "$", "data", "=", "self", "::", "getBackendData", "(", "$", "url", ",", "$", "db", ",", "'domain_rank'", ")", ";", "return", "is_array", "(", "$", "data", ")", "?", "$", "data", "[", "'rank'", "]", "[", "'data'", "]", "[", "0", "]", ":", "$", "data", ";", "}" ]
Returns the SEMRush main report data. (Only main report is public available.) @access public @param url string Domain name only, eg. "ebay.com" (/wo quotes). @param db string Optional: The database to use. Valid values are: au, br, ca, de, es, fr, it, ru, uk, us, us.bing (us is default) @return array Returns an array containing the main report data. @link http://www.semrush.com/api.html
[ "Returns", "the", "SEMRush", "main", "report", "data", ".", "(", "Only", "main", "report", "is", "public", "available", ".", ")" ]
train
https://github.com/eyecatchup/SEOstats/blob/bd3117f3ee10d6bf3c3cadd0461c15b91b15428b/SEOstats/Services/SemRush.php#L81-L86
RubixML/RubixML
src/Clusterers/GaussianMixture.php
GaussianMixture.priors
public function priors() : array { $priors = []; if (is_array($this->priors)) { $total = logsumexp($this->priors); foreach ($this->priors as $class => $probability) { $priors[$class] = exp($probability - $total); } } return $priors; }
php
public function priors() : array { $priors = []; if (is_array($this->priors)) { $total = logsumexp($this->priors); foreach ($this->priors as $class => $probability) { $priors[$class] = exp($probability - $total); } } return $priors; }
[ "public", "function", "priors", "(", ")", ":", "array", "{", "$", "priors", "=", "[", "]", ";", "if", "(", "is_array", "(", "$", "this", "->", "priors", ")", ")", "{", "$", "total", "=", "logsumexp", "(", "$", "this", "->", "priors", ")", ";", "foreach", "(", "$", "this", "->", "priors", "as", "$", "class", "=>", "$", "probability", ")", "{", "$", "priors", "[", "$", "class", "]", "=", "exp", "(", "$", "probability", "-", "$", "total", ")", ";", "}", "}", "return", "$", "priors", ";", "}" ]
Return the cluster prior probabilities. @return float[]
[ "Return", "the", "cluster", "prior", "probabilities", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Clusterers/GaussianMixture.php#L187-L200
RubixML/RubixML
src/Clusterers/GaussianMixture.php
GaussianMixture.train
public function train(Dataset $dataset) : void { DatasetIsCompatibleWithEstimator::check($dataset, $this); if ($this->logger) { $this->logger->info('Learner init ' . Params::stringify([ 'k' => $this->k, 'epochs' => $this->epochs, 'min_change' => $this->minChange, 'seeder' => $this->seeder, ])); } $samples = $dataset->samples(); $rotated = $dataset->columns(); $n = $dataset->numRows(); $this->priors = array_fill(0, $this->k, log(1. / $this->k)); [$this->means, $this->variances] = $this->initialize($dataset); $this->steps = []; $prevLoss = INF; for ($epoch = 1; $epoch <= $this->epochs; $epoch++) { $memberships = []; foreach ($dataset as $sample) { $jll = $this->jointLogLikelihood($sample); $memberships[] = array_map('exp', $jll); } $loss = 0.; for ($cluster = 0; $cluster < $this->k; $cluster++) { $mHat = array_column($memberships, $cluster); $means = $variances = []; foreach ($rotated as $values) { $a = $b = $total = 0.; foreach ($values as $i => $value) { $membership = $mHat[$i]; $a += $membership * $value; $total += $membership; } $total = $total ?: EPSILON; $mean = $a / $total; foreach ($values as $i => $value) { $b += $mHat[$i] * ($value - $mean) ** 2; } $variance = $b / $total; $means[] = $mean; $variances[] = $variance ?: EPSILON; $loss += $total; } $prior = array_sum($mHat) / $n; $this->means[$cluster] = $means; $this->variances[$cluster] = $variances; $this->priors[$cluster] = log($prior); } $this->steps[] = $loss; if ($this->logger) { $this->logger->info("Epoch $epoch complete, loss=$loss"); } if (is_nan($loss)) { break 1; } if (abs($loss - $prevLoss) < $this->minChange) { break 1; } $prevLoss = $loss; } if ($this->logger) { $this->logger->info('Training complete'); } }
php
public function train(Dataset $dataset) : void { DatasetIsCompatibleWithEstimator::check($dataset, $this); if ($this->logger) { $this->logger->info('Learner init ' . Params::stringify([ 'k' => $this->k, 'epochs' => $this->epochs, 'min_change' => $this->minChange, 'seeder' => $this->seeder, ])); } $samples = $dataset->samples(); $rotated = $dataset->columns(); $n = $dataset->numRows(); $this->priors = array_fill(0, $this->k, log(1. / $this->k)); [$this->means, $this->variances] = $this->initialize($dataset); $this->steps = []; $prevLoss = INF; for ($epoch = 1; $epoch <= $this->epochs; $epoch++) { $memberships = []; foreach ($dataset as $sample) { $jll = $this->jointLogLikelihood($sample); $memberships[] = array_map('exp', $jll); } $loss = 0.; for ($cluster = 0; $cluster < $this->k; $cluster++) { $mHat = array_column($memberships, $cluster); $means = $variances = []; foreach ($rotated as $values) { $a = $b = $total = 0.; foreach ($values as $i => $value) { $membership = $mHat[$i]; $a += $membership * $value; $total += $membership; } $total = $total ?: EPSILON; $mean = $a / $total; foreach ($values as $i => $value) { $b += $mHat[$i] * ($value - $mean) ** 2; } $variance = $b / $total; $means[] = $mean; $variances[] = $variance ?: EPSILON; $loss += $total; } $prior = array_sum($mHat) / $n; $this->means[$cluster] = $means; $this->variances[$cluster] = $variances; $this->priors[$cluster] = log($prior); } $this->steps[] = $loss; if ($this->logger) { $this->logger->info("Epoch $epoch complete, loss=$loss"); } if (is_nan($loss)) { break 1; } if (abs($loss - $prevLoss) < $this->minChange) { break 1; } $prevLoss = $loss; } if ($this->logger) { $this->logger->info('Training complete'); } }
[ "public", "function", "train", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "'Learner init '", ".", "Params", "::", "stringify", "(", "[", "'k'", "=>", "$", "this", "->", "k", ",", "'epochs'", "=>", "$", "this", "->", "epochs", ",", "'min_change'", "=>", "$", "this", "->", "minChange", ",", "'seeder'", "=>", "$", "this", "->", "seeder", ",", "]", ")", ")", ";", "}", "$", "samples", "=", "$", "dataset", "->", "samples", "(", ")", ";", "$", "rotated", "=", "$", "dataset", "->", "columns", "(", ")", ";", "$", "n", "=", "$", "dataset", "->", "numRows", "(", ")", ";", "$", "this", "->", "priors", "=", "array_fill", "(", "0", ",", "$", "this", "->", "k", ",", "log", "(", "1.", "/", "$", "this", "->", "k", ")", ")", ";", "[", "$", "this", "->", "means", ",", "$", "this", "->", "variances", "]", "=", "$", "this", "->", "initialize", "(", "$", "dataset", ")", ";", "$", "this", "->", "steps", "=", "[", "]", ";", "$", "prevLoss", "=", "INF", ";", "for", "(", "$", "epoch", "=", "1", ";", "$", "epoch", "<=", "$", "this", "->", "epochs", ";", "$", "epoch", "++", ")", "{", "$", "memberships", "=", "[", "]", ";", "foreach", "(", "$", "dataset", "as", "$", "sample", ")", "{", "$", "jll", "=", "$", "this", "->", "jointLogLikelihood", "(", "$", "sample", ")", ";", "$", "memberships", "[", "]", "=", "array_map", "(", "'exp'", ",", "$", "jll", ")", ";", "}", "$", "loss", "=", "0.", ";", "for", "(", "$", "cluster", "=", "0", ";", "$", "cluster", "<", "$", "this", "->", "k", ";", "$", "cluster", "++", ")", "{", "$", "mHat", "=", "array_column", "(", "$", "memberships", ",", "$", "cluster", ")", ";", "$", "means", "=", "$", "variances", "=", "[", "]", ";", "foreach", "(", "$", "rotated", "as", "$", "values", ")", "{", "$", "a", "=", "$", "b", "=", "$", "total", "=", "0.", ";", "foreach", "(", "$", "values", "as", "$", "i", "=>", "$", "value", ")", "{", "$", "membership", "=", "$", "mHat", "[", "$", "i", "]", ";", "$", "a", "+=", "$", "membership", "*", "$", "value", ";", "$", "total", "+=", "$", "membership", ";", "}", "$", "total", "=", "$", "total", "?", ":", "EPSILON", ";", "$", "mean", "=", "$", "a", "/", "$", "total", ";", "foreach", "(", "$", "values", "as", "$", "i", "=>", "$", "value", ")", "{", "$", "b", "+=", "$", "mHat", "[", "$", "i", "]", "*", "(", "$", "value", "-", "$", "mean", ")", "**", "2", ";", "}", "$", "variance", "=", "$", "b", "/", "$", "total", ";", "$", "means", "[", "]", "=", "$", "mean", ";", "$", "variances", "[", "]", "=", "$", "variance", "?", ":", "EPSILON", ";", "$", "loss", "+=", "$", "total", ";", "}", "$", "prior", "=", "array_sum", "(", "$", "mHat", ")", "/", "$", "n", ";", "$", "this", "->", "means", "[", "$", "cluster", "]", "=", "$", "means", ";", "$", "this", "->", "variances", "[", "$", "cluster", "]", "=", "$", "variances", ";", "$", "this", "->", "priors", "[", "$", "cluster", "]", "=", "log", "(", "$", "prior", ")", ";", "}", "$", "this", "->", "steps", "[", "]", "=", "$", "loss", ";", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "\"Epoch $epoch complete, loss=$loss\"", ")", ";", "}", "if", "(", "is_nan", "(", "$", "loss", ")", ")", "{", "break", "1", ";", "}", "if", "(", "abs", "(", "$", "loss", "-", "$", "prevLoss", ")", "<", "$", "this", "->", "minChange", ")", "{", "break", "1", ";", "}", "$", "prevLoss", "=", "$", "loss", ";", "}", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "'Training complete'", ")", ";", "}", "}" ]
Train the learner with a dataset. @param \Rubix\ML\Datasets\Dataset $dataset @throws \InvalidArgumentException
[ "Train", "the", "learner", "with", "a", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Clusterers/GaussianMixture.php#L238-L333
RubixML/RubixML
src/Clusterers/GaussianMixture.php
GaussianMixture.proba
public function proba(Dataset $dataset) : array { if (empty($this->priors)) { throw new RuntimeException('Estimator has not been trained.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); $probabilities = []; foreach ($dataset as $sample) { $jll = $this->jointLogLikelihood($sample); $total = logsumexp($jll); $dist = []; foreach ($jll as $cluster => $likelihood) { $dist[$cluster] = exp($likelihood - $total); } $probabilities[] = $dist; } return $probabilities; }
php
public function proba(Dataset $dataset) : array { if (empty($this->priors)) { throw new RuntimeException('Estimator has not been trained.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); $probabilities = []; foreach ($dataset as $sample) { $jll = $this->jointLogLikelihood($sample); $total = logsumexp($jll); $dist = []; foreach ($jll as $cluster => $likelihood) { $dist[$cluster] = exp($likelihood - $total); } $probabilities[] = $dist; } return $probabilities; }
[ "public", "function", "proba", "(", "Dataset", "$", "dataset", ")", ":", "array", "{", "if", "(", "empty", "(", "$", "this", "->", "priors", ")", ")", "{", "throw", "new", "RuntimeException", "(", "'Estimator has not been trained.'", ")", ";", "}", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "$", "probabilities", "=", "[", "]", ";", "foreach", "(", "$", "dataset", "as", "$", "sample", ")", "{", "$", "jll", "=", "$", "this", "->", "jointLogLikelihood", "(", "$", "sample", ")", ";", "$", "total", "=", "logsumexp", "(", "$", "jll", ")", ";", "$", "dist", "=", "[", "]", ";", "foreach", "(", "$", "jll", "as", "$", "cluster", "=>", "$", "likelihood", ")", "{", "$", "dist", "[", "$", "cluster", "]", "=", "exp", "(", "$", "likelihood", "-", "$", "total", ")", ";", "}", "$", "probabilities", "[", "]", "=", "$", "dist", ";", "}", "return", "$", "probabilities", ";", "}" ]
Estimate probabilities for each possible outcome. @param \Rubix\ML\Datasets\Dataset $dataset @throws \InvalidArgumentException @throws \RuntimeException @return array
[ "Estimate", "probabilities", "for", "each", "possible", "outcome", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Clusterers/GaussianMixture.php#L370-L395
RubixML/RubixML
src/Clusterers/GaussianMixture.php
GaussianMixture.initialize
protected function initialize(Dataset $dataset) : array { $centroids = $this->seeder->seed($dataset, $this->k); $kernel = new Euclidean(); $clusters = array_fill(0, $this->k, []); foreach ($dataset as $sample) { $bestDistance = INF; $bestCluster = -1; foreach ($centroids as $cluster => $centroid) { $distance = $kernel->compute($sample, $centroid); if ($distance < $bestDistance) { $bestDistance = $distance; $bestCluster = $cluster; } } $clusters[$bestCluster][] = $sample; } $means = $variances = []; foreach ($clusters as $cluster => $samples) { $mHat = $vHat = []; $columns = array_map(null, ...$samples); foreach ($columns as $values) { [$mean, $variance] = Stats::meanVar($values); $mHat[] = $mean; $vHat[] = $variance; } $means[$cluster] = $mHat; $variances[$cluster] = $vHat; } return [$means, $variances]; }
php
protected function initialize(Dataset $dataset) : array { $centroids = $this->seeder->seed($dataset, $this->k); $kernel = new Euclidean(); $clusters = array_fill(0, $this->k, []); foreach ($dataset as $sample) { $bestDistance = INF; $bestCluster = -1; foreach ($centroids as $cluster => $centroid) { $distance = $kernel->compute($sample, $centroid); if ($distance < $bestDistance) { $bestDistance = $distance; $bestCluster = $cluster; } } $clusters[$bestCluster][] = $sample; } $means = $variances = []; foreach ($clusters as $cluster => $samples) { $mHat = $vHat = []; $columns = array_map(null, ...$samples); foreach ($columns as $values) { [$mean, $variance] = Stats::meanVar($values); $mHat[] = $mean; $vHat[] = $variance; } $means[$cluster] = $mHat; $variances[$cluster] = $vHat; } return [$means, $variances]; }
[ "protected", "function", "initialize", "(", "Dataset", "$", "dataset", ")", ":", "array", "{", "$", "centroids", "=", "$", "this", "->", "seeder", "->", "seed", "(", "$", "dataset", ",", "$", "this", "->", "k", ")", ";", "$", "kernel", "=", "new", "Euclidean", "(", ")", ";", "$", "clusters", "=", "array_fill", "(", "0", ",", "$", "this", "->", "k", ",", "[", "]", ")", ";", "foreach", "(", "$", "dataset", "as", "$", "sample", ")", "{", "$", "bestDistance", "=", "INF", ";", "$", "bestCluster", "=", "-", "1", ";", "foreach", "(", "$", "centroids", "as", "$", "cluster", "=>", "$", "centroid", ")", "{", "$", "distance", "=", "$", "kernel", "->", "compute", "(", "$", "sample", ",", "$", "centroid", ")", ";", "if", "(", "$", "distance", "<", "$", "bestDistance", ")", "{", "$", "bestDistance", "=", "$", "distance", ";", "$", "bestCluster", "=", "$", "cluster", ";", "}", "}", "$", "clusters", "[", "$", "bestCluster", "]", "[", "]", "=", "$", "sample", ";", "}", "$", "means", "=", "$", "variances", "=", "[", "]", ";", "foreach", "(", "$", "clusters", "as", "$", "cluster", "=>", "$", "samples", ")", "{", "$", "mHat", "=", "$", "vHat", "=", "[", "]", ";", "$", "columns", "=", "array_map", "(", "null", ",", "...", "$", "samples", ")", ";", "foreach", "(", "$", "columns", "as", "$", "values", ")", "{", "[", "$", "mean", ",", "$", "variance", "]", "=", "Stats", "::", "meanVar", "(", "$", "values", ")", ";", "$", "mHat", "[", "]", "=", "$", "mean", ";", "$", "vHat", "[", "]", "=", "$", "variance", ";", "}", "$", "means", "[", "$", "cluster", "]", "=", "$", "mHat", ";", "$", "variances", "[", "$", "cluster", "]", "=", "$", "vHat", ";", "}", "return", "[", "$", "means", ",", "$", "variances", "]", ";", "}" ]
Initialize the gaussian components by calculating the means and variances of k initial cluster centroids generated by the seeder. @param \Rubix\ML\Datasets\Dataset $dataset @return array[]
[ "Initialize", "the", "gaussian", "components", "by", "calculating", "the", "means", "and", "variances", "of", "k", "initial", "cluster", "centroids", "generated", "by", "the", "seeder", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Clusterers/GaussianMixture.php#L437-L480
RubixML/RubixML
src/NeuralNet/Layers/PReLU.php
PReLU.initialize
public function initialize(int $fanIn) : int { $fanOut = $fanIn; $alpha = $this->initializer->initialize(1, $fanOut)->columnAsVector(0); $this->alpha = new VectorParam($alpha); $this->width = $fanOut; return $fanOut; }
php
public function initialize(int $fanIn) : int { $fanOut = $fanIn; $alpha = $this->initializer->initialize(1, $fanOut)->columnAsVector(0); $this->alpha = new VectorParam($alpha); $this->width = $fanOut; return $fanOut; }
[ "public", "function", "initialize", "(", "int", "$", "fanIn", ")", ":", "int", "{", "$", "fanOut", "=", "$", "fanIn", ";", "$", "alpha", "=", "$", "this", "->", "initializer", "->", "initialize", "(", "1", ",", "$", "fanOut", ")", "->", "columnAsVector", "(", "0", ")", ";", "$", "this", "->", "alpha", "=", "new", "VectorParam", "(", "$", "alpha", ")", ";", "$", "this", "->", "width", "=", "$", "fanOut", ";", "return", "$", "fanOut", ";", "}" ]
Initialize the layer with the fan in from the previous layer and return the fan out for this layer. @param int $fanIn @return int
[ "Initialize", "the", "layer", "with", "the", "fan", "in", "from", "the", "previous", "layer", "and", "return", "the", "fan", "out", "for", "this", "layer", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/NeuralNet/Layers/PReLU.php#L105-L116
RubixML/RubixML
src/NeuralNet/Layers/PReLU.php
PReLU.forward
public function forward(Matrix $input) : Matrix { $this->input = $input; return $this->compute($input); }
php
public function forward(Matrix $input) : Matrix { $this->input = $input; return $this->compute($input); }
[ "public", "function", "forward", "(", "Matrix", "$", "input", ")", ":", "Matrix", "{", "$", "this", "->", "input", "=", "$", "input", ";", "return", "$", "this", "->", "compute", "(", "$", "input", ")", ";", "}" ]
Compute a forward pass through the layer. @param \Rubix\Tensor\Matrix $input @return \Rubix\Tensor\Matrix
[ "Compute", "a", "forward", "pass", "through", "the", "layer", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/NeuralNet/Layers/PReLU.php#L124-L129
RubixML/RubixML
src/NeuralNet/Layers/PReLU.php
PReLU.back
public function back(Deferred $prevGradient, Optimizer $optimizer) : Deferred { if (!$this->alpha) { throw new RuntimeException('Layer has not been initlaized.'); } if (!$this->input) { throw new RuntimeException('Must perform a forward pass before' . ' backpropagating.'); } $dOut = $prevGradient->result(); $dIn = $this->input->clipUpper(0.); $dAlpha = $dOut->multiply($dIn)->sum(); $optimizer->step($this->alpha, $dAlpha); $z = $this->input; unset($this->input); return new Deferred(function () use ($z, $dOut) { return $this->differentiate($z)->multiply($dOut); }); }
php
public function back(Deferred $prevGradient, Optimizer $optimizer) : Deferred { if (!$this->alpha) { throw new RuntimeException('Layer has not been initlaized.'); } if (!$this->input) { throw new RuntimeException('Must perform a forward pass before' . ' backpropagating.'); } $dOut = $prevGradient->result(); $dIn = $this->input->clipUpper(0.); $dAlpha = $dOut->multiply($dIn)->sum(); $optimizer->step($this->alpha, $dAlpha); $z = $this->input; unset($this->input); return new Deferred(function () use ($z, $dOut) { return $this->differentiate($z)->multiply($dOut); }); }
[ "public", "function", "back", "(", "Deferred", "$", "prevGradient", ",", "Optimizer", "$", "optimizer", ")", ":", "Deferred", "{", "if", "(", "!", "$", "this", "->", "alpha", ")", "{", "throw", "new", "RuntimeException", "(", "'Layer has not been initlaized.'", ")", ";", "}", "if", "(", "!", "$", "this", "->", "input", ")", "{", "throw", "new", "RuntimeException", "(", "'Must perform a forward pass before'", ".", "' backpropagating.'", ")", ";", "}", "$", "dOut", "=", "$", "prevGradient", "->", "result", "(", ")", ";", "$", "dIn", "=", "$", "this", "->", "input", "->", "clipUpper", "(", "0.", ")", ";", "$", "dAlpha", "=", "$", "dOut", "->", "multiply", "(", "$", "dIn", ")", "->", "sum", "(", ")", ";", "$", "optimizer", "->", "step", "(", "$", "this", "->", "alpha", ",", "$", "dAlpha", ")", ";", "$", "z", "=", "$", "this", "->", "input", ";", "unset", "(", "$", "this", "->", "input", ")", ";", "return", "new", "Deferred", "(", "function", "(", ")", "use", "(", "$", "z", ",", "$", "dOut", ")", "{", "return", "$", "this", "->", "differentiate", "(", "$", "z", ")", "->", "multiply", "(", "$", "dOut", ")", ";", "}", ")", ";", "}" ]
Calculate the gradient and update the parameters of the layer. @param \Rubix\ML\NeuralNet\Deferred $prevGradient @param \Rubix\ML\NeuralNet\Optimizers\Optimizer $optimizer @throws \RuntimeException @return \Rubix\ML\NeuralNet\Deferred
[ "Calculate", "the", "gradient", "and", "update", "the", "parameters", "of", "the", "layer", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/NeuralNet/Layers/PReLU.php#L150-L176
RubixML/RubixML
src/NeuralNet/Layers/PReLU.php
PReLU.compute
protected function compute(Matrix $z) : Matrix { if (!$this->alpha) { throw new RuntimeException('Layer is not initialized.'); } $alphas = $this->alpha->w(); $computed = []; foreach ($z as $i => $row) { $alpha = $alphas[$i]; $activations = []; foreach ($row as $value) { $activations[] = $value > 0. ? $value : $alpha * $value; } $computed[] = $activations; } return Matrix::quick($computed); }
php
protected function compute(Matrix $z) : Matrix { if (!$this->alpha) { throw new RuntimeException('Layer is not initialized.'); } $alphas = $this->alpha->w(); $computed = []; foreach ($z as $i => $row) { $alpha = $alphas[$i]; $activations = []; foreach ($row as $value) { $activations[] = $value > 0. ? $value : $alpha * $value; } $computed[] = $activations; } return Matrix::quick($computed); }
[ "protected", "function", "compute", "(", "Matrix", "$", "z", ")", ":", "Matrix", "{", "if", "(", "!", "$", "this", "->", "alpha", ")", "{", "throw", "new", "RuntimeException", "(", "'Layer is not initialized.'", ")", ";", "}", "$", "alphas", "=", "$", "this", "->", "alpha", "->", "w", "(", ")", ";", "$", "computed", "=", "[", "]", ";", "foreach", "(", "$", "z", "as", "$", "i", "=>", "$", "row", ")", "{", "$", "alpha", "=", "$", "alphas", "[", "$", "i", "]", ";", "$", "activations", "=", "[", "]", ";", "foreach", "(", "$", "row", "as", "$", "value", ")", "{", "$", "activations", "[", "]", "=", "$", "value", ">", "0.", "?", "$", "value", ":", "$", "alpha", "*", "$", "value", ";", "}", "$", "computed", "[", "]", "=", "$", "activations", ";", "}", "return", "Matrix", "::", "quick", "(", "$", "computed", ")", ";", "}" ]
Compute the leaky ReLU activation function and return a matrix. @param \Rubix\Tensor\Matrix $z @throws \RuntimeException @return \Rubix\Tensor\Matrix
[ "Compute", "the", "leaky", "ReLU", "activation", "function", "and", "return", "a", "matrix", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/NeuralNet/Layers/PReLU.php#L185-L210
RubixML/RubixML
src/NeuralNet/Layers/PReLU.php
PReLU.differentiate
protected function differentiate(Matrix $z) : Matrix { if (!$this->alpha) { throw new RuntimeException('Layer has not been initlaized.'); } $alphas = $this->alpha->w(); $gradient = []; foreach ($z as $i => $row) { $alpha = $alphas[$i]; $temp = []; foreach ($row as $value) { $temp[] = $value > 0. ? 1. : $alpha; } $gradient[] = $temp; } return Matrix::quick($gradient); }
php
protected function differentiate(Matrix $z) : Matrix { if (!$this->alpha) { throw new RuntimeException('Layer has not been initlaized.'); } $alphas = $this->alpha->w(); $gradient = []; foreach ($z as $i => $row) { $alpha = $alphas[$i]; $temp = []; foreach ($row as $value) { $temp[] = $value > 0. ? 1. : $alpha; } $gradient[] = $temp; } return Matrix::quick($gradient); }
[ "protected", "function", "differentiate", "(", "Matrix", "$", "z", ")", ":", "Matrix", "{", "if", "(", "!", "$", "this", "->", "alpha", ")", "{", "throw", "new", "RuntimeException", "(", "'Layer has not been initlaized.'", ")", ";", "}", "$", "alphas", "=", "$", "this", "->", "alpha", "->", "w", "(", ")", ";", "$", "gradient", "=", "[", "]", ";", "foreach", "(", "$", "z", "as", "$", "i", "=>", "$", "row", ")", "{", "$", "alpha", "=", "$", "alphas", "[", "$", "i", "]", ";", "$", "temp", "=", "[", "]", ";", "foreach", "(", "$", "row", "as", "$", "value", ")", "{", "$", "temp", "[", "]", "=", "$", "value", ">", "0.", "?", "1.", ":", "$", "alpha", ";", "}", "$", "gradient", "[", "]", "=", "$", "temp", ";", "}", "return", "Matrix", "::", "quick", "(", "$", "gradient", ")", ";", "}" ]
Calculate the partial derivatives of the activation function. @param \Rubix\Tensor\Matrix $z @throws \RuntimeException @return \Rubix\Tensor\Matrix
[ "Calculate", "the", "partial", "derivatives", "of", "the", "activation", "function", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/NeuralNet/Layers/PReLU.php#L219-L242
RubixML/RubixML
src/Other/Traits/LoggerAware.php
LoggerAware.setLogger
public function setLogger(LoggerInterface $logger) : void { if ($this instanceof Wrapper) { $estimator = $this->base(); if ($estimator instanceof Verbose) { $estimator->setLogger($logger); } } $this->logger = $logger; }
php
public function setLogger(LoggerInterface $logger) : void { if ($this instanceof Wrapper) { $estimator = $this->base(); if ($estimator instanceof Verbose) { $estimator->setLogger($logger); } } $this->logger = $logger; }
[ "public", "function", "setLogger", "(", "LoggerInterface", "$", "logger", ")", ":", "void", "{", "if", "(", "$", "this", "instanceof", "Wrapper", ")", "{", "$", "estimator", "=", "$", "this", "->", "base", "(", ")", ";", "if", "(", "$", "estimator", "instanceof", "Verbose", ")", "{", "$", "estimator", "->", "setLogger", "(", "$", "logger", ")", ";", "}", "}", "$", "this", "->", "logger", "=", "$", "logger", ";", "}" ]
Sets a logger instance on the object. @param \Psr\Log\LoggerInterface $logger
[ "Sets", "a", "logger", "instance", "on", "the", "object", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Other/Traits/LoggerAware.php#L34-L45
RubixML/RubixML
src/Transformers/ZScaleStandardizer.php
ZScaleStandardizer.fit
public function fit(Dataset $dataset) : void { $columns = $dataset->columnsByType(DataType::CONTINUOUS); $this->means = $this->variances = $this->stddevs = []; foreach ($columns as $column => $values) { [$mean, $variance] = Stats::meanVar($values); $this->means[$column] = $mean; $this->variances[$column] = $variance; $this->stddevs[$column] = sqrt($variance ?: EPSILON); } $this->n = $dataset->numRows(); }
php
public function fit(Dataset $dataset) : void { $columns = $dataset->columnsByType(DataType::CONTINUOUS); $this->means = $this->variances = $this->stddevs = []; foreach ($columns as $column => $values) { [$mean, $variance] = Stats::meanVar($values); $this->means[$column] = $mean; $this->variances[$column] = $variance; $this->stddevs[$column] = sqrt($variance ?: EPSILON); } $this->n = $dataset->numRows(); }
[ "public", "function", "fit", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "$", "columns", "=", "$", "dataset", "->", "columnsByType", "(", "DataType", "::", "CONTINUOUS", ")", ";", "$", "this", "->", "means", "=", "$", "this", "->", "variances", "=", "$", "this", "->", "stddevs", "=", "[", "]", ";", "foreach", "(", "$", "columns", "as", "$", "column", "=>", "$", "values", ")", "{", "[", "$", "mean", ",", "$", "variance", "]", "=", "Stats", "::", "meanVar", "(", "$", "values", ")", ";", "$", "this", "->", "means", "[", "$", "column", "]", "=", "$", "mean", ";", "$", "this", "->", "variances", "[", "$", "column", "]", "=", "$", "variance", ";", "$", "this", "->", "stddevs", "[", "$", "column", "]", "=", "sqrt", "(", "$", "variance", "?", ":", "EPSILON", ")", ";", "}", "$", "this", "->", "n", "=", "$", "dataset", "->", "numRows", "(", ")", ";", "}" ]
Fit the transformer to the dataset. @param \Rubix\ML\Datasets\Dataset $dataset
[ "Fit", "the", "transformer", "to", "the", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/ZScaleStandardizer.php#L116-L131
RubixML/RubixML
src/Transformers/ZScaleStandardizer.php
ZScaleStandardizer.update
public function update(Dataset $dataset) : void { if ($this->means === null or $this->variances === null) { $this->fit($dataset); return; } $n = $dataset->numRows(); foreach ($this->means as $column => $oldMean) { $oldVariance = $this->variances[$column]; $values = $dataset->column($column); [$mean, $variance] = Stats::meanVar($values); $this->means[$column] = (($n * $mean) + ($this->n * $oldMean)) / ($this->n + $n); $varNew = ($this->n * $oldVariance + ($n * $variance) + ($this->n / ($n * ($this->n + $n))) * ($n * $oldMean - $n * $mean) ** 2) / ($this->n + $n); $this->variances[$column] = $varNew; $this->stddevs[$column] = sqrt($varNew ?: EPSILON); } $this->n += $n; }
php
public function update(Dataset $dataset) : void { if ($this->means === null or $this->variances === null) { $this->fit($dataset); return; } $n = $dataset->numRows(); foreach ($this->means as $column => $oldMean) { $oldVariance = $this->variances[$column]; $values = $dataset->column($column); [$mean, $variance] = Stats::meanVar($values); $this->means[$column] = (($n * $mean) + ($this->n * $oldMean)) / ($this->n + $n); $varNew = ($this->n * $oldVariance + ($n * $variance) + ($this->n / ($n * ($this->n + $n))) * ($n * $oldMean - $n * $mean) ** 2) / ($this->n + $n); $this->variances[$column] = $varNew; $this->stddevs[$column] = sqrt($varNew ?: EPSILON); } $this->n += $n; }
[ "public", "function", "update", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "if", "(", "$", "this", "->", "means", "===", "null", "or", "$", "this", "->", "variances", "===", "null", ")", "{", "$", "this", "->", "fit", "(", "$", "dataset", ")", ";", "return", ";", "}", "$", "n", "=", "$", "dataset", "->", "numRows", "(", ")", ";", "foreach", "(", "$", "this", "->", "means", "as", "$", "column", "=>", "$", "oldMean", ")", "{", "$", "oldVariance", "=", "$", "this", "->", "variances", "[", "$", "column", "]", ";", "$", "values", "=", "$", "dataset", "->", "column", "(", "$", "column", ")", ";", "[", "$", "mean", ",", "$", "variance", "]", "=", "Stats", "::", "meanVar", "(", "$", "values", ")", ";", "$", "this", "->", "means", "[", "$", "column", "]", "=", "(", "(", "$", "n", "*", "$", "mean", ")", "+", "(", "$", "this", "->", "n", "*", "$", "oldMean", ")", ")", "/", "(", "$", "this", "->", "n", "+", "$", "n", ")", ";", "$", "varNew", "=", "(", "$", "this", "->", "n", "*", "$", "oldVariance", "+", "(", "$", "n", "*", "$", "variance", ")", "+", "(", "$", "this", "->", "n", "/", "(", "$", "n", "*", "(", "$", "this", "->", "n", "+", "$", "n", ")", ")", ")", "*", "(", "$", "n", "*", "$", "oldMean", "-", "$", "n", "*", "$", "mean", ")", "**", "2", ")", "/", "(", "$", "this", "->", "n", "+", "$", "n", ")", ";", "$", "this", "->", "variances", "[", "$", "column", "]", "=", "$", "varNew", ";", "$", "this", "->", "stddevs", "[", "$", "column", "]", "=", "sqrt", "(", "$", "varNew", "?", ":", "EPSILON", ")", ";", "}", "$", "this", "->", "n", "+=", "$", "n", ";", "}" ]
Update the fitting of the transformer. @param \Rubix\ML\Datasets\Dataset $dataset
[ "Update", "the", "fitting", "of", "the", "transformer", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/ZScaleStandardizer.php#L138-L170
RubixML/RubixML
src/Transformers/ZScaleStandardizer.php
ZScaleStandardizer.transform
public function transform(array &$samples) : void { if ($this->means === null or $this->stddevs === null) { throw new RuntimeException('Transformer has not been fitted.'); } foreach ($samples as &$sample) { foreach ($this->stddevs as $column => $stddev) { $feature = $sample[$column]; if ($this->center) { $feature -= $this->means[$column]; } $sample[$column] = $feature / $stddev; } } }
php
public function transform(array &$samples) : void { if ($this->means === null or $this->stddevs === null) { throw new RuntimeException('Transformer has not been fitted.'); } foreach ($samples as &$sample) { foreach ($this->stddevs as $column => $stddev) { $feature = $sample[$column]; if ($this->center) { $feature -= $this->means[$column]; } $sample[$column] = $feature / $stddev; } } }
[ "public", "function", "transform", "(", "array", "&", "$", "samples", ")", ":", "void", "{", "if", "(", "$", "this", "->", "means", "===", "null", "or", "$", "this", "->", "stddevs", "===", "null", ")", "{", "throw", "new", "RuntimeException", "(", "'Transformer has not been fitted.'", ")", ";", "}", "foreach", "(", "$", "samples", "as", "&", "$", "sample", ")", "{", "foreach", "(", "$", "this", "->", "stddevs", "as", "$", "column", "=>", "$", "stddev", ")", "{", "$", "feature", "=", "$", "sample", "[", "$", "column", "]", ";", "if", "(", "$", "this", "->", "center", ")", "{", "$", "feature", "-=", "$", "this", "->", "means", "[", "$", "column", "]", ";", "}", "$", "sample", "[", "$", "column", "]", "=", "$", "feature", "/", "$", "stddev", ";", "}", "}", "}" ]
Transform the dataset in place. @param array $samples @throws \RuntimeException
[ "Transform", "the", "dataset", "in", "place", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/ZScaleStandardizer.php#L178-L195
RubixML/RubixML
src/CrossValidation/Metrics/VMeasure.php
VMeasure.score
public function score(array $predictions, array $labels) : float { if (empty($predictions)) { return 0.; } $homogeneity = (new Homogeneity())->score($predictions, $labels); $completeness = (new Completeness())->score($predictions, $labels); return 2. * ($homogeneity * $completeness) / ($homogeneity + $completeness); }
php
public function score(array $predictions, array $labels) : float { if (empty($predictions)) { return 0.; } $homogeneity = (new Homogeneity())->score($predictions, $labels); $completeness = (new Completeness())->score($predictions, $labels); return 2. * ($homogeneity * $completeness) / ($homogeneity + $completeness); }
[ "public", "function", "score", "(", "array", "$", "predictions", ",", "array", "$", "labels", ")", ":", "float", "{", "if", "(", "empty", "(", "$", "predictions", ")", ")", "{", "return", "0.", ";", "}", "$", "homogeneity", "=", "(", "new", "Homogeneity", "(", ")", ")", "->", "score", "(", "$", "predictions", ",", "$", "labels", ")", ";", "$", "completeness", "=", "(", "new", "Completeness", "(", ")", ")", "->", "score", "(", "$", "predictions", ",", "$", "labels", ")", ";", "return", "2.", "*", "(", "$", "homogeneity", "*", "$", "completeness", ")", "/", "(", "$", "homogeneity", "+", "$", "completeness", ")", ";", "}" ]
Score a set of predictions. @param array $predictions @param array $labels @return float
[ "Score", "a", "set", "of", "predictions", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/CrossValidation/Metrics/VMeasure.php#L52-L63
RubixML/RubixML
src/CrossValidation/Metrics/RandIndex.php
RandIndex.comb
public static function comb(int $n, int $k = 2) : int { return $k === 0 ? 1 : (int) (($n * self::comb($n - 1, $k - 1)) / $k); }
php
public static function comb(int $n, int $k = 2) : int { return $k === 0 ? 1 : (int) (($n * self::comb($n - 1, $k - 1)) / $k); }
[ "public", "static", "function", "comb", "(", "int", "$", "n", ",", "int", "$", "k", "=", "2", ")", ":", "int", "{", "return", "$", "k", "===", "0", "?", "1", ":", "(", "int", ")", "(", "(", "$", "n", "*", "self", "::", "comb", "(", "$", "n", "-", "1", ",", "$", "k", "-", "1", ")", ")", "/", "$", "k", ")", ";", "}" ]
Compute n choose k. @param int $n @param int $k @return int
[ "Compute", "n", "choose", "k", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/CrossValidation/Metrics/RandIndex.php#L36-L39
RubixML/RubixML
src/CrossValidation/Metrics/RandIndex.php
RandIndex.score
public function score(array $predictions, array $labels) : float { if (empty($predictions)) { return 0.; } $n = count($predictions); if ($n !== count($labels)) { throw new InvalidArgumentException('The number of labels' . ' must equal the number of predictions.'); } $table = Matrix::build((new ContingencyTable())->generate($predictions, $labels)); $sigma = $table->map([self::class, 'comb'])->sum()->sum(); $alpha = $table->sum()->map([self::class, 'comb'])->sum(); $beta = $table->transpose()->sum()->map([self::class, 'comb'])->sum(); $pHat = ($alpha * $beta) / self::comb($n); $mean = ($alpha + $beta) / 2.; return ($sigma - $pHat) / ($mean - $pHat); }
php
public function score(array $predictions, array $labels) : float { if (empty($predictions)) { return 0.; } $n = count($predictions); if ($n !== count($labels)) { throw new InvalidArgumentException('The number of labels' . ' must equal the number of predictions.'); } $table = Matrix::build((new ContingencyTable())->generate($predictions, $labels)); $sigma = $table->map([self::class, 'comb'])->sum()->sum(); $alpha = $table->sum()->map([self::class, 'comb'])->sum(); $beta = $table->transpose()->sum()->map([self::class, 'comb'])->sum(); $pHat = ($alpha * $beta) / self::comb($n); $mean = ($alpha + $beta) / 2.; return ($sigma - $pHat) / ($mean - $pHat); }
[ "public", "function", "score", "(", "array", "$", "predictions", ",", "array", "$", "labels", ")", ":", "float", "{", "if", "(", "empty", "(", "$", "predictions", ")", ")", "{", "return", "0.", ";", "}", "$", "n", "=", "count", "(", "$", "predictions", ")", ";", "if", "(", "$", "n", "!==", "count", "(", "$", "labels", ")", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'The number of labels'", ".", "' must equal the number of predictions.'", ")", ";", "}", "$", "table", "=", "Matrix", "::", "build", "(", "(", "new", "ContingencyTable", "(", ")", ")", "->", "generate", "(", "$", "predictions", ",", "$", "labels", ")", ")", ";", "$", "sigma", "=", "$", "table", "->", "map", "(", "[", "self", "::", "class", ",", "'comb'", "]", ")", "->", "sum", "(", ")", "->", "sum", "(", ")", ";", "$", "alpha", "=", "$", "table", "->", "sum", "(", ")", "->", "map", "(", "[", "self", "::", "class", ",", "'comb'", "]", ")", "->", "sum", "(", ")", ";", "$", "beta", "=", "$", "table", "->", "transpose", "(", ")", "->", "sum", "(", ")", "->", "map", "(", "[", "self", "::", "class", ",", "'comb'", "]", ")", "->", "sum", "(", ")", ";", "$", "pHat", "=", "(", "$", "alpha", "*", "$", "beta", ")", "/", "self", "::", "comb", "(", "$", "n", ")", ";", "$", "mean", "=", "(", "$", "alpha", "+", "$", "beta", ")", "/", "2.", ";", "return", "(", "$", "sigma", "-", "$", "pHat", ")", "/", "(", "$", "mean", "-", "$", "pHat", ")", ";", "}" ]
Score a set of predictions. @param array $predictions @param array $labels @throws \InvalidArgumentException @return float
[ "Score", "a", "set", "of", "predictions", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/CrossValidation/Metrics/RandIndex.php#L71-L95
RubixML/RubixML
src/Clusterers/Seeders/KMC2.php
KMC2.seed
public function seed(Dataset $dataset, int $k) : array { $centroids = $dataset->randomSubsetWithReplacement(1)->samples(); for ($i = 1; $i < $k; $i++) { $candidates = $dataset->randomSubsetWithReplacement($this->m)->samples(); $x = array_pop($candidates); $target = end($centroids); $xDistance = $this->kernel->compute($x, $target) ?: EPSILON; foreach ($candidates as $y) { $yDistance = $this->kernel->compute($y, $target); $probability = min(1., $yDistance / $xDistance); if ($probability === 1.) { $xDistance = $yDistance; $x = $y; continue 1; } $threshold = rand(0, self::PHI) / self::PHI; if ($probability > $threshold) { $xDistance = $yDistance; $x = $y; } } $centroids[] = $x; } return $centroids; }
php
public function seed(Dataset $dataset, int $k) : array { $centroids = $dataset->randomSubsetWithReplacement(1)->samples(); for ($i = 1; $i < $k; $i++) { $candidates = $dataset->randomSubsetWithReplacement($this->m)->samples(); $x = array_pop($candidates); $target = end($centroids); $xDistance = $this->kernel->compute($x, $target) ?: EPSILON; foreach ($candidates as $y) { $yDistance = $this->kernel->compute($y, $target); $probability = min(1., $yDistance / $xDistance); if ($probability === 1.) { $xDistance = $yDistance; $x = $y; continue 1; } $threshold = rand(0, self::PHI) / self::PHI; if ($probability > $threshold) { $xDistance = $yDistance; $x = $y; } } $centroids[] = $x; } return $centroids; }
[ "public", "function", "seed", "(", "Dataset", "$", "dataset", ",", "int", "$", "k", ")", ":", "array", "{", "$", "centroids", "=", "$", "dataset", "->", "randomSubsetWithReplacement", "(", "1", ")", "->", "samples", "(", ")", ";", "for", "(", "$", "i", "=", "1", ";", "$", "i", "<", "$", "k", ";", "$", "i", "++", ")", "{", "$", "candidates", "=", "$", "dataset", "->", "randomSubsetWithReplacement", "(", "$", "this", "->", "m", ")", "->", "samples", "(", ")", ";", "$", "x", "=", "array_pop", "(", "$", "candidates", ")", ";", "$", "target", "=", "end", "(", "$", "centroids", ")", ";", "$", "xDistance", "=", "$", "this", "->", "kernel", "->", "compute", "(", "$", "x", ",", "$", "target", ")", "?", ":", "EPSILON", ";", "foreach", "(", "$", "candidates", "as", "$", "y", ")", "{", "$", "yDistance", "=", "$", "this", "->", "kernel", "->", "compute", "(", "$", "y", ",", "$", "target", ")", ";", "$", "probability", "=", "min", "(", "1.", ",", "$", "yDistance", "/", "$", "xDistance", ")", ";", "if", "(", "$", "probability", "===", "1.", ")", "{", "$", "xDistance", "=", "$", "yDistance", ";", "$", "x", "=", "$", "y", ";", "continue", "1", ";", "}", "$", "threshold", "=", "rand", "(", "0", ",", "self", "::", "PHI", ")", "/", "self", "::", "PHI", ";", "if", "(", "$", "probability", ">", "$", "threshold", ")", "{", "$", "xDistance", "=", "$", "yDistance", ";", "$", "x", "=", "$", "y", ";", "}", "}", "$", "centroids", "[", "]", "=", "$", "x", ";", "}", "return", "$", "centroids", ";", "}" ]
Seed k cluster centroids from a dataset. @param \Rubix\ML\Datasets\Dataset $dataset @param int $k @throws \RuntimeException @return array
[ "Seed", "k", "cluster", "centroids", "from", "a", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Clusterers/Seeders/KMC2.php#L68-L105
RubixML/RubixML
src/AnomalyDetectors/OneClassSVM.php
OneClassSVM.train
public function train(Dataset $dataset) : void { DatasetIsCompatibleWithEstimator::check($dataset, $this); $this->model = $this->svm->train($dataset->samples()); }
php
public function train(Dataset $dataset) : void { DatasetIsCompatibleWithEstimator::check($dataset, $this); $this->model = $this->svm->train($dataset->samples()); }
[ "public", "function", "train", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "$", "this", "->", "model", "=", "$", "this", "->", "svm", "->", "train", "(", "$", "dataset", "->", "samples", "(", ")", ")", ";", "}" ]
Train the learner with a dataset. @param \Rubix\ML\Datasets\Dataset $dataset @throws \InvalidArgumentException
[ "Train", "the", "learner", "with", "a", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/AnomalyDetectors/OneClassSVM.php#L142-L147
RubixML/RubixML
src/AnomalyDetectors/OneClassSVM.php
OneClassSVM.predict
public function predict(Dataset $dataset) : array { if (!$this->model) { throw new RuntimeException('The learner has not' . ' been trained.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); $predictions = []; foreach ($dataset as $sample) { $predictions[] = $this->model->predict($sample) !== 1. ? 0 : 1; } return $predictions; }
php
public function predict(Dataset $dataset) : array { if (!$this->model) { throw new RuntimeException('The learner has not' . ' been trained.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); $predictions = []; foreach ($dataset as $sample) { $predictions[] = $this->model->predict($sample) !== 1. ? 0 : 1; } return $predictions; }
[ "public", "function", "predict", "(", "Dataset", "$", "dataset", ")", ":", "array", "{", "if", "(", "!", "$", "this", "->", "model", ")", "{", "throw", "new", "RuntimeException", "(", "'The learner has not'", ".", "' been trained.'", ")", ";", "}", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "$", "predictions", "=", "[", "]", ";", "foreach", "(", "$", "dataset", "as", "$", "sample", ")", "{", "$", "predictions", "[", "]", "=", "$", "this", "->", "model", "->", "predict", "(", "$", "sample", ")", "!==", "1.", "?", "0", ":", "1", ";", "}", "return", "$", "predictions", ";", "}" ]
Make predictions from a dataset. @param \Rubix\ML\Datasets\Dataset $dataset @throws \InvalidArgumentException @throws \RuntimeException @return array
[ "Make", "predictions", "from", "a", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/AnomalyDetectors/OneClassSVM.php#L157-L173
RubixML/RubixML
src/Other/Helpers/Params.php
Params.ints
public static function ints(int $min, int $max, int $n = 10) : array { if (($max - $min) < 0) { throw new InvalidArgumentException('Maximum cannot be' . ' less than minimum.'); } if ($n < 1) { throw new InvalidArgumentException('Cannot generate less' . ' than 1 parameter.'); } if ($n > ($max - $min + 1)) { throw new InvalidArgumentException('Cannot generate more' . ' unique integers than in range.'); } $distribution = []; while (count($distribution) < $n) { $r = rand($min, $max); if (!in_array($r, $distribution)) { $distribution[] = $r; } } return $distribution; }
php
public static function ints(int $min, int $max, int $n = 10) : array { if (($max - $min) < 0) { throw new InvalidArgumentException('Maximum cannot be' . ' less than minimum.'); } if ($n < 1) { throw new InvalidArgumentException('Cannot generate less' . ' than 1 parameter.'); } if ($n > ($max - $min + 1)) { throw new InvalidArgumentException('Cannot generate more' . ' unique integers than in range.'); } $distribution = []; while (count($distribution) < $n) { $r = rand($min, $max); if (!in_array($r, $distribution)) { $distribution[] = $r; } } return $distribution; }
[ "public", "static", "function", "ints", "(", "int", "$", "min", ",", "int", "$", "max", ",", "int", "$", "n", "=", "10", ")", ":", "array", "{", "if", "(", "(", "$", "max", "-", "$", "min", ")", "<", "0", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'Maximum cannot be'", ".", "' less than minimum.'", ")", ";", "}", "if", "(", "$", "n", "<", "1", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'Cannot generate less'", ".", "' than 1 parameter.'", ")", ";", "}", "if", "(", "$", "n", ">", "(", "$", "max", "-", "$", "min", "+", "1", ")", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'Cannot generate more'", ".", "' unique integers than in range.'", ")", ";", "}", "$", "distribution", "=", "[", "]", ";", "while", "(", "count", "(", "$", "distribution", ")", "<", "$", "n", ")", "{", "$", "r", "=", "rand", "(", "$", "min", ",", "$", "max", ")", ";", "if", "(", "!", "in_array", "(", "$", "r", ",", "$", "distribution", ")", ")", "{", "$", "distribution", "[", "]", "=", "$", "r", ";", "}", "}", "return", "$", "distribution", ";", "}" ]
Generate a random unique integer distribution. @param int $min @param int $max @param int $n @throws \InvalidArgumentException @return int[]
[ "Generate", "a", "random", "unique", "integer", "distribution", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Other/Helpers/Params.php#L32-L60
RubixML/RubixML
src/Other/Helpers/Params.php
Params.floats
public static function floats(float $min, float $max, int $n = 10) : array { if (($max - $min) < 0.) { throw new InvalidArgumentException('Maximum cannot be' . ' less than minimum.'); } if ($n < 1) { throw new InvalidArgumentException('Cannot generate less' . ' than 1 parameter.'); } $min = (int) round($min * self::PHI); $max = (int) round($max * self::PHI); $distribution = []; for ($i = 0; $i < $n; $i++) { $distribution[] = rand($min, $max) / self::PHI; } return $distribution; }
php
public static function floats(float $min, float $max, int $n = 10) : array { if (($max - $min) < 0.) { throw new InvalidArgumentException('Maximum cannot be' . ' less than minimum.'); } if ($n < 1) { throw new InvalidArgumentException('Cannot generate less' . ' than 1 parameter.'); } $min = (int) round($min * self::PHI); $max = (int) round($max * self::PHI); $distribution = []; for ($i = 0; $i < $n; $i++) { $distribution[] = rand($min, $max) / self::PHI; } return $distribution; }
[ "public", "static", "function", "floats", "(", "float", "$", "min", ",", "float", "$", "max", ",", "int", "$", "n", "=", "10", ")", ":", "array", "{", "if", "(", "(", "$", "max", "-", "$", "min", ")", "<", "0.", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'Maximum cannot be'", ".", "' less than minimum.'", ")", ";", "}", "if", "(", "$", "n", "<", "1", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'Cannot generate less'", ".", "' than 1 parameter.'", ")", ";", "}", "$", "min", "=", "(", "int", ")", "round", "(", "$", "min", "*", "self", "::", "PHI", ")", ";", "$", "max", "=", "(", "int", ")", "round", "(", "$", "max", "*", "self", "::", "PHI", ")", ";", "$", "distribution", "=", "[", "]", ";", "for", "(", "$", "i", "=", "0", ";", "$", "i", "<", "$", "n", ";", "$", "i", "++", ")", "{", "$", "distribution", "[", "]", "=", "rand", "(", "$", "min", ",", "$", "max", ")", "/", "self", "::", "PHI", ";", "}", "return", "$", "distribution", ";", "}" ]
Generate a random distribution of floating point parameters. @param float $min @param float $max @param int $n @throws \InvalidArgumentException @return float[]
[ "Generate", "a", "random", "distribution", "of", "floating", "point", "parameters", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Other/Helpers/Params.php#L71-L93
RubixML/RubixML
src/Other/Helpers/Params.php
Params.grid
public static function grid(float $min, float $max, int $n = 10) : array { if ($min > $max) { throw new InvalidArgumentException('Max cannot be less' . ' then min.'); } if ($n < 2) { throw new InvalidArgumentException('Cannot generate less' . ' than 2 parameters.'); } $interval = ($max - $min) / ($n - 1); return range($min, $max, $interval); }
php
public static function grid(float $min, float $max, int $n = 10) : array { if ($min > $max) { throw new InvalidArgumentException('Max cannot be less' . ' then min.'); } if ($n < 2) { throw new InvalidArgumentException('Cannot generate less' . ' than 2 parameters.'); } $interval = ($max - $min) / ($n - 1); return range($min, $max, $interval); }
[ "public", "static", "function", "grid", "(", "float", "$", "min", ",", "float", "$", "max", ",", "int", "$", "n", "=", "10", ")", ":", "array", "{", "if", "(", "$", "min", ">", "$", "max", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'Max cannot be less'", ".", "' then min.'", ")", ";", "}", "if", "(", "$", "n", "<", "2", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'Cannot generate less'", ".", "' than 2 parameters.'", ")", ";", "}", "$", "interval", "=", "(", "$", "max", "-", "$", "min", ")", "/", "(", "$", "n", "-", "1", ")", ";", "return", "range", "(", "$", "min", ",", "$", "max", ",", "$", "interval", ")", ";", "}" ]
Generate a grid of evenly distributed parameters. @param float $min @param float $max @param int $n @throws \InvalidArgumentException @return float[]
[ "Generate", "a", "grid", "of", "evenly", "distributed", "parameters", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Other/Helpers/Params.php#L104-L119
RubixML/RubixML
src/Other/Helpers/Params.php
Params.args
public static function args($object) : array { if (!is_object($object)) { throw new InvalidArgumentException('Argument must be' . ' an object ' . gettype($object) . ' found.'); } $reflector = new ReflectionClass($object); $constructor = $reflector->getConstructor(); if ($constructor instanceof ReflectionMethod) { $args = array_column($constructor->getParameters(), 'name'); } else { $args = []; } return $args; }
php
public static function args($object) : array { if (!is_object($object)) { throw new InvalidArgumentException('Argument must be' . ' an object ' . gettype($object) . ' found.'); } $reflector = new ReflectionClass($object); $constructor = $reflector->getConstructor(); if ($constructor instanceof ReflectionMethod) { $args = array_column($constructor->getParameters(), 'name'); } else { $args = []; } return $args; }
[ "public", "static", "function", "args", "(", "$", "object", ")", ":", "array", "{", "if", "(", "!", "is_object", "(", "$", "object", ")", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'Argument must be'", ".", "' an object '", ".", "gettype", "(", "$", "object", ")", ".", "' found.'", ")", ";", "}", "$", "reflector", "=", "new", "ReflectionClass", "(", "$", "object", ")", ";", "$", "constructor", "=", "$", "reflector", "->", "getConstructor", "(", ")", ";", "if", "(", "$", "constructor", "instanceof", "ReflectionMethod", ")", "{", "$", "args", "=", "array_column", "(", "$", "constructor", "->", "getParameters", "(", ")", ",", "'name'", ")", ";", "}", "else", "{", "$", "args", "=", "[", "]", ";", "}", "return", "$", "args", ";", "}" ]
Extract the arguments from the model constructor for display. @param object $object @throws \InvalidArgumentException @return string[]
[ "Extract", "the", "arguments", "from", "the", "model", "constructor", "for", "display", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Other/Helpers/Params.php#L128-L146
RubixML/RubixML
src/Other/Helpers/Params.php
Params.stringify
public static function stringify(array $constructor, string $equator = '=', string $separator = ' ') : string { $strings = []; foreach ($constructor as $arg => $param) { if (is_object($param)) { $param = self::shortName($param); } if (is_array($param)) { $temp = array_combine(array_keys($param), $param) ?: []; $param = '[' . self::stringify($temp) . ']'; } $strings[] = (string) $arg . $equator . (string) $param; } return implode($separator, $strings); }
php
public static function stringify(array $constructor, string $equator = '=', string $separator = ' ') : string { $strings = []; foreach ($constructor as $arg => $param) { if (is_object($param)) { $param = self::shortName($param); } if (is_array($param)) { $temp = array_combine(array_keys($param), $param) ?: []; $param = '[' . self::stringify($temp) . ']'; } $strings[] = (string) $arg . $equator . (string) $param; } return implode($separator, $strings); }
[ "public", "static", "function", "stringify", "(", "array", "$", "constructor", ",", "string", "$", "equator", "=", "'='", ",", "string", "$", "separator", "=", "' '", ")", ":", "string", "{", "$", "strings", "=", "[", "]", ";", "foreach", "(", "$", "constructor", "as", "$", "arg", "=>", "$", "param", ")", "{", "if", "(", "is_object", "(", "$", "param", ")", ")", "{", "$", "param", "=", "self", "::", "shortName", "(", "$", "param", ")", ";", "}", "if", "(", "is_array", "(", "$", "param", ")", ")", "{", "$", "temp", "=", "array_combine", "(", "array_keys", "(", "$", "param", ")", ",", "$", "param", ")", "?", ":", "[", "]", ";", "$", "param", "=", "'['", ".", "self", "::", "stringify", "(", "$", "temp", ")", ".", "']'", ";", "}", "$", "strings", "[", "]", "=", "(", "string", ")", "$", "arg", ".", "$", "equator", ".", "(", "string", ")", "$", "param", ";", "}", "return", "implode", "(", "$", "separator", ",", "$", "strings", ")", ";", "}" ]
Return a string representation of the constructor arguments from an associative constructor array. @param array $constructor @param string $equator @param string $separator @return string
[ "Return", "a", "string", "representation", "of", "the", "constructor", "arguments", "from", "an", "associative", "constructor", "array", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Other/Helpers/Params.php#L157-L176
RubixML/RubixML
src/Other/Helpers/Params.php
Params.shortName
public static function shortName($object) : string { if (!is_object($object)) { throw new InvalidArgumentException('Must provide an object' . gettype($object) . ' given.'); } return substr(strrchr(get_class($object), '\\') ?: '', 1); }
php
public static function shortName($object) : string { if (!is_object($object)) { throw new InvalidArgumentException('Must provide an object' . gettype($object) . ' given.'); } return substr(strrchr(get_class($object), '\\') ?: '', 1); }
[ "public", "static", "function", "shortName", "(", "$", "object", ")", ":", "string", "{", "if", "(", "!", "is_object", "(", "$", "object", ")", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'Must provide an object'", ".", "gettype", "(", "$", "object", ")", ".", "' given.'", ")", ";", "}", "return", "substr", "(", "strrchr", "(", "get_class", "(", "$", "object", ")", ",", "'\\\\'", ")", "?", ":", "''", ",", "1", ")", ";", "}" ]
Return the short class name from a fully qualified class name (fqcn). @param mixed $object @throws \InvalidArgumentException @return string
[ "Return", "the", "short", "class", "name", "from", "a", "fully", "qualified", "class", "name", "(", "fqcn", ")", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Other/Helpers/Params.php#L186-L194
RubixML/RubixML
src/CommitteeMachine.php
CommitteeMachine.train
public function train(Dataset $dataset) : void { if ($this->type === self::CLASSIFIER or $this->type === self::REGRESSOR) { if (!$dataset instanceof Labeled) { throw new InvalidArgumentException('This estimator requires a' . ' labeled training set.'); } } DatasetIsCompatibleWithEstimator::check($dataset, $this); if ($this->logger) { $this->logger->info('Learner init ' . Params::stringify([ 'experts' => $this->experts, 'influences' => $this->influences, 'workers' => $this->workers, ])); } Loop::run(function () use ($dataset) { $pool = new DefaultPool($this->workers); $coroutines = []; foreach ($this->experts as $index => $estimator) { $task = new CallableTask( [$this, '_train'], [$estimator, $dataset] ); $coroutines[] = call(function () use ($pool, $task, $index) { $estimator = yield $pool->enqueue($task); if ($this->logger) { $this->logger->info(Params::stringify([ $index => $estimator, ]) . ' finished training'); } return $estimator; }); } $this->experts = yield all($coroutines); return yield $pool->shutdown(); }); if ($this->type === self::CLASSIFIER and $dataset instanceof Labeled) { $this->classes = $dataset->possibleOutcomes(); } if ($this->logger) { $this->logger->info('Training complete'); } }
php
public function train(Dataset $dataset) : void { if ($this->type === self::CLASSIFIER or $this->type === self::REGRESSOR) { if (!$dataset instanceof Labeled) { throw new InvalidArgumentException('This estimator requires a' . ' labeled training set.'); } } DatasetIsCompatibleWithEstimator::check($dataset, $this); if ($this->logger) { $this->logger->info('Learner init ' . Params::stringify([ 'experts' => $this->experts, 'influences' => $this->influences, 'workers' => $this->workers, ])); } Loop::run(function () use ($dataset) { $pool = new DefaultPool($this->workers); $coroutines = []; foreach ($this->experts as $index => $estimator) { $task = new CallableTask( [$this, '_train'], [$estimator, $dataset] ); $coroutines[] = call(function () use ($pool, $task, $index) { $estimator = yield $pool->enqueue($task); if ($this->logger) { $this->logger->info(Params::stringify([ $index => $estimator, ]) . ' finished training'); } return $estimator; }); } $this->experts = yield all($coroutines); return yield $pool->shutdown(); }); if ($this->type === self::CLASSIFIER and $dataset instanceof Labeled) { $this->classes = $dataset->possibleOutcomes(); } if ($this->logger) { $this->logger->info('Training complete'); } }
[ "public", "function", "train", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "if", "(", "$", "this", "->", "type", "===", "self", "::", "CLASSIFIER", "or", "$", "this", "->", "type", "===", "self", "::", "REGRESSOR", ")", "{", "if", "(", "!", "$", "dataset", "instanceof", "Labeled", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'This estimator requires a'", ".", "' labeled training set.'", ")", ";", "}", "}", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "'Learner init '", ".", "Params", "::", "stringify", "(", "[", "'experts'", "=>", "$", "this", "->", "experts", ",", "'influences'", "=>", "$", "this", "->", "influences", ",", "'workers'", "=>", "$", "this", "->", "workers", ",", "]", ")", ")", ";", "}", "Loop", "::", "run", "(", "function", "(", ")", "use", "(", "$", "dataset", ")", "{", "$", "pool", "=", "new", "DefaultPool", "(", "$", "this", "->", "workers", ")", ";", "$", "coroutines", "=", "[", "]", ";", "foreach", "(", "$", "this", "->", "experts", "as", "$", "index", "=>", "$", "estimator", ")", "{", "$", "task", "=", "new", "CallableTask", "(", "[", "$", "this", ",", "'_train'", "]", ",", "[", "$", "estimator", ",", "$", "dataset", "]", ")", ";", "$", "coroutines", "[", "]", "=", "call", "(", "function", "(", ")", "use", "(", "$", "pool", ",", "$", "task", ",", "$", "index", ")", "{", "$", "estimator", "=", "yield", "$", "pool", "->", "enqueue", "(", "$", "task", ")", ";", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "Params", "::", "stringify", "(", "[", "$", "index", "=>", "$", "estimator", ",", "]", ")", ".", "' finished training'", ")", ";", "}", "return", "$", "estimator", ";", "}", ")", ";", "}", "$", "this", "->", "experts", "=", "yield", "all", "(", "$", "coroutines", ")", ";", "return", "yield", "$", "pool", "->", "shutdown", "(", ")", ";", "}", ")", ";", "if", "(", "$", "this", "->", "type", "===", "self", "::", "CLASSIFIER", "and", "$", "dataset", "instanceof", "Labeled", ")", "{", "$", "this", "->", "classes", "=", "$", "dataset", "->", "possibleOutcomes", "(", ")", ";", "}", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "'Training complete'", ")", ";", "}", "}" ]
Train all the experts with the dataset. @param \Rubix\ML\Datasets\Dataset $dataset @throws \InvalidArgumentException
[ "Train", "all", "the", "experts", "with", "the", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/CommitteeMachine.php#L215-L270
RubixML/RubixML
src/CommitteeMachine.php
CommitteeMachine.predict
public function predict(Dataset $dataset) : array { $aggregate = []; foreach ($this->experts as $estimator) { foreach ($estimator->predict($dataset) as $i => $prediction) { $aggregate[$i][] = $prediction; } } switch ($this->type) { case self::CLASSIFIER: return array_map([$this, 'decideClass'], $aggregate); case self::REGRESSOR: return array_map([$this, 'decideValue'], $aggregate); case self::ANOMALY_DETECTOR: return array_map([$this, 'decideAnomaly'], $aggregate); } }
php
public function predict(Dataset $dataset) : array { $aggregate = []; foreach ($this->experts as $estimator) { foreach ($estimator->predict($dataset) as $i => $prediction) { $aggregate[$i][] = $prediction; } } switch ($this->type) { case self::CLASSIFIER: return array_map([$this, 'decideClass'], $aggregate); case self::REGRESSOR: return array_map([$this, 'decideValue'], $aggregate); case self::ANOMALY_DETECTOR: return array_map([$this, 'decideAnomaly'], $aggregate); } }
[ "public", "function", "predict", "(", "Dataset", "$", "dataset", ")", ":", "array", "{", "$", "aggregate", "=", "[", "]", ";", "foreach", "(", "$", "this", "->", "experts", "as", "$", "estimator", ")", "{", "foreach", "(", "$", "estimator", "->", "predict", "(", "$", "dataset", ")", "as", "$", "i", "=>", "$", "prediction", ")", "{", "$", "aggregate", "[", "$", "i", "]", "[", "]", "=", "$", "prediction", ";", "}", "}", "switch", "(", "$", "this", "->", "type", ")", "{", "case", "self", "::", "CLASSIFIER", ":", "return", "array_map", "(", "[", "$", "this", ",", "'decideClass'", "]", ",", "$", "aggregate", ")", ";", "case", "self", "::", "REGRESSOR", ":", "return", "array_map", "(", "[", "$", "this", ",", "'decideValue'", "]", ",", "$", "aggregate", ")", ";", "case", "self", "::", "ANOMALY_DETECTOR", ":", "return", "array_map", "(", "[", "$", "this", ",", "'decideAnomaly'", "]", ",", "$", "aggregate", ")", ";", "}", "}" ]
Make predictions from a dataset. @param \Rubix\ML\Datasets\Dataset $dataset @return array
[ "Make", "predictions", "from", "a", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/CommitteeMachine.php#L278-L298
RubixML/RubixML
src/CommitteeMachine.php
CommitteeMachine.decideClass
public function decideClass(array $votes) { $scores = array_fill_keys($this->classes, 0.); foreach ($votes as $i => $vote) { $scores[$vote] += $this->influences[$i]; } return argmax($scores); }
php
public function decideClass(array $votes) { $scores = array_fill_keys($this->classes, 0.); foreach ($votes as $i => $vote) { $scores[$vote] += $this->influences[$i]; } return argmax($scores); }
[ "public", "function", "decideClass", "(", "array", "$", "votes", ")", "{", "$", "scores", "=", "array_fill_keys", "(", "$", "this", "->", "classes", ",", "0.", ")", ";", "foreach", "(", "$", "votes", "as", "$", "i", "=>", "$", "vote", ")", "{", "$", "scores", "[", "$", "vote", "]", "+=", "$", "this", "->", "influences", "[", "$", "i", "]", ";", "}", "return", "argmax", "(", "$", "scores", ")", ";", "}" ]
Decide on a class outcome. @param (int|string)[] $votes @return int|string
[ "Decide", "on", "a", "class", "outcome", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/CommitteeMachine.php#L306-L315
RubixML/RubixML
src/CommitteeMachine.php
CommitteeMachine.decideAnomaly
public function decideAnomaly(array $votes) : int { $scores = array_fill(0, 2, 0.); foreach ($votes as $i => $vote) { $scores[$vote] += $this->influences[$i]; } return argmax($scores); }
php
public function decideAnomaly(array $votes) : int { $scores = array_fill(0, 2, 0.); foreach ($votes as $i => $vote) { $scores[$vote] += $this->influences[$i]; } return argmax($scores); }
[ "public", "function", "decideAnomaly", "(", "array", "$", "votes", ")", ":", "int", "{", "$", "scores", "=", "array_fill", "(", "0", ",", "2", ",", "0.", ")", ";", "foreach", "(", "$", "votes", "as", "$", "i", "=>", "$", "vote", ")", "{", "$", "scores", "[", "$", "vote", "]", "+=", "$", "this", "->", "influences", "[", "$", "i", "]", ";", "}", "return", "argmax", "(", "$", "scores", ")", ";", "}" ]
Decide on an anomaly outcome. @param int[] $votes @return int
[ "Decide", "on", "an", "anomaly", "outcome", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/CommitteeMachine.php#L334-L343
RubixML/RubixML
src/CommitteeMachine.php
CommitteeMachine._train
public function _train(Learner $estimator, Dataset $dataset) : Learner { $estimator->train($dataset); return $estimator; }
php
public function _train(Learner $estimator, Dataset $dataset) : Learner { $estimator->train($dataset); return $estimator; }
[ "public", "function", "_train", "(", "Learner", "$", "estimator", ",", "Dataset", "$", "dataset", ")", ":", "Learner", "{", "$", "estimator", "->", "train", "(", "$", "dataset", ")", ";", "return", "$", "estimator", ";", "}" ]
Train an learner using a dataset and return it. @param \Rubix\ML\Learner $estimator @param \Rubix\ML\Datasets\Dataset $dataset @return \Rubix\ML\Learner
[ "Train", "an", "learner", "using", "a", "dataset", "and", "return", "it", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/CommitteeMachine.php#L352-L357
RubixML/RubixML
src/NeuralNet/Deferred.php
Deferred.result
public function result() : Matrix { if (!$this->result) { $this->result = call_user_func($this->computation); } return $this->result; }
php
public function result() : Matrix { if (!$this->result) { $this->result = call_user_func($this->computation); } return $this->result; }
[ "public", "function", "result", "(", ")", ":", "Matrix", "{", "if", "(", "!", "$", "this", "->", "result", ")", "{", "$", "this", "->", "result", "=", "call_user_func", "(", "$", "this", "->", "computation", ")", ";", "}", "return", "$", "this", "->", "result", ";", "}" ]
Return the result of the computation. @return \Rubix\Tensor\Matrix
[ "Return", "the", "result", "of", "the", "computation", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/NeuralNet/Deferred.php#L49-L56
RubixML/RubixML
src/Transformers/MaxAbsoluteScaler.php
MaxAbsoluteScaler.fit
public function fit(Dataset $dataset) : void { $this->maxabs = []; foreach ($dataset->types() as $column => $type) { if ($type === DataType::CONTINUOUS) { $this->maxabs[$column] = -INF; } } $this->update($dataset); }
php
public function fit(Dataset $dataset) : void { $this->maxabs = []; foreach ($dataset->types() as $column => $type) { if ($type === DataType::CONTINUOUS) { $this->maxabs[$column] = -INF; } } $this->update($dataset); }
[ "public", "function", "fit", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "$", "this", "->", "maxabs", "=", "[", "]", ";", "foreach", "(", "$", "dataset", "->", "types", "(", ")", "as", "$", "column", "=>", "$", "type", ")", "{", "if", "(", "$", "type", "===", "DataType", "::", "CONTINUOUS", ")", "{", "$", "this", "->", "maxabs", "[", "$", "column", "]", "=", "-", "INF", ";", "}", "}", "$", "this", "->", "update", "(", "$", "dataset", ")", ";", "}" ]
Fit the transformer to the dataset. @param \Rubix\ML\Datasets\Dataset $dataset
[ "Fit", "the", "transformer", "to", "the", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/MaxAbsoluteScaler.php#L55-L66
RubixML/RubixML
src/Transformers/MaxAbsoluteScaler.php
MaxAbsoluteScaler.update
public function update(Dataset $dataset) : void { if ($this->maxabs === null) { $this->fit($dataset); return; } foreach ($this->maxabs as $column => $oldMax) { $values = $dataset->column($column); $max = max(array_map('abs', $values)); $max = max($oldMax, $max); $this->maxabs[$column] = $max ?: EPSILON; } }
php
public function update(Dataset $dataset) : void { if ($this->maxabs === null) { $this->fit($dataset); return; } foreach ($this->maxabs as $column => $oldMax) { $values = $dataset->column($column); $max = max(array_map('abs', $values)); $max = max($oldMax, $max); $this->maxabs[$column] = $max ?: EPSILON; } }
[ "public", "function", "update", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "if", "(", "$", "this", "->", "maxabs", "===", "null", ")", "{", "$", "this", "->", "fit", "(", "$", "dataset", ")", ";", "return", ";", "}", "foreach", "(", "$", "this", "->", "maxabs", "as", "$", "column", "=>", "$", "oldMax", ")", "{", "$", "values", "=", "$", "dataset", "->", "column", "(", "$", "column", ")", ";", "$", "max", "=", "max", "(", "array_map", "(", "'abs'", ",", "$", "values", ")", ")", ";", "$", "max", "=", "max", "(", "$", "oldMax", ",", "$", "max", ")", ";", "$", "this", "->", "maxabs", "[", "$", "column", "]", "=", "$", "max", "?", ":", "EPSILON", ";", "}", "}" ]
Update the fitting of the transformer. @param \Rubix\ML\Datasets\Dataset $dataset
[ "Update", "the", "fitting", "of", "the", "transformer", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/MaxAbsoluteScaler.php#L73-L90
RubixML/RubixML
src/Transformers/MaxAbsoluteScaler.php
MaxAbsoluteScaler.transform
public function transform(array &$samples) : void { if ($this->maxabs === null) { throw new RuntimeException('Transformer has not been fitted.'); } foreach ($samples as &$sample) { foreach ($this->maxabs as $column => $maxabs) { $sample[$column] /= $maxabs; } } }
php
public function transform(array &$samples) : void { if ($this->maxabs === null) { throw new RuntimeException('Transformer has not been fitted.'); } foreach ($samples as &$sample) { foreach ($this->maxabs as $column => $maxabs) { $sample[$column] /= $maxabs; } } }
[ "public", "function", "transform", "(", "array", "&", "$", "samples", ")", ":", "void", "{", "if", "(", "$", "this", "->", "maxabs", "===", "null", ")", "{", "throw", "new", "RuntimeException", "(", "'Transformer has not been fitted.'", ")", ";", "}", "foreach", "(", "$", "samples", "as", "&", "$", "sample", ")", "{", "foreach", "(", "$", "this", "->", "maxabs", "as", "$", "column", "=>", "$", "maxabs", ")", "{", "$", "sample", "[", "$", "column", "]", "/=", "$", "maxabs", ";", "}", "}", "}" ]
Transform the dataset in place. @param array $samples @throws \RuntimeException
[ "Transform", "the", "dataset", "in", "place", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/MaxAbsoluteScaler.php#L98-L109
RubixML/RubixML
src/NeuralNet/CostFunctions/HuberLoss.php
HuberLoss.compute
public function compute(Tensor $expected, Tensor $output) : float { return $expected->subtract($output) ->map([$this, '_compute'])->sum()->mean(); }
php
public function compute(Tensor $expected, Tensor $output) : float { return $expected->subtract($output) ->map([$this, '_compute'])->sum()->mean(); }
[ "public", "function", "compute", "(", "Tensor", "$", "expected", ",", "Tensor", "$", "output", ")", ":", "float", "{", "return", "$", "expected", "->", "subtract", "(", "$", "output", ")", "->", "map", "(", "[", "$", "this", ",", "'_compute'", "]", ")", "->", "sum", "(", ")", "->", "mean", "(", ")", ";", "}" ]
Compute the loss. @param \Rubix\Tensor\Tensor $expected @param \Rubix\Tensor\Tensor $output @return float
[ "Compute", "the", "loss", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/NeuralNet/CostFunctions/HuberLoss.php#L69-L73
RubixML/RubixML
src/NeuralNet/CostFunctions/HuberLoss.php
HuberLoss.differentiate
public function differentiate(Tensor $expected, Tensor $output) : Tensor { $alpha = $output->subtract($expected); return $alpha->square() ->add($this->beta) ->pow(-0.5) ->multiply($alpha); }
php
public function differentiate(Tensor $expected, Tensor $output) : Tensor { $alpha = $output->subtract($expected); return $alpha->square() ->add($this->beta) ->pow(-0.5) ->multiply($alpha); }
[ "public", "function", "differentiate", "(", "Tensor", "$", "expected", ",", "Tensor", "$", "output", ")", ":", "Tensor", "{", "$", "alpha", "=", "$", "output", "->", "subtract", "(", "$", "expected", ")", ";", "return", "$", "alpha", "->", "square", "(", ")", "->", "add", "(", "$", "this", "->", "beta", ")", "->", "pow", "(", "-", "0.5", ")", "->", "multiply", "(", "$", "alpha", ")", ";", "}" ]
Calculate the gradient of the cost function with respect to the output. @param \Rubix\Tensor\Tensor $expected @param \Rubix\Tensor\Tensor $output @return \Rubix\Tensor\Tensor
[ "Calculate", "the", "gradient", "of", "the", "cost", "function", "with", "respect", "to", "the", "output", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/NeuralNet/CostFunctions/HuberLoss.php#L82-L90
RubixML/RubixML
src/Datasets/Generators/HalfMoon.php
HalfMoon.generate
public function generate(int $n) : Dataset { $r = Vector::rand($n)->multiply(180) ->add($this->rotation) ->radians(); $x = $r->cos(); $y = $r->sin(); $noise = Matrix::gaussian($n, 2) ->multiply($this->noise); $samples = Matrix::stack([$x, $y]) ->transpose() ->add($noise) ->multiply($this->scale) ->add($this->center) ->asArray(); $labels = $r->degrees()->asArray(); return Labeled::quick($samples, $labels); }
php
public function generate(int $n) : Dataset { $r = Vector::rand($n)->multiply(180) ->add($this->rotation) ->radians(); $x = $r->cos(); $y = $r->sin(); $noise = Matrix::gaussian($n, 2) ->multiply($this->noise); $samples = Matrix::stack([$x, $y]) ->transpose() ->add($noise) ->multiply($this->scale) ->add($this->center) ->asArray(); $labels = $r->degrees()->asArray(); return Labeled::quick($samples, $labels); }
[ "public", "function", "generate", "(", "int", "$", "n", ")", ":", "Dataset", "{", "$", "r", "=", "Vector", "::", "rand", "(", "$", "n", ")", "->", "multiply", "(", "180", ")", "->", "add", "(", "$", "this", "->", "rotation", ")", "->", "radians", "(", ")", ";", "$", "x", "=", "$", "r", "->", "cos", "(", ")", ";", "$", "y", "=", "$", "r", "->", "sin", "(", ")", ";", "$", "noise", "=", "Matrix", "::", "gaussian", "(", "$", "n", ",", "2", ")", "->", "multiply", "(", "$", "this", "->", "noise", ")", ";", "$", "samples", "=", "Matrix", "::", "stack", "(", "[", "$", "x", ",", "$", "y", "]", ")", "->", "transpose", "(", ")", "->", "add", "(", "$", "noise", ")", "->", "multiply", "(", "$", "this", "->", "scale", ")", "->", "add", "(", "$", "this", "->", "center", ")", "->", "asArray", "(", ")", ";", "$", "labels", "=", "$", "r", "->", "degrees", "(", ")", "->", "asArray", "(", ")", ";", "return", "Labeled", "::", "quick", "(", "$", "samples", ",", "$", "labels", ")", ";", "}" ]
Generate n data points. @param int $n @return \Rubix\ML\Datasets\Dataset
[ "Generate", "n", "data", "points", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Datasets/Generators/HalfMoon.php#L102-L124
RubixML/RubixML
src/Classifiers/GaussianNB.php
GaussianNB.train
public function train(Dataset $dataset) : void { if (!$dataset instanceof Labeled) { throw new InvalidArgumentException('This Estimator requires a' . ' Labeled training set.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); $classes = $dataset->possibleOutcomes(); $this->classes = $classes; $this->weights = array_fill_keys($classes, 0); $this->means = $this->variances = array_fill_keys($classes, []); foreach ($dataset->stratify() as $class => $stratum) { $means = $variances = []; foreach ($stratum->columns() as $values) { [$mean, $variance] = Stats::meanVar($values); $means[] = $mean; $variances[] = $variance ?: EPSILON; } $this->means[$class] = $means; $this->variances[$class] = $variances; $this->weights[$class] += $stratum->numRows(); } if ($this->fitPriors) { $this->priors = []; $total = array_sum($this->weights) ?: EPSILON; foreach ($this->weights as $class => $weight) { $this->priors[$class] = log($weight / $total); } } }
php
public function train(Dataset $dataset) : void { if (!$dataset instanceof Labeled) { throw new InvalidArgumentException('This Estimator requires a' . ' Labeled training set.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); $classes = $dataset->possibleOutcomes(); $this->classes = $classes; $this->weights = array_fill_keys($classes, 0); $this->means = $this->variances = array_fill_keys($classes, []); foreach ($dataset->stratify() as $class => $stratum) { $means = $variances = []; foreach ($stratum->columns() as $values) { [$mean, $variance] = Stats::meanVar($values); $means[] = $mean; $variances[] = $variance ?: EPSILON; } $this->means[$class] = $means; $this->variances[$class] = $variances; $this->weights[$class] += $stratum->numRows(); } if ($this->fitPriors) { $this->priors = []; $total = array_sum($this->weights) ?: EPSILON; foreach ($this->weights as $class => $weight) { $this->priors[$class] = log($weight / $total); } } }
[ "public", "function", "train", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "if", "(", "!", "$", "dataset", "instanceof", "Labeled", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'This Estimator requires a'", ".", "' Labeled training set.'", ")", ";", "}", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "$", "classes", "=", "$", "dataset", "->", "possibleOutcomes", "(", ")", ";", "$", "this", "->", "classes", "=", "$", "classes", ";", "$", "this", "->", "weights", "=", "array_fill_keys", "(", "$", "classes", ",", "0", ")", ";", "$", "this", "->", "means", "=", "$", "this", "->", "variances", "=", "array_fill_keys", "(", "$", "classes", ",", "[", "]", ")", ";", "foreach", "(", "$", "dataset", "->", "stratify", "(", ")", "as", "$", "class", "=>", "$", "stratum", ")", "{", "$", "means", "=", "$", "variances", "=", "[", "]", ";", "foreach", "(", "$", "stratum", "->", "columns", "(", ")", "as", "$", "values", ")", "{", "[", "$", "mean", ",", "$", "variance", "]", "=", "Stats", "::", "meanVar", "(", "$", "values", ")", ";", "$", "means", "[", "]", "=", "$", "mean", ";", "$", "variances", "[", "]", "=", "$", "variance", "?", ":", "EPSILON", ";", "}", "$", "this", "->", "means", "[", "$", "class", "]", "=", "$", "means", ";", "$", "this", "->", "variances", "[", "$", "class", "]", "=", "$", "variances", ";", "$", "this", "->", "weights", "[", "$", "class", "]", "+=", "$", "stratum", "->", "numRows", "(", ")", ";", "}", "if", "(", "$", "this", "->", "fitPriors", ")", "{", "$", "this", "->", "priors", "=", "[", "]", ";", "$", "total", "=", "array_sum", "(", "$", "this", "->", "weights", ")", "?", ":", "EPSILON", ";", "foreach", "(", "$", "this", "->", "weights", "as", "$", "class", "=>", "$", "weight", ")", "{", "$", "this", "->", "priors", "[", "$", "class", "]", "=", "log", "(", "$", "weight", "/", "$", "total", ")", ";", "}", "}", "}" ]
Compute the necessary statistics to estimate a probability density for each feature column. @param \Rubix\ML\Datasets\Dataset $dataset @throws \InvalidArgumentException
[ "Compute", "the", "necessary", "statistics", "to", "estimate", "a", "probability", "density", "for", "each", "feature", "column", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Classifiers/GaussianNB.php#L207-L248
RubixML/RubixML
src/Classifiers/GaussianNB.php
GaussianNB.partial
public function partial(Dataset $dataset) : void { if (empty($this->weights) or empty($this->means) or empty($this->variances)) { $this->train($dataset); return; } if (!$dataset instanceof Labeled) { throw new InvalidArgumentException('This Estimator requires a' . ' Labeled training set.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); foreach ($dataset->stratify() as $class => $stratum) { $means = $this->means[$class]; $variances = $this->variances[$class]; $oldWeight = $this->weights[$class]; $oldMeans = $this->means[$class]; $oldVariances = $this->variances[$class]; $n = $stratum->numRows(); foreach ($stratum->columns() as $column => $values) { [$mean, $variance] = Stats::meanVar($values); $means[$column] = (($n * $mean) + ($oldWeight * $oldMeans[$column])) / ($oldWeight + $n); $vHat = ($oldWeight * $oldVariances[$column] + ($n * $variance) + ($oldWeight / ($n * ($oldWeight + $n))) * ($n * $oldMeans[$column] - $n * $mean) ** 2) / ($oldWeight + $n); $variances[$column] = $vHat ?: EPSILON; } $this->means[$class] = $means; $this->variances[$class] = $variances; $this->weights[$class] += $n; } if ($this->fitPriors) { $total = array_sum($this->weights) ?: EPSILON; foreach ($this->weights as $class => $weight) { $this->priors[$class] = log($weight / $total); } } }
php
public function partial(Dataset $dataset) : void { if (empty($this->weights) or empty($this->means) or empty($this->variances)) { $this->train($dataset); return; } if (!$dataset instanceof Labeled) { throw new InvalidArgumentException('This Estimator requires a' . ' Labeled training set.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); foreach ($dataset->stratify() as $class => $stratum) { $means = $this->means[$class]; $variances = $this->variances[$class]; $oldWeight = $this->weights[$class]; $oldMeans = $this->means[$class]; $oldVariances = $this->variances[$class]; $n = $stratum->numRows(); foreach ($stratum->columns() as $column => $values) { [$mean, $variance] = Stats::meanVar($values); $means[$column] = (($n * $mean) + ($oldWeight * $oldMeans[$column])) / ($oldWeight + $n); $vHat = ($oldWeight * $oldVariances[$column] + ($n * $variance) + ($oldWeight / ($n * ($oldWeight + $n))) * ($n * $oldMeans[$column] - $n * $mean) ** 2) / ($oldWeight + $n); $variances[$column] = $vHat ?: EPSILON; } $this->means[$class] = $means; $this->variances[$class] = $variances; $this->weights[$class] += $n; } if ($this->fitPriors) { $total = array_sum($this->weights) ?: EPSILON; foreach ($this->weights as $class => $weight) { $this->priors[$class] = log($weight / $total); } } }
[ "public", "function", "partial", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "if", "(", "empty", "(", "$", "this", "->", "weights", ")", "or", "empty", "(", "$", "this", "->", "means", ")", "or", "empty", "(", "$", "this", "->", "variances", ")", ")", "{", "$", "this", "->", "train", "(", "$", "dataset", ")", ";", "return", ";", "}", "if", "(", "!", "$", "dataset", "instanceof", "Labeled", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'This Estimator requires a'", ".", "' Labeled training set.'", ")", ";", "}", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "foreach", "(", "$", "dataset", "->", "stratify", "(", ")", "as", "$", "class", "=>", "$", "stratum", ")", "{", "$", "means", "=", "$", "this", "->", "means", "[", "$", "class", "]", ";", "$", "variances", "=", "$", "this", "->", "variances", "[", "$", "class", "]", ";", "$", "oldWeight", "=", "$", "this", "->", "weights", "[", "$", "class", "]", ";", "$", "oldMeans", "=", "$", "this", "->", "means", "[", "$", "class", "]", ";", "$", "oldVariances", "=", "$", "this", "->", "variances", "[", "$", "class", "]", ";", "$", "n", "=", "$", "stratum", "->", "numRows", "(", ")", ";", "foreach", "(", "$", "stratum", "->", "columns", "(", ")", "as", "$", "column", "=>", "$", "values", ")", "{", "[", "$", "mean", ",", "$", "variance", "]", "=", "Stats", "::", "meanVar", "(", "$", "values", ")", ";", "$", "means", "[", "$", "column", "]", "=", "(", "(", "$", "n", "*", "$", "mean", ")", "+", "(", "$", "oldWeight", "*", "$", "oldMeans", "[", "$", "column", "]", ")", ")", "/", "(", "$", "oldWeight", "+", "$", "n", ")", ";", "$", "vHat", "=", "(", "$", "oldWeight", "*", "$", "oldVariances", "[", "$", "column", "]", "+", "(", "$", "n", "*", "$", "variance", ")", "+", "(", "$", "oldWeight", "/", "(", "$", "n", "*", "(", "$", "oldWeight", "+", "$", "n", ")", ")", ")", "*", "(", "$", "n", "*", "$", "oldMeans", "[", "$", "column", "]", "-", "$", "n", "*", "$", "mean", ")", "**", "2", ")", "/", "(", "$", "oldWeight", "+", "$", "n", ")", ";", "$", "variances", "[", "$", "column", "]", "=", "$", "vHat", "?", ":", "EPSILON", ";", "}", "$", "this", "->", "means", "[", "$", "class", "]", "=", "$", "means", ";", "$", "this", "->", "variances", "[", "$", "class", "]", "=", "$", "variances", ";", "$", "this", "->", "weights", "[", "$", "class", "]", "+=", "$", "n", ";", "}", "if", "(", "$", "this", "->", "fitPriors", ")", "{", "$", "total", "=", "array_sum", "(", "$", "this", "->", "weights", ")", "?", ":", "EPSILON", ";", "foreach", "(", "$", "this", "->", "weights", "as", "$", "class", "=>", "$", "weight", ")", "{", "$", "this", "->", "priors", "[", "$", "class", "]", "=", "log", "(", "$", "weight", "/", "$", "total", ")", ";", "}", "}", "}" ]
Uupdate the rolling means and variances of each feature column using an online updating algorithm. @param \Rubix\ML\Datasets\Dataset $dataset @throws \InvalidArgumentException
[ "Uupdate", "the", "rolling", "means", "and", "variances", "of", "each", "feature", "column", "using", "an", "online", "updating", "algorithm", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Classifiers/GaussianNB.php#L257-L311
RubixML/RubixML
src/Classifiers/GaussianNB.php
GaussianNB.jointLogLikelihood
protected function jointLogLikelihood(array $sample) : array { $likelihoods = []; foreach ($this->means as $class => $means) { $variances = $this->variances[$class]; $likelihood = $this->priors[$class] ?? LOG_EPSILON; foreach ($sample as $column => $feature) { $mean = $means[$column]; $variance = $variances[$column]; $pdf = -0.5 * log(TWO_PI * $variance); $pdf -= 0.5 * (($feature - $mean) ** 2) / $variance; $likelihood += $pdf; } $likelihoods[$class] = $likelihood; } return $likelihoods; }
php
protected function jointLogLikelihood(array $sample) : array { $likelihoods = []; foreach ($this->means as $class => $means) { $variances = $this->variances[$class]; $likelihood = $this->priors[$class] ?? LOG_EPSILON; foreach ($sample as $column => $feature) { $mean = $means[$column]; $variance = $variances[$column]; $pdf = -0.5 * log(TWO_PI * $variance); $pdf -= 0.5 * (($feature - $mean) ** 2) / $variance; $likelihood += $pdf; } $likelihoods[$class] = $likelihood; } return $likelihoods; }
[ "protected", "function", "jointLogLikelihood", "(", "array", "$", "sample", ")", ":", "array", "{", "$", "likelihoods", "=", "[", "]", ";", "foreach", "(", "$", "this", "->", "means", "as", "$", "class", "=>", "$", "means", ")", "{", "$", "variances", "=", "$", "this", "->", "variances", "[", "$", "class", "]", ";", "$", "likelihood", "=", "$", "this", "->", "priors", "[", "$", "class", "]", "??", "LOG_EPSILON", ";", "foreach", "(", "$", "sample", "as", "$", "column", "=>", "$", "feature", ")", "{", "$", "mean", "=", "$", "means", "[", "$", "column", "]", ";", "$", "variance", "=", "$", "variances", "[", "$", "column", "]", ";", "$", "pdf", "=", "-", "0.5", "*", "log", "(", "TWO_PI", "*", "$", "variance", ")", ";", "$", "pdf", "-=", "0.5", "*", "(", "(", "$", "feature", "-", "$", "mean", ")", "**", "2", ")", "/", "$", "variance", ";", "$", "likelihood", "+=", "$", "pdf", ";", "}", "$", "likelihoods", "[", "$", "class", "]", "=", "$", "likelihood", ";", "}", "return", "$", "likelihoods", ";", "}" ]
Calculate the joint log likelihood of a sample being a member of each class. @param array $sample @return array
[ "Calculate", "the", "joint", "log", "likelihood", "of", "a", "sample", "being", "a", "member", "of", "each", "class", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Classifiers/GaussianNB.php#L384-L407
RubixML/RubixML
src/NeuralNet/Layers/BatchNorm.php
BatchNorm.parameters
public function parameters() : Generator { if (!$this->beta or !$this->gamma) { throw new RuntimeException('Layer has not been initilaized.'); } yield $this->beta; yield $this->gamma; }
php
public function parameters() : Generator { if (!$this->beta or !$this->gamma) { throw new RuntimeException('Layer has not been initilaized.'); } yield $this->beta; yield $this->gamma; }
[ "public", "function", "parameters", "(", ")", ":", "Generator", "{", "if", "(", "!", "$", "this", "->", "beta", "or", "!", "$", "this", "->", "gamma", ")", "{", "throw", "new", "RuntimeException", "(", "'Layer has not been initilaized.'", ")", ";", "}", "yield", "$", "this", "->", "beta", ";", "yield", "$", "this", "->", "gamma", ";", "}" ]
Return the parameters of the layer. @throws \RuntimeException @return \Generator
[ "Return", "the", "parameters", "of", "the", "layer", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/NeuralNet/Layers/BatchNorm.php#L151-L159
RubixML/RubixML
src/NeuralNet/Layers/BatchNorm.php
BatchNorm.initialize
public function initialize(int $fanIn) : int { $fanOut = $fanIn; $beta = $this->betaInitializer->initialize(1, $fanOut)->columnAsVector(0); $gamma = $this->gammaInitializer->initialize(1, $fanOut)->columnAsVector(0); $this->beta = new VectorParam($beta); $this->gamma = new VectorParam($gamma); $this->width = $fanOut; return $fanOut; }
php
public function initialize(int $fanIn) : int { $fanOut = $fanIn; $beta = $this->betaInitializer->initialize(1, $fanOut)->columnAsVector(0); $gamma = $this->gammaInitializer->initialize(1, $fanOut)->columnAsVector(0); $this->beta = new VectorParam($beta); $this->gamma = new VectorParam($gamma); $this->width = $fanOut; return $fanOut; }
[ "public", "function", "initialize", "(", "int", "$", "fanIn", ")", ":", "int", "{", "$", "fanOut", "=", "$", "fanIn", ";", "$", "beta", "=", "$", "this", "->", "betaInitializer", "->", "initialize", "(", "1", ",", "$", "fanOut", ")", "->", "columnAsVector", "(", "0", ")", ";", "$", "gamma", "=", "$", "this", "->", "gammaInitializer", "->", "initialize", "(", "1", ",", "$", "fanOut", ")", "->", "columnAsVector", "(", "0", ")", ";", "$", "this", "->", "beta", "=", "new", "VectorParam", "(", "$", "beta", ")", ";", "$", "this", "->", "gamma", "=", "new", "VectorParam", "(", "$", "gamma", ")", ";", "$", "this", "->", "width", "=", "$", "fanOut", ";", "return", "$", "fanOut", ";", "}" ]
Initialize the layer with the fan in from the previous layer and return the fan out for this layer. @param int $fanIn @return int
[ "Initialize", "the", "layer", "with", "the", "fan", "in", "from", "the", "previous", "layer", "and", "return", "the", "fan", "out", "for", "this", "layer", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/NeuralNet/Layers/BatchNorm.php#L168-L181
RubixML/RubixML
src/NeuralNet/Layers/BatchNorm.php
BatchNorm.forward
public function forward(Matrix $input) : Matrix { if (!$this->beta or !$this->gamma) { throw new RuntimeException('Layer is not initialized.'); } $means = $variances = []; foreach ($input as $row) { [$means[], $variances[]] = Stats::meanVar($row); } $mean = ColumnVector::quick($means); $variance = ColumnVector::quick($variances); if (!$this->mean or !$this->variance) { $this->mean = $mean; $this->variance = $variance; } $stdDev = $variance->clipLower(EPSILON)->sqrt(); $stdInv = $stdDev->reciprocal(); $xHat = $stdInv->multiply($input->subtract($mean)); $this->mean = $this->mean->multiply($this->decay) ->add($mean->multiply(1. - $this->decay)); $this->variance = $this->variance->multiply($this->decay) ->add($variance->multiply(1. - $this->decay)); $this->stdInv = $stdInv; $this->xHat = $xHat; return $this->gamma->w()->multiply($xHat)->add($this->beta->w()); }
php
public function forward(Matrix $input) : Matrix { if (!$this->beta or !$this->gamma) { throw new RuntimeException('Layer is not initialized.'); } $means = $variances = []; foreach ($input as $row) { [$means[], $variances[]] = Stats::meanVar($row); } $mean = ColumnVector::quick($means); $variance = ColumnVector::quick($variances); if (!$this->mean or !$this->variance) { $this->mean = $mean; $this->variance = $variance; } $stdDev = $variance->clipLower(EPSILON)->sqrt(); $stdInv = $stdDev->reciprocal(); $xHat = $stdInv->multiply($input->subtract($mean)); $this->mean = $this->mean->multiply($this->decay) ->add($mean->multiply(1. - $this->decay)); $this->variance = $this->variance->multiply($this->decay) ->add($variance->multiply(1. - $this->decay)); $this->stdInv = $stdInv; $this->xHat = $xHat; return $this->gamma->w()->multiply($xHat)->add($this->beta->w()); }
[ "public", "function", "forward", "(", "Matrix", "$", "input", ")", ":", "Matrix", "{", "if", "(", "!", "$", "this", "->", "beta", "or", "!", "$", "this", "->", "gamma", ")", "{", "throw", "new", "RuntimeException", "(", "'Layer is not initialized.'", ")", ";", "}", "$", "means", "=", "$", "variances", "=", "[", "]", ";", "foreach", "(", "$", "input", "as", "$", "row", ")", "{", "[", "$", "means", "[", "]", ",", "$", "variances", "[", "]", "]", "=", "Stats", "::", "meanVar", "(", "$", "row", ")", ";", "}", "$", "mean", "=", "ColumnVector", "::", "quick", "(", "$", "means", ")", ";", "$", "variance", "=", "ColumnVector", "::", "quick", "(", "$", "variances", ")", ";", "if", "(", "!", "$", "this", "->", "mean", "or", "!", "$", "this", "->", "variance", ")", "{", "$", "this", "->", "mean", "=", "$", "mean", ";", "$", "this", "->", "variance", "=", "$", "variance", ";", "}", "$", "stdDev", "=", "$", "variance", "->", "clipLower", "(", "EPSILON", ")", "->", "sqrt", "(", ")", ";", "$", "stdInv", "=", "$", "stdDev", "->", "reciprocal", "(", ")", ";", "$", "xHat", "=", "$", "stdInv", "->", "multiply", "(", "$", "input", "->", "subtract", "(", "$", "mean", ")", ")", ";", "$", "this", "->", "mean", "=", "$", "this", "->", "mean", "->", "multiply", "(", "$", "this", "->", "decay", ")", "->", "add", "(", "$", "mean", "->", "multiply", "(", "1.", "-", "$", "this", "->", "decay", ")", ")", ";", "$", "this", "->", "variance", "=", "$", "this", "->", "variance", "->", "multiply", "(", "$", "this", "->", "decay", ")", "->", "add", "(", "$", "variance", "->", "multiply", "(", "1.", "-", "$", "this", "->", "decay", ")", ")", ";", "$", "this", "->", "stdInv", "=", "$", "stdInv", ";", "$", "this", "->", "xHat", "=", "$", "xHat", ";", "return", "$", "this", "->", "gamma", "->", "w", "(", ")", "->", "multiply", "(", "$", "xHat", ")", "->", "add", "(", "$", "this", "->", "beta", "->", "w", "(", ")", ")", ";", "}" ]
Compute a forward pass through the layer. @param \Rubix\Tensor\Matrix $input @throws \RuntimeException @return \Rubix\Tensor\Matrix
[ "Compute", "a", "forward", "pass", "through", "the", "layer", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/NeuralNet/Layers/BatchNorm.php#L190-L226
RubixML/RubixML
src/NeuralNet/Layers/BatchNorm.php
BatchNorm.infer
public function infer(Matrix $input) : Matrix { if (!$this->mean or !$this->variance or !$this->beta or !$this->gamma) { throw new RuntimeException('Layer has not been initilaized.'); } $xHat = $input->subtract($this->mean) ->divide($this->variance->clipLower(EPSILON)->sqrt()); return $this->gamma->w()->multiply($xHat)->add($this->beta->w()); }
php
public function infer(Matrix $input) : Matrix { if (!$this->mean or !$this->variance or !$this->beta or !$this->gamma) { throw new RuntimeException('Layer has not been initilaized.'); } $xHat = $input->subtract($this->mean) ->divide($this->variance->clipLower(EPSILON)->sqrt()); return $this->gamma->w()->multiply($xHat)->add($this->beta->w()); }
[ "public", "function", "infer", "(", "Matrix", "$", "input", ")", ":", "Matrix", "{", "if", "(", "!", "$", "this", "->", "mean", "or", "!", "$", "this", "->", "variance", "or", "!", "$", "this", "->", "beta", "or", "!", "$", "this", "->", "gamma", ")", "{", "throw", "new", "RuntimeException", "(", "'Layer has not been initilaized.'", ")", ";", "}", "$", "xHat", "=", "$", "input", "->", "subtract", "(", "$", "this", "->", "mean", ")", "->", "divide", "(", "$", "this", "->", "variance", "->", "clipLower", "(", "EPSILON", ")", "->", "sqrt", "(", ")", ")", ";", "return", "$", "this", "->", "gamma", "->", "w", "(", ")", "->", "multiply", "(", "$", "xHat", ")", "->", "add", "(", "$", "this", "->", "beta", "->", "w", "(", ")", ")", ";", "}" ]
Compute an inferential pass through the layer. @param \Rubix\Tensor\Matrix $input @throws \RuntimeException @return \Rubix\Tensor\Matrix
[ "Compute", "an", "inferential", "pass", "through", "the", "layer", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/NeuralNet/Layers/BatchNorm.php#L235-L245
RubixML/RubixML
src/NeuralNet/Layers/BatchNorm.php
BatchNorm.back
public function back(Deferred $prevGradient, Optimizer $optimizer) : Deferred { if (!$this->beta or !$this->gamma) { throw new RuntimeException('Layer has not been initilaized.'); } if (!$this->stdInv or !$this->xHat) { throw new RuntimeException('Must perform forward pass before' . ' backpropagating.'); } $dOut = $prevGradient->result(); $dBeta = $dOut->sum(); $dGamma = $dOut->multiply($this->xHat)->sum(); $gamma = $this->gamma->w(); $optimizer->step($this->beta, $dBeta); $optimizer->step($this->gamma, $dGamma); $stdInv = $this->stdInv; $xHat = $this->xHat; unset($this->stdInv, $this->xHat); return new Deferred(function () use ($dOut, $gamma, $stdInv, $xHat) { $dXHat = $dOut->multiply($gamma); $xHatSigma = $dXHat->multiply($xHat)->sum(); $dXHatSigma = $dXHat->sum(); return $dXHat->multiply($dOut->m()) ->subtract($dXHatSigma) ->subtract($xHat->multiply($xHatSigma)) ->multiply($stdInv->divide($dOut->m())); }); }
php
public function back(Deferred $prevGradient, Optimizer $optimizer) : Deferred { if (!$this->beta or !$this->gamma) { throw new RuntimeException('Layer has not been initilaized.'); } if (!$this->stdInv or !$this->xHat) { throw new RuntimeException('Must perform forward pass before' . ' backpropagating.'); } $dOut = $prevGradient->result(); $dBeta = $dOut->sum(); $dGamma = $dOut->multiply($this->xHat)->sum(); $gamma = $this->gamma->w(); $optimizer->step($this->beta, $dBeta); $optimizer->step($this->gamma, $dGamma); $stdInv = $this->stdInv; $xHat = $this->xHat; unset($this->stdInv, $this->xHat); return new Deferred(function () use ($dOut, $gamma, $stdInv, $xHat) { $dXHat = $dOut->multiply($gamma); $xHatSigma = $dXHat->multiply($xHat)->sum(); $dXHatSigma = $dXHat->sum(); return $dXHat->multiply($dOut->m()) ->subtract($dXHatSigma) ->subtract($xHat->multiply($xHatSigma)) ->multiply($stdInv->divide($dOut->m())); }); }
[ "public", "function", "back", "(", "Deferred", "$", "prevGradient", ",", "Optimizer", "$", "optimizer", ")", ":", "Deferred", "{", "if", "(", "!", "$", "this", "->", "beta", "or", "!", "$", "this", "->", "gamma", ")", "{", "throw", "new", "RuntimeException", "(", "'Layer has not been initilaized.'", ")", ";", "}", "if", "(", "!", "$", "this", "->", "stdInv", "or", "!", "$", "this", "->", "xHat", ")", "{", "throw", "new", "RuntimeException", "(", "'Must perform forward pass before'", ".", "' backpropagating.'", ")", ";", "}", "$", "dOut", "=", "$", "prevGradient", "->", "result", "(", ")", ";", "$", "dBeta", "=", "$", "dOut", "->", "sum", "(", ")", ";", "$", "dGamma", "=", "$", "dOut", "->", "multiply", "(", "$", "this", "->", "xHat", ")", "->", "sum", "(", ")", ";", "$", "gamma", "=", "$", "this", "->", "gamma", "->", "w", "(", ")", ";", "$", "optimizer", "->", "step", "(", "$", "this", "->", "beta", ",", "$", "dBeta", ")", ";", "$", "optimizer", "->", "step", "(", "$", "this", "->", "gamma", ",", "$", "dGamma", ")", ";", "$", "stdInv", "=", "$", "this", "->", "stdInv", ";", "$", "xHat", "=", "$", "this", "->", "xHat", ";", "unset", "(", "$", "this", "->", "stdInv", ",", "$", "this", "->", "xHat", ")", ";", "return", "new", "Deferred", "(", "function", "(", ")", "use", "(", "$", "dOut", ",", "$", "gamma", ",", "$", "stdInv", ",", "$", "xHat", ")", "{", "$", "dXHat", "=", "$", "dOut", "->", "multiply", "(", "$", "gamma", ")", ";", "$", "xHatSigma", "=", "$", "dXHat", "->", "multiply", "(", "$", "xHat", ")", "->", "sum", "(", ")", ";", "$", "dXHatSigma", "=", "$", "dXHat", "->", "sum", "(", ")", ";", "return", "$", "dXHat", "->", "multiply", "(", "$", "dOut", "->", "m", "(", ")", ")", "->", "subtract", "(", "$", "dXHatSigma", ")", "->", "subtract", "(", "$", "xHat", "->", "multiply", "(", "$", "xHatSigma", ")", ")", "->", "multiply", "(", "$", "stdInv", "->", "divide", "(", "$", "dOut", "->", "m", "(", ")", ")", ")", ";", "}", ")", ";", "}" ]
Calculate the errors and gradients of the layer and update the parameters. @param \Rubix\ML\NeuralNet\Deferred $prevGradient @param \Rubix\ML\NeuralNet\Optimizers\Optimizer $optimizer @throws \RuntimeException @return \Rubix\ML\NeuralNet\Deferred
[ "Calculate", "the", "errors", "and", "gradients", "of", "the", "layer", "and", "update", "the", "parameters", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/NeuralNet/Layers/BatchNorm.php#L255-L293
RubixML/RubixML
src/NeuralNet/Layers/BatchNorm.php
BatchNorm.read
public function read() : array { if (!$this->beta or !$this->gamma) { throw new RuntimeException('Layer has not been initilaized.'); } return [ 'beta' => clone $this->beta, 'gamma' => clone $this->gamma, ]; }
php
public function read() : array { if (!$this->beta or !$this->gamma) { throw new RuntimeException('Layer has not been initilaized.'); } return [ 'beta' => clone $this->beta, 'gamma' => clone $this->gamma, ]; }
[ "public", "function", "read", "(", ")", ":", "array", "{", "if", "(", "!", "$", "this", "->", "beta", "or", "!", "$", "this", "->", "gamma", ")", "{", "throw", "new", "RuntimeException", "(", "'Layer has not been initilaized.'", ")", ";", "}", "return", "[", "'beta'", "=>", "clone", "$", "this", "->", "beta", ",", "'gamma'", "=>", "clone", "$", "this", "->", "gamma", ",", "]", ";", "}" ]
Return the parameters of the layer in an associative array. @throws \RuntimeException @return array
[ "Return", "the", "parameters", "of", "the", "layer", "in", "an", "associative", "array", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/NeuralNet/Layers/BatchNorm.php#L301-L311
RubixML/RubixML
src/Classifiers/ClassificationTree.php
ClassificationTree.proba
public function proba(Dataset $dataset) : array { if ($this->bare()) { throw new RuntimeException('The learner has not' . ' been trained.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); $template = array_fill_keys($this->classes, 0.); $probabilities = []; foreach ($dataset as $sample) { $node = $this->search($sample); $probabilities[] = $node instanceof Outcome ? array_replace($template, $node->probabilities()) : null; } return $probabilities; }
php
public function proba(Dataset $dataset) : array { if ($this->bare()) { throw new RuntimeException('The learner has not' . ' been trained.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); $template = array_fill_keys($this->classes, 0.); $probabilities = []; foreach ($dataset as $sample) { $node = $this->search($sample); $probabilities[] = $node instanceof Outcome ? array_replace($template, $node->probabilities()) : null; } return $probabilities; }
[ "public", "function", "proba", "(", "Dataset", "$", "dataset", ")", ":", "array", "{", "if", "(", "$", "this", "->", "bare", "(", ")", ")", "{", "throw", "new", "RuntimeException", "(", "'The learner has not'", ".", "' been trained.'", ")", ";", "}", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "$", "template", "=", "array_fill_keys", "(", "$", "this", "->", "classes", ",", "0.", ")", ";", "$", "probabilities", "=", "[", "]", ";", "foreach", "(", "$", "dataset", "as", "$", "sample", ")", "{", "$", "node", "=", "$", "this", "->", "search", "(", "$", "sample", ")", ";", "$", "probabilities", "[", "]", "=", "$", "node", "instanceof", "Outcome", "?", "array_replace", "(", "$", "template", ",", "$", "node", "->", "probabilities", "(", ")", ")", ":", "null", ";", "}", "return", "$", "probabilities", ";", "}" ]
Estimate probabilities for each possible outcome. @param \Rubix\ML\Datasets\Dataset $dataset @throws \RuntimeException @throws \InvalidArgumentException @return array
[ "Estimate", "probabilities", "for", "each", "possible", "outcome", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Classifiers/ClassificationTree.php#L195-L217
RubixML/RubixML
src/Classifiers/ClassificationTree.php
ClassificationTree.terminate
protected function terminate(Labeled $dataset) : BinaryNode { $n = $dataset->numRows(); $labels = $dataset->labels(); $counts = array_count_values($labels); $outcome = argmax($counts); $probabilities = []; foreach ($counts as $class => $count) { $probabilities[$class] = $count / $n; } $impurity = 1. - (max($counts) / $n) ** 2; return new Outcome($outcome, $probabilities, $impurity, $n); }
php
protected function terminate(Labeled $dataset) : BinaryNode { $n = $dataset->numRows(); $labels = $dataset->labels(); $counts = array_count_values($labels); $outcome = argmax($counts); $probabilities = []; foreach ($counts as $class => $count) { $probabilities[$class] = $count / $n; } $impurity = 1. - (max($counts) / $n) ** 2; return new Outcome($outcome, $probabilities, $impurity, $n); }
[ "protected", "function", "terminate", "(", "Labeled", "$", "dataset", ")", ":", "BinaryNode", "{", "$", "n", "=", "$", "dataset", "->", "numRows", "(", ")", ";", "$", "labels", "=", "$", "dataset", "->", "labels", "(", ")", ";", "$", "counts", "=", "array_count_values", "(", "$", "labels", ")", ";", "$", "outcome", "=", "argmax", "(", "$", "counts", ")", ";", "$", "probabilities", "=", "[", "]", ";", "foreach", "(", "$", "counts", "as", "$", "class", "=>", "$", "count", ")", "{", "$", "probabilities", "[", "$", "class", "]", "=", "$", "count", "/", "$", "n", ";", "}", "$", "impurity", "=", "1.", "-", "(", "max", "(", "$", "counts", ")", "/", "$", "n", ")", "**", "2", ";", "return", "new", "Outcome", "(", "$", "outcome", ",", "$", "probabilities", ",", "$", "impurity", ",", "$", "n", ")", ";", "}" ]
Terminate the branch by selecting the class outcome with the highest probability. @param \Rubix\ML\Datasets\Labeled $dataset @return \Rubix\ML\Graph\Nodes\BinaryNode
[ "Terminate", "the", "branch", "by", "selecting", "the", "class", "outcome", "with", "the", "highest", "probability", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Classifiers/ClassificationTree.php#L264-L283
RubixML/RubixML
src/Classifiers/ClassificationTree.php
ClassificationTree.splitImpurity
protected function splitImpurity(array $groups) : float { $n = array_sum(array_map('count', $groups)); $impurity = 0.; foreach ($groups as $dataset) { $k = $dataset->numRows(); if ($k < 2) { continue 1; } $counts = array_count_values($dataset->labels()); $p = 0.; foreach ($counts as $count) { $p += 1. - ($count / $n) ** 2; } $impurity += ($k / $n) * $p; } return $impurity; }
php
protected function splitImpurity(array $groups) : float { $n = array_sum(array_map('count', $groups)); $impurity = 0.; foreach ($groups as $dataset) { $k = $dataset->numRows(); if ($k < 2) { continue 1; } $counts = array_count_values($dataset->labels()); $p = 0.; foreach ($counts as $count) { $p += 1. - ($count / $n) ** 2; } $impurity += ($k / $n) * $p; } return $impurity; }
[ "protected", "function", "splitImpurity", "(", "array", "$", "groups", ")", ":", "float", "{", "$", "n", "=", "array_sum", "(", "array_map", "(", "'count'", ",", "$", "groups", ")", ")", ";", "$", "impurity", "=", "0.", ";", "foreach", "(", "$", "groups", "as", "$", "dataset", ")", "{", "$", "k", "=", "$", "dataset", "->", "numRows", "(", ")", ";", "if", "(", "$", "k", "<", "2", ")", "{", "continue", "1", ";", "}", "$", "counts", "=", "array_count_values", "(", "$", "dataset", "->", "labels", "(", ")", ")", ";", "$", "p", "=", "0.", ";", "foreach", "(", "$", "counts", "as", "$", "count", ")", "{", "$", "p", "+=", "1.", "-", "(", "$", "count", "/", "$", "n", ")", "**", "2", ";", "}", "$", "impurity", "+=", "(", "$", "k", "/", "$", "n", ")", "*", "$", "p", ";", "}", "return", "$", "impurity", ";", "}" ]
Calculate the Gini impurity for a given split. @param array $groups @return float
[ "Calculate", "the", "Gini", "impurity", "for", "a", "given", "split", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Classifiers/ClassificationTree.php#L291-L316
RubixML/RubixML
src/Datasets/DataFrame.php
DataFrame.columnType
public function columnType(int $index) : int { if (empty($this->samples)) { throw new RuntimeException('Cannot determine data type' . ' of an empty data frame.'); } $sample = reset($this->samples); if (!isset($sample[$index])) { throw new InvalidArgumentException("Column $index does" . ' not exist.'); } return DataType::determine($sample[$index]); }
php
public function columnType(int $index) : int { if (empty($this->samples)) { throw new RuntimeException('Cannot determine data type' . ' of an empty data frame.'); } $sample = reset($this->samples); if (!isset($sample[$index])) { throw new InvalidArgumentException("Column $index does" . ' not exist.'); } return DataType::determine($sample[$index]); }
[ "public", "function", "columnType", "(", "int", "$", "index", ")", ":", "int", "{", "if", "(", "empty", "(", "$", "this", "->", "samples", ")", ")", "{", "throw", "new", "RuntimeException", "(", "'Cannot determine data type'", ".", "' of an empty data frame.'", ")", ";", "}", "$", "sample", "=", "reset", "(", "$", "this", "->", "samples", ")", ";", "if", "(", "!", "isset", "(", "$", "sample", "[", "$", "index", "]", ")", ")", "{", "throw", "new", "InvalidArgumentException", "(", "\"Column $index does\"", ".", "' not exist.'", ")", ";", "}", "return", "DataType", "::", "determine", "(", "$", "sample", "[", "$", "index", "]", ")", ";", "}" ]
Get the datatype for a feature column given a column index. @param int $index @throws \InvalidArgumentException @throws \RuntimeException @return int
[ "Get", "the", "datatype", "for", "a", "feature", "column", "given", "a", "column", "index", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Datasets/DataFrame.php#L145-L160
RubixML/RubixML
src/Datasets/DataFrame.php
DataFrame.columns
public function columns() : array { if ($this->numRows() > 1) { return array_map(null, ...$this->samples); } $n = $this->numColumns(); $columns = []; for ($i = 0; $i < $n; $i++) { $columns[] = array_column($this->samples, $i); } return $columns; }
php
public function columns() : array { if ($this->numRows() > 1) { return array_map(null, ...$this->samples); } $n = $this->numColumns(); $columns = []; for ($i = 0; $i < $n; $i++) { $columns[] = array_column($this->samples, $i); } return $columns; }
[ "public", "function", "columns", "(", ")", ":", "array", "{", "if", "(", "$", "this", "->", "numRows", "(", ")", ">", "1", ")", "{", "return", "array_map", "(", "null", ",", "...", "$", "this", "->", "samples", ")", ";", "}", "$", "n", "=", "$", "this", "->", "numColumns", "(", ")", ";", "$", "columns", "=", "[", "]", ";", "for", "(", "$", "i", "=", "0", ";", "$", "i", "<", "$", "n", ";", "$", "i", "++", ")", "{", "$", "columns", "[", "]", "=", "array_column", "(", "$", "this", "->", "samples", ",", "$", "i", ")", ";", "}", "return", "$", "columns", ";", "}" ]
Rotate the dataframe and return it in an array. i.e. rows become columns and columns become rows. @return array
[ "Rotate", "the", "dataframe", "and", "return", "it", "in", "an", "array", ".", "i", ".", "e", ".", "rows", "become", "columns", "and", "columns", "become", "rows", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Datasets/DataFrame.php#L189-L204
RubixML/RubixML
src/Datasets/DataFrame.php
DataFrame.columnsByType
public function columnsByType(int $type) : array { $n = $this->numColumns(); $columns = []; for ($i = 0; $i < $n; $i++) { if ($this->columnType($i) === $type) { $columns[$i] = $this->column($i); } } return $columns; }
php
public function columnsByType(int $type) : array { $n = $this->numColumns(); $columns = []; for ($i = 0; $i < $n; $i++) { if ($this->columnType($i) === $type) { $columns[$i] = $this->column($i); } } return $columns; }
[ "public", "function", "columnsByType", "(", "int", "$", "type", ")", ":", "array", "{", "$", "n", "=", "$", "this", "->", "numColumns", "(", ")", ";", "$", "columns", "=", "[", "]", ";", "for", "(", "$", "i", "=", "0", ";", "$", "i", "<", "$", "n", ";", "$", "i", "++", ")", "{", "if", "(", "$", "this", "->", "columnType", "(", "$", "i", ")", "===", "$", "type", ")", "{", "$", "columns", "[", "$", "i", "]", "=", "$", "this", "->", "column", "(", "$", "i", ")", ";", "}", "}", "return", "$", "columns", ";", "}" ]
Return the columns that match a given data type. @param int $type @return array
[ "Return", "the", "columns", "that", "match", "a", "given", "data", "type", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Datasets/DataFrame.php#L212-L225
RubixML/RubixML
src/Transformers/StopWordFilter.php
StopWordFilter.transform
public function transform(array &$samples) : void { foreach ($samples as &$sample) { foreach ($sample as &$feature) { if (is_string($feature)) { $feature = preg_replace($this->stopWords, '', $feature); } } } }
php
public function transform(array &$samples) : void { foreach ($samples as &$sample) { foreach ($sample as &$feature) { if (is_string($feature)) { $feature = preg_replace($this->stopWords, '', $feature); } } } }
[ "public", "function", "transform", "(", "array", "&", "$", "samples", ")", ":", "void", "{", "foreach", "(", "$", "samples", "as", "&", "$", "sample", ")", "{", "foreach", "(", "$", "sample", "as", "&", "$", "feature", ")", "{", "if", "(", "is_string", "(", "$", "feature", ")", ")", "{", "$", "feature", "=", "preg_replace", "(", "$", "this", "->", "stopWords", ",", "''", ",", "$", "feature", ")", ";", "}", "}", "}", "}" ]
Transform the dataset in place. @param array $samples
[ "Transform", "the", "dataset", "in", "place", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/StopWordFilter.php#L50-L59
RubixML/RubixML
src/Transformers/PolynomialExpander.php
PolynomialExpander.transform
public function transform(array &$samples) : void { foreach ($samples as &$sample) { $vector = []; foreach ($sample as $feature) { for ($i = 1; $i <= $this->degree; $i++) { $vector[] = $feature ** $i; } } $sample = $vector; } }
php
public function transform(array &$samples) : void { foreach ($samples as &$sample) { $vector = []; foreach ($sample as $feature) { for ($i = 1; $i <= $this->degree; $i++) { $vector[] = $feature ** $i; } } $sample = $vector; } }
[ "public", "function", "transform", "(", "array", "&", "$", "samples", ")", ":", "void", "{", "foreach", "(", "$", "samples", "as", "&", "$", "sample", ")", "{", "$", "vector", "=", "[", "]", ";", "foreach", "(", "$", "sample", "as", "$", "feature", ")", "{", "for", "(", "$", "i", "=", "1", ";", "$", "i", "<=", "$", "this", "->", "degree", ";", "$", "i", "++", ")", "{", "$", "vector", "[", "]", "=", "$", "feature", "**", "$", "i", ";", "}", "}", "$", "sample", "=", "$", "vector", ";", "}", "}" ]
Transform the dataset in place. @param array $samples
[ "Transform", "the", "dataset", "in", "place", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/PolynomialExpander.php#L49-L62
RubixML/RubixML
src/NeuralNet/Layers/Continuous.php
Continuous.parameters
public function parameters() : Generator { if (!$this->weights or !$this->biases) { throw new RuntimeException('Layer is not initialized'); } yield $this->weights; yield $this->biases; }
php
public function parameters() : Generator { if (!$this->weights or !$this->biases) { throw new RuntimeException('Layer is not initialized'); } yield $this->weights; yield $this->biases; }
[ "public", "function", "parameters", "(", ")", ":", "Generator", "{", "if", "(", "!", "$", "this", "->", "weights", "or", "!", "$", "this", "->", "biases", ")", "{", "throw", "new", "RuntimeException", "(", "'Layer is not initialized'", ")", ";", "}", "yield", "$", "this", "->", "weights", ";", "yield", "$", "this", "->", "biases", ";", "}" ]
Return the parameters of the layer. @throws \RuntimeException @return \Generator
[ "Return", "the", "parameters", "of", "the", "layer", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/NeuralNet/Layers/Continuous.php#L128-L136
RubixML/RubixML
src/NeuralNet/Layers/Continuous.php
Continuous.forward
public function forward(Matrix $input) : Matrix { if (!$this->weights or !$this->biases) { throw new RuntimeException('Layer is not initialized'); } $this->input = $input; $this->z = $this->weights->w()->matmul($input) ->add($this->biases->w()); return $this->z; }
php
public function forward(Matrix $input) : Matrix { if (!$this->weights or !$this->biases) { throw new RuntimeException('Layer is not initialized'); } $this->input = $input; $this->z = $this->weights->w()->matmul($input) ->add($this->biases->w()); return $this->z; }
[ "public", "function", "forward", "(", "Matrix", "$", "input", ")", ":", "Matrix", "{", "if", "(", "!", "$", "this", "->", "weights", "or", "!", "$", "this", "->", "biases", ")", "{", "throw", "new", "RuntimeException", "(", "'Layer is not initialized'", ")", ";", "}", "$", "this", "->", "input", "=", "$", "input", ";", "$", "this", "->", "z", "=", "$", "this", "->", "weights", "->", "w", "(", ")", "->", "matmul", "(", "$", "input", ")", "->", "add", "(", "$", "this", "->", "biases", "->", "w", "(", ")", ")", ";", "return", "$", "this", "->", "z", ";", "}" ]
Compute a forward pass through the layer. @param \Rubix\Tensor\Matrix $input @throws \RuntimeException @return \Rubix\Tensor\Matrix
[ "Compute", "a", "forward", "pass", "through", "the", "layer", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/NeuralNet/Layers/Continuous.php#L165-L177
RubixML/RubixML
src/NeuralNet/Layers/Continuous.php
Continuous.back
public function back(array $labels, Optimizer $optimizer) : array { if (!$this->weights or !$this->biases) { throw new RuntimeException('Layer is not initialized'); } if (!$this->input or !$this->z) { throw new RuntimeException('Must perform forward pass before' . ' backpropagating.'); } $expected = Vector::quick($labels); $penalties = $this->weights->w()->sum() ->multiply($this->alpha); $dL = $this->costFn ->differentiate($expected, $this->z) ->add($penalties) ->divide($this->z->n()); $dW = $dL->matmul($this->input->transpose()); $dB = $dL->sum(); $w = $this->weights->w(); $optimizer->step($this->weights, $dW); $optimizer->step($this->biases, $dB); $loss = $this->costFn->compute($expected, $this->z); $gradient = new Deferred(function () use ($w, $dL) { return $w->transpose()->matmul($dL); }); unset($this->input, $this->z); return [$gradient, $loss]; }
php
public function back(array $labels, Optimizer $optimizer) : array { if (!$this->weights or !$this->biases) { throw new RuntimeException('Layer is not initialized'); } if (!$this->input or !$this->z) { throw new RuntimeException('Must perform forward pass before' . ' backpropagating.'); } $expected = Vector::quick($labels); $penalties = $this->weights->w()->sum() ->multiply($this->alpha); $dL = $this->costFn ->differentiate($expected, $this->z) ->add($penalties) ->divide($this->z->n()); $dW = $dL->matmul($this->input->transpose()); $dB = $dL->sum(); $w = $this->weights->w(); $optimizer->step($this->weights, $dW); $optimizer->step($this->biases, $dB); $loss = $this->costFn->compute($expected, $this->z); $gradient = new Deferred(function () use ($w, $dL) { return $w->transpose()->matmul($dL); }); unset($this->input, $this->z); return [$gradient, $loss]; }
[ "public", "function", "back", "(", "array", "$", "labels", ",", "Optimizer", "$", "optimizer", ")", ":", "array", "{", "if", "(", "!", "$", "this", "->", "weights", "or", "!", "$", "this", "->", "biases", ")", "{", "throw", "new", "RuntimeException", "(", "'Layer is not initialized'", ")", ";", "}", "if", "(", "!", "$", "this", "->", "input", "or", "!", "$", "this", "->", "z", ")", "{", "throw", "new", "RuntimeException", "(", "'Must perform forward pass before'", ".", "' backpropagating.'", ")", ";", "}", "$", "expected", "=", "Vector", "::", "quick", "(", "$", "labels", ")", ";", "$", "penalties", "=", "$", "this", "->", "weights", "->", "w", "(", ")", "->", "sum", "(", ")", "->", "multiply", "(", "$", "this", "->", "alpha", ")", ";", "$", "dL", "=", "$", "this", "->", "costFn", "->", "differentiate", "(", "$", "expected", ",", "$", "this", "->", "z", ")", "->", "add", "(", "$", "penalties", ")", "->", "divide", "(", "$", "this", "->", "z", "->", "n", "(", ")", ")", ";", "$", "dW", "=", "$", "dL", "->", "matmul", "(", "$", "this", "->", "input", "->", "transpose", "(", ")", ")", ";", "$", "dB", "=", "$", "dL", "->", "sum", "(", ")", ";", "$", "w", "=", "$", "this", "->", "weights", "->", "w", "(", ")", ";", "$", "optimizer", "->", "step", "(", "$", "this", "->", "weights", ",", "$", "dW", ")", ";", "$", "optimizer", "->", "step", "(", "$", "this", "->", "biases", ",", "$", "dB", ")", ";", "$", "loss", "=", "$", "this", "->", "costFn", "->", "compute", "(", "$", "expected", ",", "$", "this", "->", "z", ")", ";", "$", "gradient", "=", "new", "Deferred", "(", "function", "(", ")", "use", "(", "$", "w", ",", "$", "dL", ")", "{", "return", "$", "w", "->", "transpose", "(", ")", "->", "matmul", "(", "$", "dL", ")", ";", "}", ")", ";", "unset", "(", "$", "this", "->", "input", ",", "$", "this", "->", "z", ")", ";", "return", "[", "$", "gradient", ",", "$", "loss", "]", ";", "}" ]
Calculate the gradients for each output neuron and update. @param array $labels @param \Rubix\ML\NeuralNet\Optimizers\Optimizer $optimizer @throws \RuntimeException @return array
[ "Calculate", "the", "gradients", "for", "each", "output", "neuron", "and", "update", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/NeuralNet/Layers/Continuous.php#L204-L242
RubixML/RubixML
src/NeuralNet/Optimizers/Momentum.php
Momentum.step
public function step(Parameter $param, Tensor $gradient) : void { $velocity = $this->cache[$param->id()]; $velocity = $gradient->multiply($this->rate) ->add($velocity->multiply($this->decay)); $this->cache[$param->id()] = $velocity; $param->update($velocity); }
php
public function step(Parameter $param, Tensor $gradient) : void { $velocity = $this->cache[$param->id()]; $velocity = $gradient->multiply($this->rate) ->add($velocity->multiply($this->decay)); $this->cache[$param->id()] = $velocity; $param->update($velocity); }
[ "public", "function", "step", "(", "Parameter", "$", "param", ",", "Tensor", "$", "gradient", ")", ":", "void", "{", "$", "velocity", "=", "$", "this", "->", "cache", "[", "$", "param", "->", "id", "(", ")", "]", ";", "$", "velocity", "=", "$", "gradient", "->", "multiply", "(", "$", "this", "->", "rate", ")", "->", "add", "(", "$", "velocity", "->", "multiply", "(", "$", "this", "->", "decay", ")", ")", ";", "$", "this", "->", "cache", "[", "$", "param", "->", "id", "(", ")", "]", "=", "$", "velocity", ";", "$", "param", "->", "update", "(", "$", "velocity", ")", ";", "}" ]
Take a step of gradient descent for a given parameter. @param \Rubix\ML\NeuralNet\Parameters\Parameter $param @param \Rubix\Tensor\Tensor $gradient
[ "Take", "a", "step", "of", "gradient", "descent", "for", "a", "given", "parameter", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/NeuralNet/Optimizers/Momentum.php#L86-L96
RubixML/RubixML
src/Transformers/VarianceThresholdFilter.php
VarianceThresholdFilter.fit
public function fit(Dataset $dataset) : void { $columns = $dataset->columnsByType(DataType::CONTINUOUS); $this->selected = []; foreach ($columns as $column => $values) { if (Stats::variance($values) <= $this->threshold) { continue 1; } $this->selected[$column] = true; } }
php
public function fit(Dataset $dataset) : void { $columns = $dataset->columnsByType(DataType::CONTINUOUS); $this->selected = []; foreach ($columns as $column => $values) { if (Stats::variance($values) <= $this->threshold) { continue 1; } $this->selected[$column] = true; } }
[ "public", "function", "fit", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "$", "columns", "=", "$", "dataset", "->", "columnsByType", "(", "DataType", "::", "CONTINUOUS", ")", ";", "$", "this", "->", "selected", "=", "[", "]", ";", "foreach", "(", "$", "columns", "as", "$", "column", "=>", "$", "values", ")", "{", "if", "(", "Stats", "::", "variance", "(", "$", "values", ")", "<=", "$", "this", "->", "threshold", ")", "{", "continue", "1", ";", "}", "$", "this", "->", "selected", "[", "$", "column", "]", "=", "true", ";", "}", "}" ]
Fit the transformer to the dataset. @param \Rubix\ML\Datasets\Dataset $dataset
[ "Fit", "the", "transformer", "to", "the", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/VarianceThresholdFilter.php#L78-L91
RubixML/RubixML
src/Transformers/VarianceThresholdFilter.php
VarianceThresholdFilter.transform
public function transform(array &$samples) : void { if ($this->selected === null) { throw new RuntimeException('Transformer has not been fitted.'); } foreach ($samples as &$sample) { $sample = array_intersect_key($sample, $this->selected); } }
php
public function transform(array &$samples) : void { if ($this->selected === null) { throw new RuntimeException('Transformer has not been fitted.'); } foreach ($samples as &$sample) { $sample = array_intersect_key($sample, $this->selected); } }
[ "public", "function", "transform", "(", "array", "&", "$", "samples", ")", ":", "void", "{", "if", "(", "$", "this", "->", "selected", "===", "null", ")", "{", "throw", "new", "RuntimeException", "(", "'Transformer has not been fitted.'", ")", ";", "}", "foreach", "(", "$", "samples", "as", "&", "$", "sample", ")", "{", "$", "sample", "=", "array_intersect_key", "(", "$", "sample", ",", "$", "this", "->", "selected", ")", ";", "}", "}" ]
Transform the dataset in place. @param array $samples @throws \RuntimeException
[ "Transform", "the", "dataset", "in", "place", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/VarianceThresholdFilter.php#L99-L108
RubixML/RubixML
src/Classifiers/LogisticRegression.php
LogisticRegression.partial
public function partial(Dataset $dataset) : void { if (!$this->network) { $this->train($dataset); return; } if (!$dataset instanceof Labeled) { throw new InvalidArgumentException('This estimator requires a' . ' labeled training set.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); if ($this->logger) { $this->logger->info('Learner init ' . Params::stringify([ 'batch_size' => $this->batchSize, 'optimizer' => $this->optimizer, 'alpha' => $this->alpha, 'epochs' => $this->epochs, 'min_change' => $this->minChange, 'cost_fn' => $this->costFn, ])); } $n = $dataset->numRows(); $randomize = $n > $this->batchSize ? true : false; $previous = INF; for ($epoch = 1; $epoch <= $this->epochs; $epoch++) { if ($randomize) { $dataset->randomize(); } $batches = $dataset->batch($this->batchSize); $loss = 0.; foreach ($batches as $batch) { $loss += $this->network->roundtrip($batch); } $loss /= $n; $this->steps[] = $loss; if ($this->logger) { $this->logger->info("Epoch $epoch complete, loss=$loss"); } if (is_nan($loss)) { break 1; } if (abs($previous - $loss) < $this->minChange) { break 1; } $previous = $loss; } if ($this->logger) { $this->logger->info('Training complete'); } }
php
public function partial(Dataset $dataset) : void { if (!$this->network) { $this->train($dataset); return; } if (!$dataset instanceof Labeled) { throw new InvalidArgumentException('This estimator requires a' . ' labeled training set.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); if ($this->logger) { $this->logger->info('Learner init ' . Params::stringify([ 'batch_size' => $this->batchSize, 'optimizer' => $this->optimizer, 'alpha' => $this->alpha, 'epochs' => $this->epochs, 'min_change' => $this->minChange, 'cost_fn' => $this->costFn, ])); } $n = $dataset->numRows(); $randomize = $n > $this->batchSize ? true : false; $previous = INF; for ($epoch = 1; $epoch <= $this->epochs; $epoch++) { if ($randomize) { $dataset->randomize(); } $batches = $dataset->batch($this->batchSize); $loss = 0.; foreach ($batches as $batch) { $loss += $this->network->roundtrip($batch); } $loss /= $n; $this->steps[] = $loss; if ($this->logger) { $this->logger->info("Epoch $epoch complete, loss=$loss"); } if (is_nan($loss)) { break 1; } if (abs($previous - $loss) < $this->minChange) { break 1; } $previous = $loss; } if ($this->logger) { $this->logger->info('Training complete'); } }
[ "public", "function", "partial", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "if", "(", "!", "$", "this", "->", "network", ")", "{", "$", "this", "->", "train", "(", "$", "dataset", ")", ";", "return", ";", "}", "if", "(", "!", "$", "dataset", "instanceof", "Labeled", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'This estimator requires a'", ".", "' labeled training set.'", ")", ";", "}", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "'Learner init '", ".", "Params", "::", "stringify", "(", "[", "'batch_size'", "=>", "$", "this", "->", "batchSize", ",", "'optimizer'", "=>", "$", "this", "->", "optimizer", ",", "'alpha'", "=>", "$", "this", "->", "alpha", ",", "'epochs'", "=>", "$", "this", "->", "epochs", ",", "'min_change'", "=>", "$", "this", "->", "minChange", ",", "'cost_fn'", "=>", "$", "this", "->", "costFn", ",", "]", ")", ")", ";", "}", "$", "n", "=", "$", "dataset", "->", "numRows", "(", ")", ";", "$", "randomize", "=", "$", "n", ">", "$", "this", "->", "batchSize", "?", "true", ":", "false", ";", "$", "previous", "=", "INF", ";", "for", "(", "$", "epoch", "=", "1", ";", "$", "epoch", "<=", "$", "this", "->", "epochs", ";", "$", "epoch", "++", ")", "{", "if", "(", "$", "randomize", ")", "{", "$", "dataset", "->", "randomize", "(", ")", ";", "}", "$", "batches", "=", "$", "dataset", "->", "batch", "(", "$", "this", "->", "batchSize", ")", ";", "$", "loss", "=", "0.", ";", "foreach", "(", "$", "batches", "as", "$", "batch", ")", "{", "$", "loss", "+=", "$", "this", "->", "network", "->", "roundtrip", "(", "$", "batch", ")", ";", "}", "$", "loss", "/=", "$", "n", ";", "$", "this", "->", "steps", "[", "]", "=", "$", "loss", ";", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "\"Epoch $epoch complete, loss=$loss\"", ")", ";", "}", "if", "(", "is_nan", "(", "$", "loss", ")", ")", "{", "break", "1", ";", "}", "if", "(", "abs", "(", "$", "previous", "-", "$", "loss", ")", "<", "$", "this", "->", "minChange", ")", "{", "break", "1", ";", "}", "$", "previous", "=", "$", "loss", ";", "}", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "'Training complete'", ")", ";", "}", "}" ]
Perform mini-batch gradient descent with given optimizer over the training set and update the input weights accordingly. @param \Rubix\ML\Datasets\Dataset $dataset @throws \InvalidArgumentException
[ "Perform", "mini", "-", "batch", "gradient", "descent", "with", "given", "optimizer", "over", "the", "training", "set", "and", "update", "the", "input", "weights", "accordingly", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Classifiers/LogisticRegression.php#L237-L304
RubixML/RubixML
src/Classifiers/LogisticRegression.php
LogisticRegression.proba
public function proba(Dataset $dataset) : array { if (!$this->network or !$this->classes) { throw new RuntimeException('The learner has not' . ' been trained.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); $xT = Matrix::quick($dataset->samples())->transpose(); $y = $this->network->infer($xT)->row(0); [$classA, $classB] = $this->classes; $probabilities = []; foreach ($y as $activation) { $probabilities[] = [ $classA => 1. - $activation, $classB => $activation, ]; } return $probabilities; }
php
public function proba(Dataset $dataset) : array { if (!$this->network or !$this->classes) { throw new RuntimeException('The learner has not' . ' been trained.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); $xT = Matrix::quick($dataset->samples())->transpose(); $y = $this->network->infer($xT)->row(0); [$classA, $classB] = $this->classes; $probabilities = []; foreach ($y as $activation) { $probabilities[] = [ $classA => 1. - $activation, $classB => $activation, ]; } return $probabilities; }
[ "public", "function", "proba", "(", "Dataset", "$", "dataset", ")", ":", "array", "{", "if", "(", "!", "$", "this", "->", "network", "or", "!", "$", "this", "->", "classes", ")", "{", "throw", "new", "RuntimeException", "(", "'The learner has not'", ".", "' been trained.'", ")", ";", "}", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "$", "xT", "=", "Matrix", "::", "quick", "(", "$", "dataset", "->", "samples", "(", ")", ")", "->", "transpose", "(", ")", ";", "$", "y", "=", "$", "this", "->", "network", "->", "infer", "(", "$", "xT", ")", "->", "row", "(", "0", ")", ";", "[", "$", "classA", ",", "$", "classB", "]", "=", "$", "this", "->", "classes", ";", "$", "probabilities", "=", "[", "]", ";", "foreach", "(", "$", "y", "as", "$", "activation", ")", "{", "$", "probabilities", "[", "]", "=", "[", "$", "classA", "=>", "1.", "-", "$", "activation", ",", "$", "classB", "=>", "$", "activation", ",", "]", ";", "}", "return", "$", "probabilities", ";", "}" ]
Estimate probabilities for each possible outcome. @param \Rubix\ML\Datasets\Dataset $dataset @throws \InvalidArgumentException @throws \RuntimeException @return array
[ "Estimate", "probabilities", "for", "each", "possible", "outcome", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Classifiers/LogisticRegression.php#L325-L350
RubixML/RubixML
src/Kernels/Distance/Diagonal.php
Diagonal.compute
public function compute(array $a, array $b) : float { $deltas = []; foreach ($a as $i => $value) { $deltas[] = abs($value - $b[$i]); } return max($deltas); }
php
public function compute(array $a, array $b) : float { $deltas = []; foreach ($a as $i => $value) { $deltas[] = abs($value - $b[$i]); } return max($deltas); }
[ "public", "function", "compute", "(", "array", "$", "a", ",", "array", "$", "b", ")", ":", "float", "{", "$", "deltas", "=", "[", "]", ";", "foreach", "(", "$", "a", "as", "$", "i", "=>", "$", "value", ")", "{", "$", "deltas", "[", "]", "=", "abs", "(", "$", "value", "-", "$", "b", "[", "$", "i", "]", ")", ";", "}", "return", "max", "(", "$", "deltas", ")", ";", "}" ]
Compute the distance between two vectors. @param array $a @param array $b @return float
[ "Compute", "the", "distance", "between", "two", "vectors", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Kernels/Distance/Diagonal.php#L25-L34
RubixML/RubixML
src/Transformers/NumericStringConverter.php
NumericStringConverter.transform
public function transform(array &$samples) : void { foreach ($samples as &$sample) { foreach ($sample as &$feature) { if (is_string($feature) and is_numeric($feature)) { $feature = (int) $feature == $feature ? (int) $feature : (float) $feature; } } } }
php
public function transform(array &$samples) : void { foreach ($samples as &$sample) { foreach ($sample as &$feature) { if (is_string($feature) and is_numeric($feature)) { $feature = (int) $feature == $feature ? (int) $feature : (float) $feature; } } } }
[ "public", "function", "transform", "(", "array", "&", "$", "samples", ")", ":", "void", "{", "foreach", "(", "$", "samples", "as", "&", "$", "sample", ")", "{", "foreach", "(", "$", "sample", "as", "&", "$", "feature", ")", "{", "if", "(", "is_string", "(", "$", "feature", ")", "and", "is_numeric", "(", "$", "feature", ")", ")", "{", "$", "feature", "=", "(", "int", ")", "$", "feature", "==", "$", "feature", "?", "(", "int", ")", "$", "feature", ":", "(", "float", ")", "$", "feature", ";", "}", "}", "}", "}" ]
Transform the dataset in place. @param array $samples
[ "Transform", "the", "dataset", "in", "place", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/NumericStringConverter.php#L23-L34
RubixML/RubixML
src/Classifiers/MultiLayerPerceptron.php
MultiLayerPerceptron.proba
public function proba(Dataset $dataset) : array { if (!$this->network or !$this->classes) { throw new RuntimeException('The learner has not' . ' been trained.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); $xT = Matrix::quick($dataset->samples())->transpose(); $yT = $this->network->infer($xT)->transpose(); $probabilities = []; foreach ($yT as $activations) { $probabilities[] = array_combine($this->classes, $activations); } return $probabilities; }
php
public function proba(Dataset $dataset) : array { if (!$this->network or !$this->classes) { throw new RuntimeException('The learner has not' . ' been trained.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); $xT = Matrix::quick($dataset->samples())->transpose(); $yT = $this->network->infer($xT)->transpose(); $probabilities = []; foreach ($yT as $activations) { $probabilities[] = array_combine($this->classes, $activations); } return $probabilities; }
[ "public", "function", "proba", "(", "Dataset", "$", "dataset", ")", ":", "array", "{", "if", "(", "!", "$", "this", "->", "network", "or", "!", "$", "this", "->", "classes", ")", "{", "throw", "new", "RuntimeException", "(", "'The learner has not'", ".", "' been trained.'", ")", ";", "}", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "$", "xT", "=", "Matrix", "::", "quick", "(", "$", "dataset", "->", "samples", "(", ")", ")", "->", "transpose", "(", ")", ";", "$", "yT", "=", "$", "this", "->", "network", "->", "infer", "(", "$", "xT", ")", "->", "transpose", "(", ")", ";", "$", "probabilities", "=", "[", "]", ";", "foreach", "(", "$", "yT", "as", "$", "activations", ")", "{", "$", "probabilities", "[", "]", "=", "array_combine", "(", "$", "this", "->", "classes", ",", "$", "activations", ")", ";", "}", "return", "$", "probabilities", ";", "}" ]
Estimate probabilities for each possible outcome. @param \Rubix\ML\Datasets\Dataset $dataset @throws \InvalidArgumentException @throws \RuntimeException @return array
[ "Estimate", "probabilities", "for", "each", "possible", "outcome", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Classifiers/MultiLayerPerceptron.php#L462-L482
RubixML/RubixML
src/Transformers/L1Normalizer.php
L1Normalizer.transform
public function transform(array &$samples) : void { foreach ($samples as &$sample) { $norm = array_sum(array_map('abs', $sample)) ?: EPSILON; foreach ($sample as &$feature) { $feature /= $norm; } } }
php
public function transform(array &$samples) : void { foreach ($samples as &$sample) { $norm = array_sum(array_map('abs', $sample)) ?: EPSILON; foreach ($sample as &$feature) { $feature /= $norm; } } }
[ "public", "function", "transform", "(", "array", "&", "$", "samples", ")", ":", "void", "{", "foreach", "(", "$", "samples", "as", "&", "$", "sample", ")", "{", "$", "norm", "=", "array_sum", "(", "array_map", "(", "'abs'", ",", "$", "sample", ")", ")", "?", ":", "EPSILON", ";", "foreach", "(", "$", "sample", "as", "&", "$", "feature", ")", "{", "$", "feature", "/=", "$", "norm", ";", "}", "}", "}" ]
Transform the dataset in place. @param array $samples
[ "Transform", "the", "dataset", "in", "place", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/L1Normalizer.php#L24-L33
RubixML/RubixML
src/Regressors/GradientBoost.php
GradientBoost.compatibility
public function compatibility() : array { $compatibility = array_intersect($this->base->compatibility(), $this->booster->compatibility()); return array_values($compatibility); }
php
public function compatibility() : array { $compatibility = array_intersect($this->base->compatibility(), $this->booster->compatibility()); return array_values($compatibility); }
[ "public", "function", "compatibility", "(", ")", ":", "array", "{", "$", "compatibility", "=", "array_intersect", "(", "$", "this", "->", "base", "->", "compatibility", "(", ")", ",", "$", "this", "->", "booster", "->", "compatibility", "(", ")", ")", ";", "return", "array_values", "(", "$", "compatibility", ")", ";", "}" ]
Return the data types that this estimator is compatible with. @return int[]
[ "Return", "the", "data", "types", "that", "this", "estimator", "is", "compatible", "with", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Regressors/GradientBoost.php#L192-L197
RubixML/RubixML
src/Regressors/GradientBoost.php
GradientBoost.train
public function train(Dataset $dataset) : void { if (!$dataset instanceof Labeled) { throw new InvalidArgumentException('This estimator requires a' . ' labeled training set.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); if ($this->logger) { $this->logger->info('Learner init ' . Params::stringify([ 'booster' => $this->booster, 'rate' => $this->rate, 'estimators' => $this->estimators, 'ratio' => $this->ratio, 'min_change' => $this->minChange, 'tolerance' => $this->tolerance, 'base' => $this->base, ])); } $n = $dataset->numRows(); $p = (int) round($this->ratio * $n); if ($this->logger) { $this->logger->info('Training ' . Params::shortName($this->base) . ' base estimator'); } $this->base->train($dataset); $predictions = $this->base->predict($dataset); $yHat = []; foreach ($predictions as $i => $prediction) { $yHat[] = $dataset->label($i) - $prediction; } $residual = Labeled::quick($dataset->samples(), $yHat); if ($this->logger) { $this->logger->info("Boosting with $this->estimators " . Params::shortName($this->booster) . ($this->estimators > 1 ? 's' : '')); } $this->ensemble = $this->steps = []; $previous = INF; for ($epoch = 1; $epoch <= $this->estimators; $epoch++) { $booster = clone $this->booster; $subset = $residual->randomize()->head($p); $booster->train($subset); $predictions = $booster->predict($residual); $labels = $residual->labels(); $loss = 0.; foreach ($predictions as $i => $prediction) { $label = $labels[$i]; $loss += ($label - $prediction) ** 2; $yHat[$i] = $label - ($this->rate * $prediction); } $loss /= $n; $this->ensemble[] = $booster; $this->steps[] = $loss; if ($this->logger) { $this->logger->info("Epoch $epoch complete, loss=$loss"); } if (is_nan($loss)) { break 1; } if (abs($previous - $loss) < $this->minChange) { break 1; } if ($loss < $this->tolerance) { break 1; } $residual = Labeled::quick($residual->samples(), $yHat); $previous = $loss; } if ($this->logger) { $this->logger->info('Training complete'); } }
php
public function train(Dataset $dataset) : void { if (!$dataset instanceof Labeled) { throw new InvalidArgumentException('This estimator requires a' . ' labeled training set.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); if ($this->logger) { $this->logger->info('Learner init ' . Params::stringify([ 'booster' => $this->booster, 'rate' => $this->rate, 'estimators' => $this->estimators, 'ratio' => $this->ratio, 'min_change' => $this->minChange, 'tolerance' => $this->tolerance, 'base' => $this->base, ])); } $n = $dataset->numRows(); $p = (int) round($this->ratio * $n); if ($this->logger) { $this->logger->info('Training ' . Params::shortName($this->base) . ' base estimator'); } $this->base->train($dataset); $predictions = $this->base->predict($dataset); $yHat = []; foreach ($predictions as $i => $prediction) { $yHat[] = $dataset->label($i) - $prediction; } $residual = Labeled::quick($dataset->samples(), $yHat); if ($this->logger) { $this->logger->info("Boosting with $this->estimators " . Params::shortName($this->booster) . ($this->estimators > 1 ? 's' : '')); } $this->ensemble = $this->steps = []; $previous = INF; for ($epoch = 1; $epoch <= $this->estimators; $epoch++) { $booster = clone $this->booster; $subset = $residual->randomize()->head($p); $booster->train($subset); $predictions = $booster->predict($residual); $labels = $residual->labels(); $loss = 0.; foreach ($predictions as $i => $prediction) { $label = $labels[$i]; $loss += ($label - $prediction) ** 2; $yHat[$i] = $label - ($this->rate * $prediction); } $loss /= $n; $this->ensemble[] = $booster; $this->steps[] = $loss; if ($this->logger) { $this->logger->info("Epoch $epoch complete, loss=$loss"); } if (is_nan($loss)) { break 1; } if (abs($previous - $loss) < $this->minChange) { break 1; } if ($loss < $this->tolerance) { break 1; } $residual = Labeled::quick($residual->samples(), $yHat); $previous = $loss; } if ($this->logger) { $this->logger->info('Training complete'); } }
[ "public", "function", "train", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "if", "(", "!", "$", "dataset", "instanceof", "Labeled", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'This estimator requires a'", ".", "' labeled training set.'", ")", ";", "}", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "'Learner init '", ".", "Params", "::", "stringify", "(", "[", "'booster'", "=>", "$", "this", "->", "booster", ",", "'rate'", "=>", "$", "this", "->", "rate", ",", "'estimators'", "=>", "$", "this", "->", "estimators", ",", "'ratio'", "=>", "$", "this", "->", "ratio", ",", "'min_change'", "=>", "$", "this", "->", "minChange", ",", "'tolerance'", "=>", "$", "this", "->", "tolerance", ",", "'base'", "=>", "$", "this", "->", "base", ",", "]", ")", ")", ";", "}", "$", "n", "=", "$", "dataset", "->", "numRows", "(", ")", ";", "$", "p", "=", "(", "int", ")", "round", "(", "$", "this", "->", "ratio", "*", "$", "n", ")", ";", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "'Training '", ".", "Params", "::", "shortName", "(", "$", "this", "->", "base", ")", ".", "' base estimator'", ")", ";", "}", "$", "this", "->", "base", "->", "train", "(", "$", "dataset", ")", ";", "$", "predictions", "=", "$", "this", "->", "base", "->", "predict", "(", "$", "dataset", ")", ";", "$", "yHat", "=", "[", "]", ";", "foreach", "(", "$", "predictions", "as", "$", "i", "=>", "$", "prediction", ")", "{", "$", "yHat", "[", "]", "=", "$", "dataset", "->", "label", "(", "$", "i", ")", "-", "$", "prediction", ";", "}", "$", "residual", "=", "Labeled", "::", "quick", "(", "$", "dataset", "->", "samples", "(", ")", ",", "$", "yHat", ")", ";", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "\"Boosting with $this->estimators \"", ".", "Params", "::", "shortName", "(", "$", "this", "->", "booster", ")", ".", "(", "$", "this", "->", "estimators", ">", "1", "?", "'s'", ":", "''", ")", ")", ";", "}", "$", "this", "->", "ensemble", "=", "$", "this", "->", "steps", "=", "[", "]", ";", "$", "previous", "=", "INF", ";", "for", "(", "$", "epoch", "=", "1", ";", "$", "epoch", "<=", "$", "this", "->", "estimators", ";", "$", "epoch", "++", ")", "{", "$", "booster", "=", "clone", "$", "this", "->", "booster", ";", "$", "subset", "=", "$", "residual", "->", "randomize", "(", ")", "->", "head", "(", "$", "p", ")", ";", "$", "booster", "->", "train", "(", "$", "subset", ")", ";", "$", "predictions", "=", "$", "booster", "->", "predict", "(", "$", "residual", ")", ";", "$", "labels", "=", "$", "residual", "->", "labels", "(", ")", ";", "$", "loss", "=", "0.", ";", "foreach", "(", "$", "predictions", "as", "$", "i", "=>", "$", "prediction", ")", "{", "$", "label", "=", "$", "labels", "[", "$", "i", "]", ";", "$", "loss", "+=", "(", "$", "label", "-", "$", "prediction", ")", "**", "2", ";", "$", "yHat", "[", "$", "i", "]", "=", "$", "label", "-", "(", "$", "this", "->", "rate", "*", "$", "prediction", ")", ";", "}", "$", "loss", "/=", "$", "n", ";", "$", "this", "->", "ensemble", "[", "]", "=", "$", "booster", ";", "$", "this", "->", "steps", "[", "]", "=", "$", "loss", ";", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "\"Epoch $epoch complete, loss=$loss\"", ")", ";", "}", "if", "(", "is_nan", "(", "$", "loss", ")", ")", "{", "break", "1", ";", "}", "if", "(", "abs", "(", "$", "previous", "-", "$", "loss", ")", "<", "$", "this", "->", "minChange", ")", "{", "break", "1", ";", "}", "if", "(", "$", "loss", "<", "$", "this", "->", "tolerance", ")", "{", "break", "1", ";", "}", "$", "residual", "=", "Labeled", "::", "quick", "(", "$", "residual", "->", "samples", "(", ")", ",", "$", "yHat", ")", ";", "$", "previous", "=", "$", "loss", ";", "}", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "'Training complete'", ")", ";", "}", "}" ]
Train the estimator with a dataset. @param \Rubix\ML\Datasets\Dataset $dataset @throws \InvalidArgumentException
[ "Train", "the", "estimator", "with", "a", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Regressors/GradientBoost.php#L225-L325
RubixML/RubixML
src/Regressors/GradientBoost.php
GradientBoost.predict
public function predict(Dataset $dataset) : array { if (empty($this->ensemble)) { throw new RuntimeException('Estimator has not been trained.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); $predictions = $this->base->predict($dataset); foreach ($this->ensemble as $estimator) { foreach ($estimator->predict($dataset) as $j => $prediction) { $predictions[$j] += $this->rate * $prediction; } } return $predictions; }
php
public function predict(Dataset $dataset) : array { if (empty($this->ensemble)) { throw new RuntimeException('Estimator has not been trained.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); $predictions = $this->base->predict($dataset); foreach ($this->ensemble as $estimator) { foreach ($estimator->predict($dataset) as $j => $prediction) { $predictions[$j] += $this->rate * $prediction; } } return $predictions; }
[ "public", "function", "predict", "(", "Dataset", "$", "dataset", ")", ":", "array", "{", "if", "(", "empty", "(", "$", "this", "->", "ensemble", ")", ")", "{", "throw", "new", "RuntimeException", "(", "'Estimator has not been trained.'", ")", ";", "}", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "$", "predictions", "=", "$", "this", "->", "base", "->", "predict", "(", "$", "dataset", ")", ";", "foreach", "(", "$", "this", "->", "ensemble", "as", "$", "estimator", ")", "{", "foreach", "(", "$", "estimator", "->", "predict", "(", "$", "dataset", ")", "as", "$", "j", "=>", "$", "prediction", ")", "{", "$", "predictions", "[", "$", "j", "]", "+=", "$", "this", "->", "rate", "*", "$", "prediction", ";", "}", "}", "return", "$", "predictions", ";", "}" ]
Make a prediction from a dataset. @param \Rubix\ML\Datasets\Dataset $dataset @throws \RuntimeException @throws \InvalidArgumentException @return array
[ "Make", "a", "prediction", "from", "a", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Regressors/GradientBoost.php#L335-L352
RubixML/RubixML
src/Kernels/Distance/Canberra.php
Canberra.compute
public function compute(array $a, array $b) : float { $distance = 0.; foreach ($a as $i => $valueA) { $valueB = $b[$i]; $distance += abs($valueA - $valueB) / ((abs($valueA) + abs($valueB)) ?: EPSILON); } return $distance; }
php
public function compute(array $a, array $b) : float { $distance = 0.; foreach ($a as $i => $valueA) { $valueB = $b[$i]; $distance += abs($valueA - $valueB) / ((abs($valueA) + abs($valueB)) ?: EPSILON); } return $distance; }
[ "public", "function", "compute", "(", "array", "$", "a", ",", "array", "$", "b", ")", ":", "float", "{", "$", "distance", "=", "0.", ";", "foreach", "(", "$", "a", "as", "$", "i", "=>", "$", "valueA", ")", "{", "$", "valueB", "=", "$", "b", "[", "$", "i", "]", ";", "$", "distance", "+=", "abs", "(", "$", "valueA", "-", "$", "valueB", ")", "/", "(", "(", "abs", "(", "$", "valueA", ")", "+", "abs", "(", "$", "valueB", ")", ")", "?", ":", "EPSILON", ")", ";", "}", "return", "$", "distance", ";", "}" ]
Compute the distance between two vectors. @param array $a @param array $b @return float
[ "Compute", "the", "distance", "between", "two", "vectors", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Kernels/Distance/Canberra.php#L26-L38
RubixML/RubixML
src/Transformers/MinMaxNormalizer.php
MinMaxNormalizer.fit
public function fit(Dataset $dataset) : void { $this->minimums = $this->maximums = $this->scales = $this->mins = []; foreach ($dataset->types() as $column => $type) { if ($type === DataType::CONTINUOUS) { $this->minimums[$column] = INF; $this->maximums[$column] = -INF; } } $this->update($dataset); }
php
public function fit(Dataset $dataset) : void { $this->minimums = $this->maximums = $this->scales = $this->mins = []; foreach ($dataset->types() as $column => $type) { if ($type === DataType::CONTINUOUS) { $this->minimums[$column] = INF; $this->maximums[$column] = -INF; } } $this->update($dataset); }
[ "public", "function", "fit", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "$", "this", "->", "minimums", "=", "$", "this", "->", "maximums", "=", "$", "this", "->", "scales", "=", "$", "this", "->", "mins", "=", "[", "]", ";", "foreach", "(", "$", "dataset", "->", "types", "(", ")", "as", "$", "column", "=>", "$", "type", ")", "{", "if", "(", "$", "type", "===", "DataType", "::", "CONTINUOUS", ")", "{", "$", "this", "->", "minimums", "[", "$", "column", "]", "=", "INF", ";", "$", "this", "->", "maximums", "[", "$", "column", "]", "=", "-", "INF", ";", "}", "}", "$", "this", "->", "update", "(", "$", "dataset", ")", ";", "}" ]
Fit the transformer to the dataset. @param \Rubix\ML\Datasets\Dataset $dataset
[ "Fit", "the", "transformer", "to", "the", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/MinMaxNormalizer.php#L117-L129
RubixML/RubixML
src/Transformers/MinMaxNormalizer.php
MinMaxNormalizer.update
public function update(Dataset $dataset) : void { if ($this->minimums === null or $this->maximums === null) { $this->fit($dataset); return; } $columns = $dataset->columnsByType(DataType::CONTINUOUS); foreach ($columns as $column => $values) { $min = min($values); $max = max($values); $min = min($min, $this->minimums[$column]); $max = max($max, $this->maximums[$column]); $scale = ($this->max - $this->min) / (($max - $min) ?: EPSILON); $minHat = $this->min - $min * $scale; $this->minimums[$column] = $min; $this->maximums[$column] = $max; $this->scales[$column] = $scale; $this->mins[$column] = $minHat; } }
php
public function update(Dataset $dataset) : void { if ($this->minimums === null or $this->maximums === null) { $this->fit($dataset); return; } $columns = $dataset->columnsByType(DataType::CONTINUOUS); foreach ($columns as $column => $values) { $min = min($values); $max = max($values); $min = min($min, $this->minimums[$column]); $max = max($max, $this->maximums[$column]); $scale = ($this->max - $this->min) / (($max - $min) ?: EPSILON); $minHat = $this->min - $min * $scale; $this->minimums[$column] = $min; $this->maximums[$column] = $max; $this->scales[$column] = $scale; $this->mins[$column] = $minHat; } }
[ "public", "function", "update", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "if", "(", "$", "this", "->", "minimums", "===", "null", "or", "$", "this", "->", "maximums", "===", "null", ")", "{", "$", "this", "->", "fit", "(", "$", "dataset", ")", ";", "return", ";", "}", "$", "columns", "=", "$", "dataset", "->", "columnsByType", "(", "DataType", "::", "CONTINUOUS", ")", ";", "foreach", "(", "$", "columns", "as", "$", "column", "=>", "$", "values", ")", "{", "$", "min", "=", "min", "(", "$", "values", ")", ";", "$", "max", "=", "max", "(", "$", "values", ")", ";", "$", "min", "=", "min", "(", "$", "min", ",", "$", "this", "->", "minimums", "[", "$", "column", "]", ")", ";", "$", "max", "=", "max", "(", "$", "max", ",", "$", "this", "->", "maximums", "[", "$", "column", "]", ")", ";", "$", "scale", "=", "(", "$", "this", "->", "max", "-", "$", "this", "->", "min", ")", "/", "(", "(", "$", "max", "-", "$", "min", ")", "?", ":", "EPSILON", ")", ";", "$", "minHat", "=", "$", "this", "->", "min", "-", "$", "min", "*", "$", "scale", ";", "$", "this", "->", "minimums", "[", "$", "column", "]", "=", "$", "min", ";", "$", "this", "->", "maximums", "[", "$", "column", "]", "=", "$", "max", ";", "$", "this", "->", "scales", "[", "$", "column", "]", "=", "$", "scale", ";", "$", "this", "->", "mins", "[", "$", "column", "]", "=", "$", "minHat", ";", "}", "}" ]
Update the fitting of the transformer. @param \Rubix\ML\Datasets\Dataset $dataset
[ "Update", "the", "fitting", "of", "the", "transformer", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/MinMaxNormalizer.php#L136-L163
RubixML/RubixML
src/Transformers/MinMaxNormalizer.php
MinMaxNormalizer.transform
public function transform(array &$samples) : void { if ($this->mins === null or $this->scales === null) { throw new RuntimeException('Transformer has not been fitted.'); } foreach ($samples as &$sample) { foreach ($this->scales as $column => $scale) { $sample[$column] *= $scale; $sample[$column] += $this->mins[$column]; } } }
php
public function transform(array &$samples) : void { if ($this->mins === null or $this->scales === null) { throw new RuntimeException('Transformer has not been fitted.'); } foreach ($samples as &$sample) { foreach ($this->scales as $column => $scale) { $sample[$column] *= $scale; $sample[$column] += $this->mins[$column]; } } }
[ "public", "function", "transform", "(", "array", "&", "$", "samples", ")", ":", "void", "{", "if", "(", "$", "this", "->", "mins", "===", "null", "or", "$", "this", "->", "scales", "===", "null", ")", "{", "throw", "new", "RuntimeException", "(", "'Transformer has not been fitted.'", ")", ";", "}", "foreach", "(", "$", "samples", "as", "&", "$", "sample", ")", "{", "foreach", "(", "$", "this", "->", "scales", "as", "$", "column", "=>", "$", "scale", ")", "{", "$", "sample", "[", "$", "column", "]", "*=", "$", "scale", ";", "$", "sample", "[", "$", "column", "]", "+=", "$", "this", "->", "mins", "[", "$", "column", "]", ";", "}", "}", "}" ]
Transform the dataset in place. @param array $samples @throws \RuntimeException
[ "Transform", "the", "dataset", "in", "place", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/MinMaxNormalizer.php#L171-L183
RubixML/RubixML
src/Embedders/TSNE.php
TSNE.predict
public function predict(Dataset $dataset) : array { DatasetIsCompatibleWithEstimator::check($dataset, $this); if ($this->logger) { $this->logger->info('Embedder initialized w/ ' . Params::stringify([ 'dimensions' => $this->dimensions, 'perplexity' => $this->perplexity, 'exaggeration' => $this->exaggeration, 'rate' => $this->rate, 'kernel' => $this->kernel, 'epochs' => $this->epochs, 'min_gradient' => $this->minGradient, 'window' => $this->window, ])); } $n = $dataset->numRows(); $x = Matrix::build($dataset->samples()); if ($this->logger) { $this->logger->info('Computing affinity matrix'); } $distances = $this->pairwiseDistances($x); $p = $this->highAffinities($distances) ->multiply($this->exaggeration); $y = $yHat = Matrix::gaussian($n, $this->dimensions) ->multiply(self::Y_INIT); $velocity = Matrix::zeros($n, $this->dimensions); $gains = Matrix::ones($n, $this->dimensions)->asArray(); $momentum = self::INIT_MOMENTUM; $this->steps = []; for ($epoch = 1; $epoch <= $this->epochs; $epoch++) { $distances = $this->pairwiseDistances($y); $gradient = $this->gradient($p, $y, $distances); $magnitude = $gradient->l2Norm(); $direction = $velocity->multiply($gradient); foreach ($gains as $i => &$row) { $temp = $direction[$i]; foreach ($row as $j => &$gain) { $gain = $temp[$j] < 0. ? $gain + self::GAIN_ACCELERATE : $gain * self::GAIN_BRAKE; $gain = max(self::MIN_GAIN, $gain); } } $gradient = $gradient->multiply(Matrix::quick($gains)); $velocity = $velocity->multiply($momentum) ->subtract($gradient->multiply($this->rate)); $y = $y->add($velocity); $this->steps[] = $magnitude; if ($this->logger) { $this->logger->info("Epoch $epoch complete," . " gradient=$magnitude"); } if (is_nan($magnitude)) { break 1; } if ($magnitude < $this->minGradient) { break 1; } if ($epoch > $this->window) { $window = array_slice($this->steps, -$this->window); $worst = $window; sort($worst); if ($window === $worst) { break 1; } } if ($epoch === $this->early) { $p = $p->divide($this->exaggeration); $momentum += self::MOMENTUM_BOOST; if ($this->logger) { $this->logger->info('Early exaggeration' . ' stage exhausted'); } } } if ($this->logger) { $this->logger->info('Embedding complete'); } return $y->asArray(); }
php
public function predict(Dataset $dataset) : array { DatasetIsCompatibleWithEstimator::check($dataset, $this); if ($this->logger) { $this->logger->info('Embedder initialized w/ ' . Params::stringify([ 'dimensions' => $this->dimensions, 'perplexity' => $this->perplexity, 'exaggeration' => $this->exaggeration, 'rate' => $this->rate, 'kernel' => $this->kernel, 'epochs' => $this->epochs, 'min_gradient' => $this->minGradient, 'window' => $this->window, ])); } $n = $dataset->numRows(); $x = Matrix::build($dataset->samples()); if ($this->logger) { $this->logger->info('Computing affinity matrix'); } $distances = $this->pairwiseDistances($x); $p = $this->highAffinities($distances) ->multiply($this->exaggeration); $y = $yHat = Matrix::gaussian($n, $this->dimensions) ->multiply(self::Y_INIT); $velocity = Matrix::zeros($n, $this->dimensions); $gains = Matrix::ones($n, $this->dimensions)->asArray(); $momentum = self::INIT_MOMENTUM; $this->steps = []; for ($epoch = 1; $epoch <= $this->epochs; $epoch++) { $distances = $this->pairwiseDistances($y); $gradient = $this->gradient($p, $y, $distances); $magnitude = $gradient->l2Norm(); $direction = $velocity->multiply($gradient); foreach ($gains as $i => &$row) { $temp = $direction[$i]; foreach ($row as $j => &$gain) { $gain = $temp[$j] < 0. ? $gain + self::GAIN_ACCELERATE : $gain * self::GAIN_BRAKE; $gain = max(self::MIN_GAIN, $gain); } } $gradient = $gradient->multiply(Matrix::quick($gains)); $velocity = $velocity->multiply($momentum) ->subtract($gradient->multiply($this->rate)); $y = $y->add($velocity); $this->steps[] = $magnitude; if ($this->logger) { $this->logger->info("Epoch $epoch complete," . " gradient=$magnitude"); } if (is_nan($magnitude)) { break 1; } if ($magnitude < $this->minGradient) { break 1; } if ($epoch > $this->window) { $window = array_slice($this->steps, -$this->window); $worst = $window; sort($worst); if ($window === $worst) { break 1; } } if ($epoch === $this->early) { $p = $p->divide($this->exaggeration); $momentum += self::MOMENTUM_BOOST; if ($this->logger) { $this->logger->info('Early exaggeration' . ' stage exhausted'); } } } if ($this->logger) { $this->logger->info('Embedding complete'); } return $y->asArray(); }
[ "public", "function", "predict", "(", "Dataset", "$", "dataset", ")", ":", "array", "{", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "'Embedder initialized w/ '", ".", "Params", "::", "stringify", "(", "[", "'dimensions'", "=>", "$", "this", "->", "dimensions", ",", "'perplexity'", "=>", "$", "this", "->", "perplexity", ",", "'exaggeration'", "=>", "$", "this", "->", "exaggeration", ",", "'rate'", "=>", "$", "this", "->", "rate", ",", "'kernel'", "=>", "$", "this", "->", "kernel", ",", "'epochs'", "=>", "$", "this", "->", "epochs", ",", "'min_gradient'", "=>", "$", "this", "->", "minGradient", ",", "'window'", "=>", "$", "this", "->", "window", ",", "]", ")", ")", ";", "}", "$", "n", "=", "$", "dataset", "->", "numRows", "(", ")", ";", "$", "x", "=", "Matrix", "::", "build", "(", "$", "dataset", "->", "samples", "(", ")", ")", ";", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "'Computing affinity matrix'", ")", ";", "}", "$", "distances", "=", "$", "this", "->", "pairwiseDistances", "(", "$", "x", ")", ";", "$", "p", "=", "$", "this", "->", "highAffinities", "(", "$", "distances", ")", "->", "multiply", "(", "$", "this", "->", "exaggeration", ")", ";", "$", "y", "=", "$", "yHat", "=", "Matrix", "::", "gaussian", "(", "$", "n", ",", "$", "this", "->", "dimensions", ")", "->", "multiply", "(", "self", "::", "Y_INIT", ")", ";", "$", "velocity", "=", "Matrix", "::", "zeros", "(", "$", "n", ",", "$", "this", "->", "dimensions", ")", ";", "$", "gains", "=", "Matrix", "::", "ones", "(", "$", "n", ",", "$", "this", "->", "dimensions", ")", "->", "asArray", "(", ")", ";", "$", "momentum", "=", "self", "::", "INIT_MOMENTUM", ";", "$", "this", "->", "steps", "=", "[", "]", ";", "for", "(", "$", "epoch", "=", "1", ";", "$", "epoch", "<=", "$", "this", "->", "epochs", ";", "$", "epoch", "++", ")", "{", "$", "distances", "=", "$", "this", "->", "pairwiseDistances", "(", "$", "y", ")", ";", "$", "gradient", "=", "$", "this", "->", "gradient", "(", "$", "p", ",", "$", "y", ",", "$", "distances", ")", ";", "$", "magnitude", "=", "$", "gradient", "->", "l2Norm", "(", ")", ";", "$", "direction", "=", "$", "velocity", "->", "multiply", "(", "$", "gradient", ")", ";", "foreach", "(", "$", "gains", "as", "$", "i", "=>", "&", "$", "row", ")", "{", "$", "temp", "=", "$", "direction", "[", "$", "i", "]", ";", "foreach", "(", "$", "row", "as", "$", "j", "=>", "&", "$", "gain", ")", "{", "$", "gain", "=", "$", "temp", "[", "$", "j", "]", "<", "0.", "?", "$", "gain", "+", "self", "::", "GAIN_ACCELERATE", ":", "$", "gain", "*", "self", "::", "GAIN_BRAKE", ";", "$", "gain", "=", "max", "(", "self", "::", "MIN_GAIN", ",", "$", "gain", ")", ";", "}", "}", "$", "gradient", "=", "$", "gradient", "->", "multiply", "(", "Matrix", "::", "quick", "(", "$", "gains", ")", ")", ";", "$", "velocity", "=", "$", "velocity", "->", "multiply", "(", "$", "momentum", ")", "->", "subtract", "(", "$", "gradient", "->", "multiply", "(", "$", "this", "->", "rate", ")", ")", ";", "$", "y", "=", "$", "y", "->", "add", "(", "$", "velocity", ")", ";", "$", "this", "->", "steps", "[", "]", "=", "$", "magnitude", ";", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "\"Epoch $epoch complete,\"", ".", "\" gradient=$magnitude\"", ")", ";", "}", "if", "(", "is_nan", "(", "$", "magnitude", ")", ")", "{", "break", "1", ";", "}", "if", "(", "$", "magnitude", "<", "$", "this", "->", "minGradient", ")", "{", "break", "1", ";", "}", "if", "(", "$", "epoch", ">", "$", "this", "->", "window", ")", "{", "$", "window", "=", "array_slice", "(", "$", "this", "->", "steps", ",", "-", "$", "this", "->", "window", ")", ";", "$", "worst", "=", "$", "window", ";", "sort", "(", "$", "worst", ")", ";", "if", "(", "$", "window", "===", "$", "worst", ")", "{", "break", "1", ";", "}", "}", "if", "(", "$", "epoch", "===", "$", "this", "->", "early", ")", "{", "$", "p", "=", "$", "p", "->", "divide", "(", "$", "this", "->", "exaggeration", ")", ";", "$", "momentum", "+=", "self", "::", "MOMENTUM_BOOST", ";", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "'Early exaggeration'", ".", "' stage exhausted'", ")", ";", "}", "}", "}", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "'Embedding complete'", ")", ";", "}", "return", "$", "y", "->", "asArray", "(", ")", ";", "}" ]
Embed a high dimensional sample matrix into a lower dimensional one. @param \Rubix\ML\Datasets\Dataset $dataset @throws \InvalidArgumentException @return array[]
[ "Embed", "a", "high", "dimensional", "sample", "matrix", "into", "a", "lower", "dimensional", "one", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Embedders/TSNE.php#L254-L365
RubixML/RubixML
src/Embedders/TSNE.php
TSNE.pairwiseDistances
protected function pairwiseDistances(Matrix $samples) : Matrix { $distances = []; foreach ($samples as $a) { $temp = []; foreach ($samples as $b) { $temp[] = $this->kernel->compute($a, $b); } $distances[] = $temp; } return Matrix::quick($distances); }
php
protected function pairwiseDistances(Matrix $samples) : Matrix { $distances = []; foreach ($samples as $a) { $temp = []; foreach ($samples as $b) { $temp[] = $this->kernel->compute($a, $b); } $distances[] = $temp; } return Matrix::quick($distances); }
[ "protected", "function", "pairwiseDistances", "(", "Matrix", "$", "samples", ")", ":", "Matrix", "{", "$", "distances", "=", "[", "]", ";", "foreach", "(", "$", "samples", "as", "$", "a", ")", "{", "$", "temp", "=", "[", "]", ";", "foreach", "(", "$", "samples", "as", "$", "b", ")", "{", "$", "temp", "[", "]", "=", "$", "this", "->", "kernel", "->", "compute", "(", "$", "a", ",", "$", "b", ")", ";", "}", "$", "distances", "[", "]", "=", "$", "temp", ";", "}", "return", "Matrix", "::", "quick", "(", "$", "distances", ")", ";", "}" ]
Calculate the pairwise distances for each sample. @param \Rubix\Tensor\Matrix $samples @return \Rubix\Tensor\Matrix
[ "Calculate", "the", "pairwise", "distances", "for", "each", "sample", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Embedders/TSNE.php#L373-L388
RubixML/RubixML
src/Embedders/TSNE.php
TSNE.highAffinities
protected function highAffinities(Matrix $distances) : Matrix { $zeros = array_fill(0, count($distances), 0); $p = []; foreach ($distances as $i => $row) { $affinities = $zeros; $minBeta = -INF; $maxBeta = INF; $beta = 1.; for ($l = 0; $l < self::BINARY_PRECISION; $l++) { $affinities = []; $pSigma = 0.; foreach ($row as $j => $distance) { if ($i !== $j) { $affinity = exp(-$distance * $beta); } else { $affinity = 0.; } $affinities[] = $affinity; $pSigma += $affinity; } if ($pSigma === 0.) { $pSigma = EPSILON; } $distSigma = 0.; foreach ($affinities as $j => &$prob) { $prob /= $pSigma; $distSigma += $row[$j] * $prob; } $entropy = log($pSigma) + $beta * $distSigma; $diff = $entropy - $this->entropy; if (abs($diff) < self::SEARCH_TOLERANCE) { break 1; } if ($diff > 0.) { $minBeta = $beta; if ($maxBeta === INF) { $beta *= 2.; } else { $beta = ($beta + $maxBeta) / 2.; } } else { $maxBeta = $beta; if ($minBeta === -INF) { $beta /= 2.; } else { $beta = ($beta + $minBeta) / 2.; } } } $p[] = $affinities; } $p = Matrix::quick($p); $pHat = $p->add($p->transpose()); $sigma = $pHat->sum() ->clipLower(EPSILON); return $pHat->divide($sigma); }
php
protected function highAffinities(Matrix $distances) : Matrix { $zeros = array_fill(0, count($distances), 0); $p = []; foreach ($distances as $i => $row) { $affinities = $zeros; $minBeta = -INF; $maxBeta = INF; $beta = 1.; for ($l = 0; $l < self::BINARY_PRECISION; $l++) { $affinities = []; $pSigma = 0.; foreach ($row as $j => $distance) { if ($i !== $j) { $affinity = exp(-$distance * $beta); } else { $affinity = 0.; } $affinities[] = $affinity; $pSigma += $affinity; } if ($pSigma === 0.) { $pSigma = EPSILON; } $distSigma = 0.; foreach ($affinities as $j => &$prob) { $prob /= $pSigma; $distSigma += $row[$j] * $prob; } $entropy = log($pSigma) + $beta * $distSigma; $diff = $entropy - $this->entropy; if (abs($diff) < self::SEARCH_TOLERANCE) { break 1; } if ($diff > 0.) { $minBeta = $beta; if ($maxBeta === INF) { $beta *= 2.; } else { $beta = ($beta + $maxBeta) / 2.; } } else { $maxBeta = $beta; if ($minBeta === -INF) { $beta /= 2.; } else { $beta = ($beta + $minBeta) / 2.; } } } $p[] = $affinities; } $p = Matrix::quick($p); $pHat = $p->add($p->transpose()); $sigma = $pHat->sum() ->clipLower(EPSILON); return $pHat->divide($sigma); }
[ "protected", "function", "highAffinities", "(", "Matrix", "$", "distances", ")", ":", "Matrix", "{", "$", "zeros", "=", "array_fill", "(", "0", ",", "count", "(", "$", "distances", ")", ",", "0", ")", ";", "$", "p", "=", "[", "]", ";", "foreach", "(", "$", "distances", "as", "$", "i", "=>", "$", "row", ")", "{", "$", "affinities", "=", "$", "zeros", ";", "$", "minBeta", "=", "-", "INF", ";", "$", "maxBeta", "=", "INF", ";", "$", "beta", "=", "1.", ";", "for", "(", "$", "l", "=", "0", ";", "$", "l", "<", "self", "::", "BINARY_PRECISION", ";", "$", "l", "++", ")", "{", "$", "affinities", "=", "[", "]", ";", "$", "pSigma", "=", "0.", ";", "foreach", "(", "$", "row", "as", "$", "j", "=>", "$", "distance", ")", "{", "if", "(", "$", "i", "!==", "$", "j", ")", "{", "$", "affinity", "=", "exp", "(", "-", "$", "distance", "*", "$", "beta", ")", ";", "}", "else", "{", "$", "affinity", "=", "0.", ";", "}", "$", "affinities", "[", "]", "=", "$", "affinity", ";", "$", "pSigma", "+=", "$", "affinity", ";", "}", "if", "(", "$", "pSigma", "===", "0.", ")", "{", "$", "pSigma", "=", "EPSILON", ";", "}", "$", "distSigma", "=", "0.", ";", "foreach", "(", "$", "affinities", "as", "$", "j", "=>", "&", "$", "prob", ")", "{", "$", "prob", "/=", "$", "pSigma", ";", "$", "distSigma", "+=", "$", "row", "[", "$", "j", "]", "*", "$", "prob", ";", "}", "$", "entropy", "=", "log", "(", "$", "pSigma", ")", "+", "$", "beta", "*", "$", "distSigma", ";", "$", "diff", "=", "$", "entropy", "-", "$", "this", "->", "entropy", ";", "if", "(", "abs", "(", "$", "diff", ")", "<", "self", "::", "SEARCH_TOLERANCE", ")", "{", "break", "1", ";", "}", "if", "(", "$", "diff", ">", "0.", ")", "{", "$", "minBeta", "=", "$", "beta", ";", "if", "(", "$", "maxBeta", "===", "INF", ")", "{", "$", "beta", "*=", "2.", ";", "}", "else", "{", "$", "beta", "=", "(", "$", "beta", "+", "$", "maxBeta", ")", "/", "2.", ";", "}", "}", "else", "{", "$", "maxBeta", "=", "$", "beta", ";", "if", "(", "$", "minBeta", "===", "-", "INF", ")", "{", "$", "beta", "/=", "2.", ";", "}", "else", "{", "$", "beta", "=", "(", "$", "beta", "+", "$", "minBeta", ")", "/", "2.", ";", "}", "}", "}", "$", "p", "[", "]", "=", "$", "affinities", ";", "}", "$", "p", "=", "Matrix", "::", "quick", "(", "$", "p", ")", ";", "$", "pHat", "=", "$", "p", "->", "add", "(", "$", "p", "->", "transpose", "(", ")", ")", ";", "$", "sigma", "=", "$", "pHat", "->", "sum", "(", ")", "->", "clipLower", "(", "EPSILON", ")", ";", "return", "$", "pHat", "->", "divide", "(", "$", "sigma", ")", ";", "}" ]
Calculate the joint likelihood of each sample in the high dimensional space as being nearest neighbor to each other sample. @param \Rubix\Tensor\Matrix $distances @return \Rubix\Tensor\Matrix
[ "Calculate", "the", "joint", "likelihood", "of", "each", "sample", "in", "the", "high", "dimensional", "space", "as", "being", "nearest", "neighbor", "to", "each", "other", "sample", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Embedders/TSNE.php#L397-L473
RubixML/RubixML
src/Embedders/TSNE.php
TSNE.gradient
protected function gradient(Matrix $p, Matrix $y, Matrix $distances) : Matrix { $q = $distances->square() ->divide($this->degrees) ->add(1.) ->pow((1. + $this->degrees) / -2.); $qSigma = $q->sum()->multiply(2.); $q = $q->divide($qSigma) ->clipLower(EPSILON); $pqd = $p->subtract($q) ->multiply($distances); $c = 2. * (1. + $this->degrees) / $this->degrees; $gradient = []; foreach ($pqd->asVectors() as $i => $row) { $yHat = $y->rowAsVector($i) ->subtract($y); $gradient[] = $row->matmul($yHat) ->multiply($c) ->row(0); } return Matrix::quick($gradient); }
php
protected function gradient(Matrix $p, Matrix $y, Matrix $distances) : Matrix { $q = $distances->square() ->divide($this->degrees) ->add(1.) ->pow((1. + $this->degrees) / -2.); $qSigma = $q->sum()->multiply(2.); $q = $q->divide($qSigma) ->clipLower(EPSILON); $pqd = $p->subtract($q) ->multiply($distances); $c = 2. * (1. + $this->degrees) / $this->degrees; $gradient = []; foreach ($pqd->asVectors() as $i => $row) { $yHat = $y->rowAsVector($i) ->subtract($y); $gradient[] = $row->matmul($yHat) ->multiply($c) ->row(0); } return Matrix::quick($gradient); }
[ "protected", "function", "gradient", "(", "Matrix", "$", "p", ",", "Matrix", "$", "y", ",", "Matrix", "$", "distances", ")", ":", "Matrix", "{", "$", "q", "=", "$", "distances", "->", "square", "(", ")", "->", "divide", "(", "$", "this", "->", "degrees", ")", "->", "add", "(", "1.", ")", "->", "pow", "(", "(", "1.", "+", "$", "this", "->", "degrees", ")", "/", "-", "2.", ")", ";", "$", "qSigma", "=", "$", "q", "->", "sum", "(", ")", "->", "multiply", "(", "2.", ")", ";", "$", "q", "=", "$", "q", "->", "divide", "(", "$", "qSigma", ")", "->", "clipLower", "(", "EPSILON", ")", ";", "$", "pqd", "=", "$", "p", "->", "subtract", "(", "$", "q", ")", "->", "multiply", "(", "$", "distances", ")", ";", "$", "c", "=", "2.", "*", "(", "1.", "+", "$", "this", "->", "degrees", ")", "/", "$", "this", "->", "degrees", ";", "$", "gradient", "=", "[", "]", ";", "foreach", "(", "$", "pqd", "->", "asVectors", "(", ")", "as", "$", "i", "=>", "$", "row", ")", "{", "$", "yHat", "=", "$", "y", "->", "rowAsVector", "(", "$", "i", ")", "->", "subtract", "(", "$", "y", ")", ";", "$", "gradient", "[", "]", "=", "$", "row", "->", "matmul", "(", "$", "yHat", ")", "->", "multiply", "(", "$", "c", ")", "->", "row", "(", "0", ")", ";", "}", "return", "Matrix", "::", "quick", "(", "$", "gradient", ")", ";", "}" ]
Compute the gradient of the KL Divergence cost function with respect to the embedding. @param \Rubix\Tensor\Matrix $p @param \Rubix\Tensor\Matrix $y @param \Rubix\Tensor\Matrix $distances @return \Rubix\Tensor\Matrix
[ "Compute", "the", "gradient", "of", "the", "KL", "Divergence", "cost", "function", "with", "respect", "to", "the", "embedding", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Embedders/TSNE.php#L484-L513
RubixML/RubixML
src/Kernels/Distance/Jaccard.php
Jaccard.compute
public function compute(array $a, array $b) : float { $distance = $min = $max = 0.; foreach ($a as $i => $valueA) { $valueB = $b[$i]; $min += min($valueA, $valueB); $max += max($valueA, $valueB); } return 1. - ($min / ($max ?: EPSILON)); }
php
public function compute(array $a, array $b) : float { $distance = $min = $max = 0.; foreach ($a as $i => $valueA) { $valueB = $b[$i]; $min += min($valueA, $valueB); $max += max($valueA, $valueB); } return 1. - ($min / ($max ?: EPSILON)); }
[ "public", "function", "compute", "(", "array", "$", "a", ",", "array", "$", "b", ")", ":", "float", "{", "$", "distance", "=", "$", "min", "=", "$", "max", "=", "0.", ";", "foreach", "(", "$", "a", "as", "$", "i", "=>", "$", "valueA", ")", "{", "$", "valueB", "=", "$", "b", "[", "$", "i", "]", ";", "$", "min", "+=", "min", "(", "$", "valueA", ",", "$", "valueB", ")", ";", "$", "max", "+=", "max", "(", "$", "valueA", ",", "$", "valueB", ")", ";", "}", "return", "1.", "-", "(", "$", "min", "/", "(", "$", "max", "?", ":", "EPSILON", ")", ")", ";", "}" ]
Compute the distance between two vectors. @param array $a @param array $b @return float
[ "Compute", "the", "distance", "between", "two", "vectors", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Kernels/Distance/Jaccard.php#L27-L39
RubixML/RubixML
src/Transformers/RobustStandardizer.php
RobustStandardizer.fit
public function fit(Dataset $dataset) : void { $columns = $dataset->columnsByType(DataType::CONTINUOUS); $this->medians = $this->mads = []; foreach ($columns as $column => $values) { [$median, $mad] = Stats::medMad($values); $this->medians[$column] = $median; $this->mads[$column] = $mad ?: EPSILON; } }
php
public function fit(Dataset $dataset) : void { $columns = $dataset->columnsByType(DataType::CONTINUOUS); $this->medians = $this->mads = []; foreach ($columns as $column => $values) { [$median, $mad] = Stats::medMad($values); $this->medians[$column] = $median; $this->mads[$column] = $mad ?: EPSILON; } }
[ "public", "function", "fit", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "$", "columns", "=", "$", "dataset", "->", "columnsByType", "(", "DataType", "::", "CONTINUOUS", ")", ";", "$", "this", "->", "medians", "=", "$", "this", "->", "mads", "=", "[", "]", ";", "foreach", "(", "$", "columns", "as", "$", "column", "=>", "$", "values", ")", "{", "[", "$", "median", ",", "$", "mad", "]", "=", "Stats", "::", "medMad", "(", "$", "values", ")", ";", "$", "this", "->", "medians", "[", "$", "column", "]", "=", "$", "median", ";", "$", "this", "->", "mads", "[", "$", "column", "]", "=", "$", "mad", "?", ":", "EPSILON", ";", "}", "}" ]
Fit the transformer to the dataset. @param \Rubix\ML\Datasets\Dataset $dataset
[ "Fit", "the", "transformer", "to", "the", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/RobustStandardizer.php#L91-L103
RubixML/RubixML
src/Transformers/RobustStandardizer.php
RobustStandardizer.transform
public function transform(array &$samples) : void { if ($this->medians === null or $this->mads === null) { throw new RuntimeException('Transformer has not been fitted.'); } foreach ($samples as &$sample) { foreach ($this->mads as $column => $mad) { $feature = $sample[$column]; if ($this->center) { $feature -= $this->medians[$column]; } $sample[$column] = $feature / $mad; } } }
php
public function transform(array &$samples) : void { if ($this->medians === null or $this->mads === null) { throw new RuntimeException('Transformer has not been fitted.'); } foreach ($samples as &$sample) { foreach ($this->mads as $column => $mad) { $feature = $sample[$column]; if ($this->center) { $feature -= $this->medians[$column]; } $sample[$column] = $feature / $mad; } } }
[ "public", "function", "transform", "(", "array", "&", "$", "samples", ")", ":", "void", "{", "if", "(", "$", "this", "->", "medians", "===", "null", "or", "$", "this", "->", "mads", "===", "null", ")", "{", "throw", "new", "RuntimeException", "(", "'Transformer has not been fitted.'", ")", ";", "}", "foreach", "(", "$", "samples", "as", "&", "$", "sample", ")", "{", "foreach", "(", "$", "this", "->", "mads", "as", "$", "column", "=>", "$", "mad", ")", "{", "$", "feature", "=", "$", "sample", "[", "$", "column", "]", ";", "if", "(", "$", "this", "->", "center", ")", "{", "$", "feature", "-=", "$", "this", "->", "medians", "[", "$", "column", "]", ";", "}", "$", "sample", "[", "$", "column", "]", "=", "$", "feature", "/", "$", "mad", ";", "}", "}", "}" ]
Transform the dataset in place. @param array $samples @throws \RuntimeException
[ "Transform", "the", "dataset", "in", "place", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/RobustStandardizer.php#L111-L128
RubixML/RubixML
src/CrossValidation/Reports/ContingencyTable.php
ContingencyTable.generate
public function generate(array $predictions, array $labels) : array { if (count($predictions) !== count($labels)) { throw new InvalidArgumentException('The number of labels' . ' must equal the number of predictions.'); } $classes = array_unique($labels); $clusters = array_unique($predictions); $table = array_fill_keys($clusters, array_fill_keys($classes, 0)); foreach ($labels as $i => $class) { $table[$predictions[$i]][$class]++; } return $table; }
php
public function generate(array $predictions, array $labels) : array { if (count($predictions) !== count($labels)) { throw new InvalidArgumentException('The number of labels' . ' must equal the number of predictions.'); } $classes = array_unique($labels); $clusters = array_unique($predictions); $table = array_fill_keys($clusters, array_fill_keys($classes, 0)); foreach ($labels as $i => $class) { $table[$predictions[$i]][$class]++; } return $table; }
[ "public", "function", "generate", "(", "array", "$", "predictions", ",", "array", "$", "labels", ")", ":", "array", "{", "if", "(", "count", "(", "$", "predictions", ")", "!==", "count", "(", "$", "labels", ")", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'The number of labels'", ".", "' must equal the number of predictions.'", ")", ";", "}", "$", "classes", "=", "array_unique", "(", "$", "labels", ")", ";", "$", "clusters", "=", "array_unique", "(", "$", "predictions", ")", ";", "$", "table", "=", "array_fill_keys", "(", "$", "clusters", ",", "array_fill_keys", "(", "$", "classes", ",", "0", ")", ")", ";", "foreach", "(", "$", "labels", "as", "$", "i", "=>", "$", "class", ")", "{", "$", "table", "[", "$", "predictions", "[", "$", "i", "]", "]", "[", "$", "class", "]", "++", ";", "}", "return", "$", "table", ";", "}" ]
Generate the report. @param array $predictions @param array $labels @throws \InvalidArgumentException @return array
[ "Generate", "the", "report", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/CrossValidation/Reports/ContingencyTable.php#L41-L58
RubixML/RubixML
src/Transformers/ImageVectorizer.php
ImageVectorizer.transform
public function transform(array &$samples) : void { foreach ($samples as &$sample) { $vectors = []; foreach ($sample as $column => $image) { if (is_resource($image) ? get_resource_type($image) === 'gd' : false) { $width = imagesx($image); $height = imagesy($image); $vector = []; for ($x = 0; $x < $width; $x++) { for ($y = 0; $y < $height; $y++) { $pixel = imagecolorat($image, $x, $y); $vector[] = $pixel & 0xFF; for ($i = 8; $i <= $this->mu; $i *= 2) { $vector[] = ($pixel >> $i) & 0xFF; } } } unset($sample[$column]); imagedestroy($image); $vectors[] = $vector; } } $sample = array_merge($sample, ...$vectors); } }
php
public function transform(array &$samples) : void { foreach ($samples as &$sample) { $vectors = []; foreach ($sample as $column => $image) { if (is_resource($image) ? get_resource_type($image) === 'gd' : false) { $width = imagesx($image); $height = imagesy($image); $vector = []; for ($x = 0; $x < $width; $x++) { for ($y = 0; $y < $height; $y++) { $pixel = imagecolorat($image, $x, $y); $vector[] = $pixel & 0xFF; for ($i = 8; $i <= $this->mu; $i *= 2) { $vector[] = ($pixel >> $i) & 0xFF; } } } unset($sample[$column]); imagedestroy($image); $vectors[] = $vector; } } $sample = array_merge($sample, ...$vectors); } }
[ "public", "function", "transform", "(", "array", "&", "$", "samples", ")", ":", "void", "{", "foreach", "(", "$", "samples", "as", "&", "$", "sample", ")", "{", "$", "vectors", "=", "[", "]", ";", "foreach", "(", "$", "sample", "as", "$", "column", "=>", "$", "image", ")", "{", "if", "(", "is_resource", "(", "$", "image", ")", "?", "get_resource_type", "(", "$", "image", ")", "===", "'gd'", ":", "false", ")", "{", "$", "width", "=", "imagesx", "(", "$", "image", ")", ";", "$", "height", "=", "imagesy", "(", "$", "image", ")", ";", "$", "vector", "=", "[", "]", ";", "for", "(", "$", "x", "=", "0", ";", "$", "x", "<", "$", "width", ";", "$", "x", "++", ")", "{", "for", "(", "$", "y", "=", "0", ";", "$", "y", "<", "$", "height", ";", "$", "y", "++", ")", "{", "$", "pixel", "=", "imagecolorat", "(", "$", "image", ",", "$", "x", ",", "$", "y", ")", ";", "$", "vector", "[", "]", "=", "$", "pixel", "&", "0xFF", ";", "for", "(", "$", "i", "=", "8", ";", "$", "i", "<=", "$", "this", "->", "mu", ";", "$", "i", "*=", "2", ")", "{", "$", "vector", "[", "]", "=", "(", "$", "pixel", ">>", "$", "i", ")", "&", "0xFF", ";", "}", "}", "}", "unset", "(", "$", "sample", "[", "$", "column", "]", ")", ";", "imagedestroy", "(", "$", "image", ")", ";", "$", "vectors", "[", "]", "=", "$", "vector", ";", "}", "}", "$", "sample", "=", "array_merge", "(", "$", "sample", ",", "...", "$", "vectors", ")", ";", "}", "}" ]
Transform the dataset in place. @param array $samples
[ "Transform", "the", "dataset", "in", "place", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/ImageVectorizer.php#L62-L96
RubixML/RubixML
src/BootstrapAggregator.php
BootstrapAggregator.train
public function train(Dataset $dataset) : void { if ($this->type() === self::CLASSIFIER or $this->type() === self::REGRESSOR) { if (!$dataset instanceof Labeled) { throw new InvalidArgumentException('This estimator requires a' . ' labeled training set.'); } } DatasetIsCompatibleWithEstimator::check($dataset, $this); $p = (int) round($this->ratio * $dataset->numRows()); $this->ensemble = []; Loop::run(function () use ($dataset, $p) { $pool = new DefaultPool($this->workers); $coroutines = []; for ($i = 0; $i < $this->estimators; $i++) { $estimator = clone $this->base; $subset = $dataset->randomSubsetWithReplacement($p); $task = new CallableTask( [$this, '_train'], [$estimator, $subset] ); $coroutines[] = call(function () use ($pool, $task) { return yield $pool->enqueue($task); }); } $this->ensemble = yield all($coroutines); return yield $pool->shutdown(); }); }
php
public function train(Dataset $dataset) : void { if ($this->type() === self::CLASSIFIER or $this->type() === self::REGRESSOR) { if (!$dataset instanceof Labeled) { throw new InvalidArgumentException('This estimator requires a' . ' labeled training set.'); } } DatasetIsCompatibleWithEstimator::check($dataset, $this); $p = (int) round($this->ratio * $dataset->numRows()); $this->ensemble = []; Loop::run(function () use ($dataset, $p) { $pool = new DefaultPool($this->workers); $coroutines = []; for ($i = 0; $i < $this->estimators; $i++) { $estimator = clone $this->base; $subset = $dataset->randomSubsetWithReplacement($p); $task = new CallableTask( [$this, '_train'], [$estimator, $subset] ); $coroutines[] = call(function () use ($pool, $task) { return yield $pool->enqueue($task); }); } $this->ensemble = yield all($coroutines); return yield $pool->shutdown(); }); }
[ "public", "function", "train", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "if", "(", "$", "this", "->", "type", "(", ")", "===", "self", "::", "CLASSIFIER", "or", "$", "this", "->", "type", "(", ")", "===", "self", "::", "REGRESSOR", ")", "{", "if", "(", "!", "$", "dataset", "instanceof", "Labeled", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'This estimator requires a'", ".", "' labeled training set.'", ")", ";", "}", "}", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "$", "p", "=", "(", "int", ")", "round", "(", "$", "this", "->", "ratio", "*", "$", "dataset", "->", "numRows", "(", ")", ")", ";", "$", "this", "->", "ensemble", "=", "[", "]", ";", "Loop", "::", "run", "(", "function", "(", ")", "use", "(", "$", "dataset", ",", "$", "p", ")", "{", "$", "pool", "=", "new", "DefaultPool", "(", "$", "this", "->", "workers", ")", ";", "$", "coroutines", "=", "[", "]", ";", "for", "(", "$", "i", "=", "0", ";", "$", "i", "<", "$", "this", "->", "estimators", ";", "$", "i", "++", ")", "{", "$", "estimator", "=", "clone", "$", "this", "->", "base", ";", "$", "subset", "=", "$", "dataset", "->", "randomSubsetWithReplacement", "(", "$", "p", ")", ";", "$", "task", "=", "new", "CallableTask", "(", "[", "$", "this", ",", "'_train'", "]", ",", "[", "$", "estimator", ",", "$", "subset", "]", ")", ";", "$", "coroutines", "[", "]", "=", "call", "(", "function", "(", ")", "use", "(", "$", "pool", ",", "$", "task", ")", "{", "return", "yield", "$", "pool", "->", "enqueue", "(", "$", "task", ")", ";", "}", ")", ";", "}", "$", "this", "->", "ensemble", "=", "yield", "all", "(", "$", "coroutines", ")", ";", "return", "yield", "$", "pool", "->", "shutdown", "(", ")", ";", "}", ")", ";", "}" ]
Instantiate and train each base estimator in the ensemble on a bootstrap training set. @param \Rubix\ML\Datasets\Dataset $dataset @throws \InvalidArgumentException
[ "Instantiate", "and", "train", "each", "base", "estimator", "in", "the", "ensemble", "on", "a", "bootstrap", "training", "set", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/BootstrapAggregator.php#L141-L180
RubixML/RubixML
src/BootstrapAggregator.php
BootstrapAggregator.predict
public function predict(Dataset $dataset) : array { if (empty($this->ensemble)) { throw new RuntimeException('Estimator has not been trained.'); } $aggregate = []; foreach ($this->ensemble as $estimator) { foreach ($estimator->predict($dataset) as $i => $prediction) { $aggregate[$i][] = $prediction; } } switch ($this->type()) { case self::CLASSIFIER: return array_map([self::class, 'decideClass'], $aggregate); case self::REGRESSOR: return array_map([Stats::class, 'mean'], $aggregate); case self::ANOMALY_DETECTOR: return array_map([self::class, 'decideAnomaly'], $aggregate); } }
php
public function predict(Dataset $dataset) : array { if (empty($this->ensemble)) { throw new RuntimeException('Estimator has not been trained.'); } $aggregate = []; foreach ($this->ensemble as $estimator) { foreach ($estimator->predict($dataset) as $i => $prediction) { $aggregate[$i][] = $prediction; } } switch ($this->type()) { case self::CLASSIFIER: return array_map([self::class, 'decideClass'], $aggregate); case self::REGRESSOR: return array_map([Stats::class, 'mean'], $aggregate); case self::ANOMALY_DETECTOR: return array_map([self::class, 'decideAnomaly'], $aggregate); } }
[ "public", "function", "predict", "(", "Dataset", "$", "dataset", ")", ":", "array", "{", "if", "(", "empty", "(", "$", "this", "->", "ensemble", ")", ")", "{", "throw", "new", "RuntimeException", "(", "'Estimator has not been trained.'", ")", ";", "}", "$", "aggregate", "=", "[", "]", ";", "foreach", "(", "$", "this", "->", "ensemble", "as", "$", "estimator", ")", "{", "foreach", "(", "$", "estimator", "->", "predict", "(", "$", "dataset", ")", "as", "$", "i", "=>", "$", "prediction", ")", "{", "$", "aggregate", "[", "$", "i", "]", "[", "]", "=", "$", "prediction", ";", "}", "}", "switch", "(", "$", "this", "->", "type", "(", ")", ")", "{", "case", "self", "::", "CLASSIFIER", ":", "return", "array_map", "(", "[", "self", "::", "class", ",", "'decideClass'", "]", ",", "$", "aggregate", ")", ";", "case", "self", "::", "REGRESSOR", ":", "return", "array_map", "(", "[", "Stats", "::", "class", ",", "'mean'", "]", ",", "$", "aggregate", ")", ";", "case", "self", "::", "ANOMALY_DETECTOR", ":", "return", "array_map", "(", "[", "self", "::", "class", ",", "'decideAnomaly'", "]", ",", "$", "aggregate", ")", ";", "}", "}" ]
Make predictions from a dataset. @param \Rubix\ML\Datasets\Dataset $dataset @throws \RuntimeException @return array
[ "Make", "predictions", "from", "a", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/BootstrapAggregator.php#L189-L214
RubixML/RubixML
src/Graph/Nodes/Isolator.php
Isolator.split
public static function split(Dataset $dataset) : self { $column = rand(0, $dataset->numColumns() - 1); $sample = $dataset[rand(0, count($dataset) - 1)]; $value = $sample[$column]; $groups = $dataset->partition($column, $value); return new self($column, $value, $groups); }
php
public static function split(Dataset $dataset) : self { $column = rand(0, $dataset->numColumns() - 1); $sample = $dataset[rand(0, count($dataset) - 1)]; $value = $sample[$column]; $groups = $dataset->partition($column, $value); return new self($column, $value, $groups); }
[ "public", "static", "function", "split", "(", "Dataset", "$", "dataset", ")", ":", "self", "{", "$", "column", "=", "rand", "(", "0", ",", "$", "dataset", "->", "numColumns", "(", ")", "-", "1", ")", ";", "$", "sample", "=", "$", "dataset", "[", "rand", "(", "0", ",", "count", "(", "$", "dataset", ")", "-", "1", ")", "]", ";", "$", "value", "=", "$", "sample", "[", "$", "column", "]", ";", "$", "groups", "=", "$", "dataset", "->", "partition", "(", "$", "column", ",", "$", "value", ")", ";", "return", "new", "self", "(", "$", "column", ",", "$", "value", ",", "$", "groups", ")", ";", "}" ]
Factory method to build a isolator node from a dataset using a random split of the dataset. @param \Rubix\ML\Datasets\Dataset $dataset @return self
[ "Factory", "method", "to", "build", "a", "isolator", "node", "from", "a", "dataset", "using", "a", "random", "split", "of", "the", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Graph/Nodes/Isolator.php#L48-L59
RubixML/RubixML
src/Transformers/OneHotEncoder.php
OneHotEncoder.fit
public function fit(Dataset $dataset) : void { $columns = $dataset->columnsByType(DataType::CATEGORICAL); $this->categories = []; foreach ($columns as $column => $values) { $categories = array_values(array_unique($values)); $this->categories[$column] = array_flip($categories); } }
php
public function fit(Dataset $dataset) : void { $columns = $dataset->columnsByType(DataType::CATEGORICAL); $this->categories = []; foreach ($columns as $column => $values) { $categories = array_values(array_unique($values)); $this->categories[$column] = array_flip($categories); } }
[ "public", "function", "fit", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "$", "columns", "=", "$", "dataset", "->", "columnsByType", "(", "DataType", "::", "CATEGORICAL", ")", ";", "$", "this", "->", "categories", "=", "[", "]", ";", "foreach", "(", "$", "columns", "as", "$", "column", "=>", "$", "values", ")", "{", "$", "categories", "=", "array_values", "(", "array_unique", "(", "$", "values", ")", ")", ";", "$", "this", "->", "categories", "[", "$", "column", "]", "=", "array_flip", "(", "$", "categories", ")", ";", "}", "}" ]
Fit the transformer to the dataset. @param \Rubix\ML\Datasets\Dataset $dataset
[ "Fit", "the", "transformer", "to", "the", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/OneHotEncoder.php#L45-L56
RubixML/RubixML
src/Transformers/OneHotEncoder.php
OneHotEncoder.transform
public function transform(array &$samples) : void { if ($this->categories === null) { throw new RuntimeException('Transformer has not been fitted.'); } $templates = []; foreach ($this->categories as $column => $categories) { $templates[$column] = array_fill(0, count($categories), 0); } foreach ($samples as &$sample) { $vector = []; foreach ($this->categories as $column => $categories) { $category = $sample[$column]; $features = $templates[$column]; if (isset($categories[$category])) { $features[$categories[$category]] = 1; } $vector = array_merge($vector, $features); unset($sample[$column]); } $sample = array_merge($sample, $vector); } }
php
public function transform(array &$samples) : void { if ($this->categories === null) { throw new RuntimeException('Transformer has not been fitted.'); } $templates = []; foreach ($this->categories as $column => $categories) { $templates[$column] = array_fill(0, count($categories), 0); } foreach ($samples as &$sample) { $vector = []; foreach ($this->categories as $column => $categories) { $category = $sample[$column]; $features = $templates[$column]; if (isset($categories[$category])) { $features[$categories[$category]] = 1; } $vector = array_merge($vector, $features); unset($sample[$column]); } $sample = array_merge($sample, $vector); } }
[ "public", "function", "transform", "(", "array", "&", "$", "samples", ")", ":", "void", "{", "if", "(", "$", "this", "->", "categories", "===", "null", ")", "{", "throw", "new", "RuntimeException", "(", "'Transformer has not been fitted.'", ")", ";", "}", "$", "templates", "=", "[", "]", ";", "foreach", "(", "$", "this", "->", "categories", "as", "$", "column", "=>", "$", "categories", ")", "{", "$", "templates", "[", "$", "column", "]", "=", "array_fill", "(", "0", ",", "count", "(", "$", "categories", ")", ",", "0", ")", ";", "}", "foreach", "(", "$", "samples", "as", "&", "$", "sample", ")", "{", "$", "vector", "=", "[", "]", ";", "foreach", "(", "$", "this", "->", "categories", "as", "$", "column", "=>", "$", "categories", ")", "{", "$", "category", "=", "$", "sample", "[", "$", "column", "]", ";", "$", "features", "=", "$", "templates", "[", "$", "column", "]", ";", "if", "(", "isset", "(", "$", "categories", "[", "$", "category", "]", ")", ")", "{", "$", "features", "[", "$", "categories", "[", "$", "category", "]", "]", "=", "1", ";", "}", "$", "vector", "=", "array_merge", "(", "$", "vector", ",", "$", "features", ")", ";", "unset", "(", "$", "sample", "[", "$", "column", "]", ")", ";", "}", "$", "sample", "=", "array_merge", "(", "$", "sample", ",", "$", "vector", ")", ";", "}", "}" ]
Transform the dataset in place. @param array $samples @throws \RuntimeException
[ "Transform", "the", "dataset", "in", "place", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/OneHotEncoder.php#L64-L94
RubixML/RubixML
src/Clusterers/MeanShift.php
MeanShift.estimateRadius
public static function estimateRadius(Dataset $dataset, float $percentile = 30., ?Distance $kernel = null) : float { if ($percentile < 0. or $percentile > 100.) { throw new InvalidArgumentException('Percentile must be between' . " 0 and 100, $percentile given."); } $kernel = $kernel ?? new Euclidean(); $distances = []; foreach ($dataset as $sampleA) { foreach ($dataset as $sampleB) { $distances[] = $kernel->compute($sampleA, $sampleB); } } return Stats::percentile($distances, $percentile); }
php
public static function estimateRadius(Dataset $dataset, float $percentile = 30., ?Distance $kernel = null) : float { if ($percentile < 0. or $percentile > 100.) { throw new InvalidArgumentException('Percentile must be between' . " 0 and 100, $percentile given."); } $kernel = $kernel ?? new Euclidean(); $distances = []; foreach ($dataset as $sampleA) { foreach ($dataset as $sampleB) { $distances[] = $kernel->compute($sampleA, $sampleB); } } return Stats::percentile($distances, $percentile); }
[ "public", "static", "function", "estimateRadius", "(", "Dataset", "$", "dataset", ",", "float", "$", "percentile", "=", "30.", ",", "?", "Distance", "$", "kernel", "=", "null", ")", ":", "float", "{", "if", "(", "$", "percentile", "<", "0.", "or", "$", "percentile", ">", "100.", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'Percentile must be between'", ".", "\" 0 and 100, $percentile given.\"", ")", ";", "}", "$", "kernel", "=", "$", "kernel", "??", "new", "Euclidean", "(", ")", ";", "$", "distances", "=", "[", "]", ";", "foreach", "(", "$", "dataset", "as", "$", "sampleA", ")", "{", "foreach", "(", "$", "dataset", "as", "$", "sampleB", ")", "{", "$", "distances", "[", "]", "=", "$", "kernel", "->", "compute", "(", "$", "sampleA", ",", "$", "sampleB", ")", ";", "}", "}", "return", "Stats", "::", "percentile", "(", "$", "distances", ",", "$", "percentile", ")", ";", "}" ]
Estimate the radius of a cluster that encompasses a certain percentage of the total training samples. > **Note**: Since radius estimation scales quadratically in the number of samples, for large datasets you can speed up the process by running it on a sample subset of the training data. @param \Rubix\ML\Datasets\Dataset $dataset @param float $percentile @param \Rubix\ML\Kernels\Distance\Distance|null $kernel @throws \InvalidArgumentException @return float
[ "Estimate", "the", "radius", "of", "a", "cluster", "that", "encompasses", "a", "certain", "percentage", "of", "the", "total", "training", "samples", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Clusterers/MeanShift.php#L137-L155
RubixML/RubixML
src/Clusterers/MeanShift.php
MeanShift.train
public function train(Dataset $dataset) : void { DatasetIsCompatibleWithEstimator::check($dataset, $this); if ($this->logger) { $this->logger->info('Learner init ' . Params::stringify([ 'radius' => $this->radius, 'kernel' => $this->kernel, 'max_leaf_size' => $this->maxLeafSize, 'epochs' => $this->epochs, 'min_change' => $this->minChange, 'seeder' => $this->seeder, 'ratio' => $this->ratio, ])); } $n = $dataset->numRows(); $tree = new BallTree($this->maxLeafSize, $this->kernel); if ($this->seeder and $n > self::MIN_SEEDS) { $k = (int) round($this->ratio * $n); $centroids = $this->seeder->seed($dataset, $k); } else { $centroids = $dataset->samples(); } $tree->grow($dataset); $previous = $centroids; $this->steps = []; for ($epoch = 1; $epoch <= $this->epochs; $epoch++) { foreach ($centroids as $i => &$centroid) { [$samples, $labels, $distances] = $tree->range($centroid, $this->radius); $step = Matrix::quick($samples) ->transpose() ->mean() ->asArray(); $mu2 = Stats::mean($distances) ** 2; $weight = exp(-$mu2 / $this->delta); foreach ($centroid as $column => &$mean) { $mean = ($weight * $step[$column]) / $weight; } foreach ($centroids as $j => $neighbor) { if ($i === $j) { continue 1; } $distance = $this->kernel->compute($centroid, $neighbor); if ($distance < $this->radius) { unset($centroids[$j]); } } } $shift = $this->shift($centroids, $previous); $this->steps[] = $shift; if ($this->logger) { $this->logger->info("Epoch $epoch complete, shift=$shift"); } if (is_nan($shift)) { break 1; } if ($shift < $this->minChange) { break 1; } $previous = $centroids; } $this->centroids = array_values($centroids); if ($this->logger) { $this->logger->info('Training complete'); } }
php
public function train(Dataset $dataset) : void { DatasetIsCompatibleWithEstimator::check($dataset, $this); if ($this->logger) { $this->logger->info('Learner init ' . Params::stringify([ 'radius' => $this->radius, 'kernel' => $this->kernel, 'max_leaf_size' => $this->maxLeafSize, 'epochs' => $this->epochs, 'min_change' => $this->minChange, 'seeder' => $this->seeder, 'ratio' => $this->ratio, ])); } $n = $dataset->numRows(); $tree = new BallTree($this->maxLeafSize, $this->kernel); if ($this->seeder and $n > self::MIN_SEEDS) { $k = (int) round($this->ratio * $n); $centroids = $this->seeder->seed($dataset, $k); } else { $centroids = $dataset->samples(); } $tree->grow($dataset); $previous = $centroids; $this->steps = []; for ($epoch = 1; $epoch <= $this->epochs; $epoch++) { foreach ($centroids as $i => &$centroid) { [$samples, $labels, $distances] = $tree->range($centroid, $this->radius); $step = Matrix::quick($samples) ->transpose() ->mean() ->asArray(); $mu2 = Stats::mean($distances) ** 2; $weight = exp(-$mu2 / $this->delta); foreach ($centroid as $column => &$mean) { $mean = ($weight * $step[$column]) / $weight; } foreach ($centroids as $j => $neighbor) { if ($i === $j) { continue 1; } $distance = $this->kernel->compute($centroid, $neighbor); if ($distance < $this->radius) { unset($centroids[$j]); } } } $shift = $this->shift($centroids, $previous); $this->steps[] = $shift; if ($this->logger) { $this->logger->info("Epoch $epoch complete, shift=$shift"); } if (is_nan($shift)) { break 1; } if ($shift < $this->minChange) { break 1; } $previous = $centroids; } $this->centroids = array_values($centroids); if ($this->logger) { $this->logger->info('Training complete'); } }
[ "public", "function", "train", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "'Learner init '", ".", "Params", "::", "stringify", "(", "[", "'radius'", "=>", "$", "this", "->", "radius", ",", "'kernel'", "=>", "$", "this", "->", "kernel", ",", "'max_leaf_size'", "=>", "$", "this", "->", "maxLeafSize", ",", "'epochs'", "=>", "$", "this", "->", "epochs", ",", "'min_change'", "=>", "$", "this", "->", "minChange", ",", "'seeder'", "=>", "$", "this", "->", "seeder", ",", "'ratio'", "=>", "$", "this", "->", "ratio", ",", "]", ")", ")", ";", "}", "$", "n", "=", "$", "dataset", "->", "numRows", "(", ")", ";", "$", "tree", "=", "new", "BallTree", "(", "$", "this", "->", "maxLeafSize", ",", "$", "this", "->", "kernel", ")", ";", "if", "(", "$", "this", "->", "seeder", "and", "$", "n", ">", "self", "::", "MIN_SEEDS", ")", "{", "$", "k", "=", "(", "int", ")", "round", "(", "$", "this", "->", "ratio", "*", "$", "n", ")", ";", "$", "centroids", "=", "$", "this", "->", "seeder", "->", "seed", "(", "$", "dataset", ",", "$", "k", ")", ";", "}", "else", "{", "$", "centroids", "=", "$", "dataset", "->", "samples", "(", ")", ";", "}", "$", "tree", "->", "grow", "(", "$", "dataset", ")", ";", "$", "previous", "=", "$", "centroids", ";", "$", "this", "->", "steps", "=", "[", "]", ";", "for", "(", "$", "epoch", "=", "1", ";", "$", "epoch", "<=", "$", "this", "->", "epochs", ";", "$", "epoch", "++", ")", "{", "foreach", "(", "$", "centroids", "as", "$", "i", "=>", "&", "$", "centroid", ")", "{", "[", "$", "samples", ",", "$", "labels", ",", "$", "distances", "]", "=", "$", "tree", "->", "range", "(", "$", "centroid", ",", "$", "this", "->", "radius", ")", ";", "$", "step", "=", "Matrix", "::", "quick", "(", "$", "samples", ")", "->", "transpose", "(", ")", "->", "mean", "(", ")", "->", "asArray", "(", ")", ";", "$", "mu2", "=", "Stats", "::", "mean", "(", "$", "distances", ")", "**", "2", ";", "$", "weight", "=", "exp", "(", "-", "$", "mu2", "/", "$", "this", "->", "delta", ")", ";", "foreach", "(", "$", "centroid", "as", "$", "column", "=>", "&", "$", "mean", ")", "{", "$", "mean", "=", "(", "$", "weight", "*", "$", "step", "[", "$", "column", "]", ")", "/", "$", "weight", ";", "}", "foreach", "(", "$", "centroids", "as", "$", "j", "=>", "$", "neighbor", ")", "{", "if", "(", "$", "i", "===", "$", "j", ")", "{", "continue", "1", ";", "}", "$", "distance", "=", "$", "this", "->", "kernel", "->", "compute", "(", "$", "centroid", ",", "$", "neighbor", ")", ";", "if", "(", "$", "distance", "<", "$", "this", "->", "radius", ")", "{", "unset", "(", "$", "centroids", "[", "$", "j", "]", ")", ";", "}", "}", "}", "$", "shift", "=", "$", "this", "->", "shift", "(", "$", "centroids", ",", "$", "previous", ")", ";", "$", "this", "->", "steps", "[", "]", "=", "$", "shift", ";", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "\"Epoch $epoch complete, shift=$shift\"", ")", ";", "}", "if", "(", "is_nan", "(", "$", "shift", ")", ")", "{", "break", "1", ";", "}", "if", "(", "$", "shift", "<", "$", "this", "->", "minChange", ")", "{", "break", "1", ";", "}", "$", "previous", "=", "$", "centroids", ";", "}", "$", "this", "->", "centroids", "=", "array_values", "(", "$", "centroids", ")", ";", "if", "(", "$", "this", "->", "logger", ")", "{", "$", "this", "->", "logger", "->", "info", "(", "'Training complete'", ")", ";", "}", "}" ]
Train the learner with a dataset. @param \Rubix\ML\Datasets\Dataset $dataset @throws \InvalidArgumentException
[ "Train", "the", "learner", "with", "a", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Clusterers/MeanShift.php#L269-L357
RubixML/RubixML
src/Clusterers/MeanShift.php
MeanShift.assign
protected function assign(array $sample) : int { $bestDistance = INF; $bestCluster = -1; foreach ($this->centroids as $cluster => $centroid) { $distance = $this->kernel->compute($sample, $centroid); if ($distance < $bestDistance) { $bestDistance = $distance; $bestCluster = $cluster; } } return (int) $bestCluster; }
php
protected function assign(array $sample) : int { $bestDistance = INF; $bestCluster = -1; foreach ($this->centroids as $cluster => $centroid) { $distance = $this->kernel->compute($sample, $centroid); if ($distance < $bestDistance) { $bestDistance = $distance; $bestCluster = $cluster; } } return (int) $bestCluster; }
[ "protected", "function", "assign", "(", "array", "$", "sample", ")", ":", "int", "{", "$", "bestDistance", "=", "INF", ";", "$", "bestCluster", "=", "-", "1", ";", "foreach", "(", "$", "this", "->", "centroids", "as", "$", "cluster", "=>", "$", "centroid", ")", "{", "$", "distance", "=", "$", "this", "->", "kernel", "->", "compute", "(", "$", "sample", ",", "$", "centroid", ")", ";", "if", "(", "$", "distance", "<", "$", "bestDistance", ")", "{", "$", "bestDistance", "=", "$", "distance", ";", "$", "bestCluster", "=", "$", "cluster", ";", "}", "}", "return", "(", "int", ")", "$", "bestCluster", ";", "}" ]
Label a given sample based on its distance from a particular centroid. @param array $sample @return int
[ "Label", "a", "given", "sample", "based", "on", "its", "distance", "from", "a", "particular", "centroid", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Clusterers/MeanShift.php#L403-L418
RubixML/RubixML
src/Clusterers/MeanShift.php
MeanShift.membership
protected function membership(array $sample) : array { $membership = $distances = []; foreach ($this->centroids as $centroid) { $distances[] = $this->kernel->compute($sample, $centroid); } $total = array_sum($distances) ?: EPSILON; foreach ($distances as $distance) { $membership[] = $distance / $total; } return $membership; }
php
protected function membership(array $sample) : array { $membership = $distances = []; foreach ($this->centroids as $centroid) { $distances[] = $this->kernel->compute($sample, $centroid); } $total = array_sum($distances) ?: EPSILON; foreach ($distances as $distance) { $membership[] = $distance / $total; } return $membership; }
[ "protected", "function", "membership", "(", "array", "$", "sample", ")", ":", "array", "{", "$", "membership", "=", "$", "distances", "=", "[", "]", ";", "foreach", "(", "$", "this", "->", "centroids", "as", "$", "centroid", ")", "{", "$", "distances", "[", "]", "=", "$", "this", "->", "kernel", "->", "compute", "(", "$", "sample", ",", "$", "centroid", ")", ";", "}", "$", "total", "=", "array_sum", "(", "$", "distances", ")", "?", ":", "EPSILON", ";", "foreach", "(", "$", "distances", "as", "$", "distance", ")", "{", "$", "membership", "[", "]", "=", "$", "distance", "/", "$", "total", ";", "}", "return", "$", "membership", ";", "}" ]
Return the membership of a sample to each of the centroids. @param array $sample @return array
[ "Return", "the", "membership", "of", "a", "sample", "to", "each", "of", "the", "centroids", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Clusterers/MeanShift.php#L426-L441
RubixML/RubixML
src/Clusterers/MeanShift.php
MeanShift.shift
protected function shift(array $current, array $previous) : float { $shift = 0.; foreach ($current as $cluster => $centroid) { $prevCentroid = $previous[$cluster]; foreach ($centroid as $column => $mean) { $shift += abs($prevCentroid[$column] - $mean); } } return $shift; }
php
protected function shift(array $current, array $previous) : float { $shift = 0.; foreach ($current as $cluster => $centroid) { $prevCentroid = $previous[$cluster]; foreach ($centroid as $column => $mean) { $shift += abs($prevCentroid[$column] - $mean); } } return $shift; }
[ "protected", "function", "shift", "(", "array", "$", "current", ",", "array", "$", "previous", ")", ":", "float", "{", "$", "shift", "=", "0.", ";", "foreach", "(", "$", "current", "as", "$", "cluster", "=>", "$", "centroid", ")", "{", "$", "prevCentroid", "=", "$", "previous", "[", "$", "cluster", "]", ";", "foreach", "(", "$", "centroid", "as", "$", "column", "=>", "$", "mean", ")", "{", "$", "shift", "+=", "abs", "(", "$", "prevCentroid", "[", "$", "column", "]", "-", "$", "mean", ")", ";", "}", "}", "return", "$", "shift", ";", "}" ]
Calculate the magnitude (l1) of centroid shift from the previous epoch. @param array $current @param array $previous @return float
[ "Calculate", "the", "magnitude", "(", "l1", ")", "of", "centroid", "shift", "from", "the", "previous", "epoch", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Clusterers/MeanShift.php#L450-L463
RubixML/RubixML
src/AnomalyDetectors/LODA.php
LODA.train
public function train(Dataset $dataset) : void { DatasetIsCompatibleWithEstimator::check($dataset, $this); [$m, $n] = $dataset->shape(); $this->r = Matrix::gaussian($n, $this->estimators); if ($n >= self::MIN_SPARSE_DIMENSIONS) { $mask = Matrix::rand($n, $this->estimators) ->less(sqrt($n) / $n); $this->r = $this->r->multiply($mask); } $z = Matrix::quick($dataset->samples()) ->matmul($this->r) ->transpose(); foreach ($z as $values) { $min = min($values) - EPSILON; $max = max($values) + EPSILON; $edges = Vector::linspace($min, $max, $this->bins + 1)->asArray(); $edges[] = INF; $counts = array_fill(0, $this->bins + 2, 0); $interior = array_slice($edges, 1, $this->bins); foreach ($values as $value) { foreach ($interior as $k => $edge) { if ($value < $edge) { $counts[$k]++; continue 2; } } $counts[$this->bins]++; } $densities = []; foreach ($counts as $count) { $densities[] = $count > 0 ? -log($count / $m) : -LOG_EPSILON; } $this->histograms[] = [$edges, $counts, $densities]; } $this->n = $m; }
php
public function train(Dataset $dataset) : void { DatasetIsCompatibleWithEstimator::check($dataset, $this); [$m, $n] = $dataset->shape(); $this->r = Matrix::gaussian($n, $this->estimators); if ($n >= self::MIN_SPARSE_DIMENSIONS) { $mask = Matrix::rand($n, $this->estimators) ->less(sqrt($n) / $n); $this->r = $this->r->multiply($mask); } $z = Matrix::quick($dataset->samples()) ->matmul($this->r) ->transpose(); foreach ($z as $values) { $min = min($values) - EPSILON; $max = max($values) + EPSILON; $edges = Vector::linspace($min, $max, $this->bins + 1)->asArray(); $edges[] = INF; $counts = array_fill(0, $this->bins + 2, 0); $interior = array_slice($edges, 1, $this->bins); foreach ($values as $value) { foreach ($interior as $k => $edge) { if ($value < $edge) { $counts[$k]++; continue 2; } } $counts[$this->bins]++; } $densities = []; foreach ($counts as $count) { $densities[] = $count > 0 ? -log($count / $m) : -LOG_EPSILON; } $this->histograms[] = [$edges, $counts, $densities]; } $this->n = $m; }
[ "public", "function", "train", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "[", "$", "m", ",", "$", "n", "]", "=", "$", "dataset", "->", "shape", "(", ")", ";", "$", "this", "->", "r", "=", "Matrix", "::", "gaussian", "(", "$", "n", ",", "$", "this", "->", "estimators", ")", ";", "if", "(", "$", "n", ">=", "self", "::", "MIN_SPARSE_DIMENSIONS", ")", "{", "$", "mask", "=", "Matrix", "::", "rand", "(", "$", "n", ",", "$", "this", "->", "estimators", ")", "->", "less", "(", "sqrt", "(", "$", "n", ")", "/", "$", "n", ")", ";", "$", "this", "->", "r", "=", "$", "this", "->", "r", "->", "multiply", "(", "$", "mask", ")", ";", "}", "$", "z", "=", "Matrix", "::", "quick", "(", "$", "dataset", "->", "samples", "(", ")", ")", "->", "matmul", "(", "$", "this", "->", "r", ")", "->", "transpose", "(", ")", ";", "foreach", "(", "$", "z", "as", "$", "values", ")", "{", "$", "min", "=", "min", "(", "$", "values", ")", "-", "EPSILON", ";", "$", "max", "=", "max", "(", "$", "values", ")", "+", "EPSILON", ";", "$", "edges", "=", "Vector", "::", "linspace", "(", "$", "min", ",", "$", "max", ",", "$", "this", "->", "bins", "+", "1", ")", "->", "asArray", "(", ")", ";", "$", "edges", "[", "]", "=", "INF", ";", "$", "counts", "=", "array_fill", "(", "0", ",", "$", "this", "->", "bins", "+", "2", ",", "0", ")", ";", "$", "interior", "=", "array_slice", "(", "$", "edges", ",", "1", ",", "$", "this", "->", "bins", ")", ";", "foreach", "(", "$", "values", "as", "$", "value", ")", "{", "foreach", "(", "$", "interior", "as", "$", "k", "=>", "$", "edge", ")", "{", "if", "(", "$", "value", "<", "$", "edge", ")", "{", "$", "counts", "[", "$", "k", "]", "++", ";", "continue", "2", ";", "}", "}", "$", "counts", "[", "$", "this", "->", "bins", "]", "++", ";", "}", "$", "densities", "=", "[", "]", ";", "foreach", "(", "$", "counts", "as", "$", "count", ")", "{", "$", "densities", "[", "]", "=", "$", "count", ">", "0", "?", "-", "log", "(", "$", "count", "/", "$", "m", ")", ":", "-", "LOG_EPSILON", ";", "}", "$", "this", "->", "histograms", "[", "]", "=", "[", "$", "edges", ",", "$", "counts", ",", "$", "densities", "]", ";", "}", "$", "this", "->", "n", "=", "$", "m", ";", "}" ]
Train the learner with a dataset. @param \Rubix\ML\Datasets\Dataset $dataset @throws \InvalidArgumentException
[ "Train", "the", "learner", "with", "a", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/AnomalyDetectors/LODA.php#L160-L215
RubixML/RubixML
src/AnomalyDetectors/LODA.php
LODA.partial
public function partial(Dataset $dataset) : void { if (!$this->r or !$this->histograms or !$this->n) { $this->train($dataset); return; } DatasetIsCompatibleWithEstimator::check($dataset, $this); $this->n += $dataset->numRows(); $z = Matrix::quick($dataset->samples()) ->matmul($this->r) ->transpose(); foreach ($z as $i => $values) { [$edges, $counts, $densities] = $this->histograms[$i]; $interior = array_slice($edges, 1, $this->bins); foreach ($values as $value) { foreach ($interior as $k => $edge) { if ($value < $edge) { $counts[$k]++; continue 2; } } $counts[$this->bins]++; } foreach ($counts as $j => $count) { $densities[$j] = $count > 0 ? -log($count / $this->n) : -LOG_EPSILON; } $this->histograms[$i] = [$edges, $counts, $densities]; } }
php
public function partial(Dataset $dataset) : void { if (!$this->r or !$this->histograms or !$this->n) { $this->train($dataset); return; } DatasetIsCompatibleWithEstimator::check($dataset, $this); $this->n += $dataset->numRows(); $z = Matrix::quick($dataset->samples()) ->matmul($this->r) ->transpose(); foreach ($z as $i => $values) { [$edges, $counts, $densities] = $this->histograms[$i]; $interior = array_slice($edges, 1, $this->bins); foreach ($values as $value) { foreach ($interior as $k => $edge) { if ($value < $edge) { $counts[$k]++; continue 2; } } $counts[$this->bins]++; } foreach ($counts as $j => $count) { $densities[$j] = $count > 0 ? -log($count / $this->n) : -LOG_EPSILON; } $this->histograms[$i] = [$edges, $counts, $densities]; } }
[ "public", "function", "partial", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "if", "(", "!", "$", "this", "->", "r", "or", "!", "$", "this", "->", "histograms", "or", "!", "$", "this", "->", "n", ")", "{", "$", "this", "->", "train", "(", "$", "dataset", ")", ";", "return", ";", "}", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "$", "this", "->", "n", "+=", "$", "dataset", "->", "numRows", "(", ")", ";", "$", "z", "=", "Matrix", "::", "quick", "(", "$", "dataset", "->", "samples", "(", ")", ")", "->", "matmul", "(", "$", "this", "->", "r", ")", "->", "transpose", "(", ")", ";", "foreach", "(", "$", "z", "as", "$", "i", "=>", "$", "values", ")", "{", "[", "$", "edges", ",", "$", "counts", ",", "$", "densities", "]", "=", "$", "this", "->", "histograms", "[", "$", "i", "]", ";", "$", "interior", "=", "array_slice", "(", "$", "edges", ",", "1", ",", "$", "this", "->", "bins", ")", ";", "foreach", "(", "$", "values", "as", "$", "value", ")", "{", "foreach", "(", "$", "interior", "as", "$", "k", "=>", "$", "edge", ")", "{", "if", "(", "$", "value", "<", "$", "edge", ")", "{", "$", "counts", "[", "$", "k", "]", "++", ";", "continue", "2", ";", "}", "}", "$", "counts", "[", "$", "this", "->", "bins", "]", "++", ";", "}", "foreach", "(", "$", "counts", "as", "$", "j", "=>", "$", "count", ")", "{", "$", "densities", "[", "$", "j", "]", "=", "$", "count", ">", "0", "?", "-", "log", "(", "$", "count", "/", "$", "this", "->", "n", ")", ":", "-", "LOG_EPSILON", ";", "}", "$", "this", "->", "histograms", "[", "$", "i", "]", "=", "[", "$", "edges", ",", "$", "counts", ",", "$", "densities", "]", ";", "}", "}" ]
Perform a partial train on the learner. @param \Rubix\ML\Datasets\Dataset $dataset @throws \InvalidArgumentException
[ "Perform", "a", "partial", "train", "on", "the", "learner", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/AnomalyDetectors/LODA.php#L223-L264
RubixML/RubixML
src/AnomalyDetectors/LODA.php
LODA.rank
public function rank(Dataset $dataset) : array { if (!$this->r or !$this->histograms) { throw new RuntimeException('The learner has not' . ' been trained.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); $z = Matrix::quick($dataset->samples()) ->matmul($this->r) ->transpose(); return $this->logLikelihood($z); }
php
public function rank(Dataset $dataset) : array { if (!$this->r or !$this->histograms) { throw new RuntimeException('The learner has not' . ' been trained.'); } DatasetIsCompatibleWithEstimator::check($dataset, $this); $z = Matrix::quick($dataset->samples()) ->matmul($this->r) ->transpose(); return $this->logLikelihood($z); }
[ "public", "function", "rank", "(", "Dataset", "$", "dataset", ")", ":", "array", "{", "if", "(", "!", "$", "this", "->", "r", "or", "!", "$", "this", "->", "histograms", ")", "{", "throw", "new", "RuntimeException", "(", "'The learner has not'", ".", "' been trained.'", ")", ";", "}", "DatasetIsCompatibleWithEstimator", "::", "check", "(", "$", "dataset", ",", "$", "this", ")", ";", "$", "z", "=", "Matrix", "::", "quick", "(", "$", "dataset", "->", "samples", "(", ")", ")", "->", "matmul", "(", "$", "this", "->", "r", ")", "->", "transpose", "(", ")", ";", "return", "$", "this", "->", "logLikelihood", "(", "$", "z", ")", ";", "}" ]
Apply an arbitrary unnormalized scoring function over the dataset. @param \Rubix\ML\Datasets\Dataset $dataset @throws \InvalidArgumentException @throws \RuntimeException @return array
[ "Apply", "an", "arbitrary", "unnormalized", "scoring", "function", "over", "the", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/AnomalyDetectors/LODA.php#L287-L301
RubixML/RubixML
src/AnomalyDetectors/LODA.php
LODA.logLikelihood
protected function logLikelihood(Matrix $z) : array { $likelihoods = array_fill(0, $z->n(), 0.); foreach ($z as $i => $values) { [$edges, $counts, $densities] = $this->histograms[$i]; foreach ($values as $j => $value) { foreach ($edges as $k => $edge) { if ($value < $edge) { $likelihoods[$j] += $densities[$k]; break 1; } } } } foreach ($likelihoods as &$likelihood) { $likelihood /= $this->estimators; } return $likelihoods; }
php
protected function logLikelihood(Matrix $z) : array { $likelihoods = array_fill(0, $z->n(), 0.); foreach ($z as $i => $values) { [$edges, $counts, $densities] = $this->histograms[$i]; foreach ($values as $j => $value) { foreach ($edges as $k => $edge) { if ($value < $edge) { $likelihoods[$j] += $densities[$k]; break 1; } } } } foreach ($likelihoods as &$likelihood) { $likelihood /= $this->estimators; } return $likelihoods; }
[ "protected", "function", "logLikelihood", "(", "Matrix", "$", "z", ")", ":", "array", "{", "$", "likelihoods", "=", "array_fill", "(", "0", ",", "$", "z", "->", "n", "(", ")", ",", "0.", ")", ";", "foreach", "(", "$", "z", "as", "$", "i", "=>", "$", "values", ")", "{", "[", "$", "edges", ",", "$", "counts", ",", "$", "densities", "]", "=", "$", "this", "->", "histograms", "[", "$", "i", "]", ";", "foreach", "(", "$", "values", "as", "$", "j", "=>", "$", "value", ")", "{", "foreach", "(", "$", "edges", "as", "$", "k", "=>", "$", "edge", ")", "{", "if", "(", "$", "value", "<", "$", "edge", ")", "{", "$", "likelihoods", "[", "$", "j", "]", "+=", "$", "densities", "[", "$", "k", "]", ";", "break", "1", ";", "}", "}", "}", "}", "foreach", "(", "$", "likelihoods", "as", "&", "$", "likelihood", ")", "{", "$", "likelihood", "/=", "$", "this", "->", "estimators", ";", "}", "return", "$", "likelihoods", ";", "}" ]
Return the negative log likelihoods of each projection. @param \Rubix\Tensor\Matrix $z @return array
[ "Return", "the", "negative", "log", "likelihoods", "of", "each", "projection", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/AnomalyDetectors/LODA.php#L309-L332
RubixML/RubixML
src/Kernels/Distance/Euclidean.php
Euclidean.compute
public function compute(array $a, array $b) : float { $distance = 0.; foreach ($a as $i => $value) { $distance += ($value - $b[$i]) ** 2; } return sqrt($distance); }
php
public function compute(array $a, array $b) : float { $distance = 0.; foreach ($a as $i => $value) { $distance += ($value - $b[$i]) ** 2; } return sqrt($distance); }
[ "public", "function", "compute", "(", "array", "$", "a", ",", "array", "$", "b", ")", ":", "float", "{", "$", "distance", "=", "0.", ";", "foreach", "(", "$", "a", "as", "$", "i", "=>", "$", "value", ")", "{", "$", "distance", "+=", "(", "$", "value", "-", "$", "b", "[", "$", "i", "]", ")", "**", "2", ";", "}", "return", "sqrt", "(", "$", "distance", ")", ";", "}" ]
Compute the distance between two vectors. @param array $a @param array $b @return float
[ "Compute", "the", "distance", "between", "two", "vectors", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Kernels/Distance/Euclidean.php#L24-L33
RubixML/RubixML
src/Transformers/GaussianRandomProjector.php
GaussianRandomProjector.minDimensions
public static function minDimensions(int $n, float $maxDistortion = 0.1) : int { return (int) round(4. * log($n) / ($maxDistortion ** 2 / 2. - $maxDistortion ** 3 / 3.)); }
php
public static function minDimensions(int $n, float $maxDistortion = 0.1) : int { return (int) round(4. * log($n) / ($maxDistortion ** 2 / 2. - $maxDistortion ** 3 / 3.)); }
[ "public", "static", "function", "minDimensions", "(", "int", "$", "n", ",", "float", "$", "maxDistortion", "=", "0.1", ")", ":", "int", "{", "return", "(", "int", ")", "round", "(", "4.", "*", "log", "(", "$", "n", ")", "/", "(", "$", "maxDistortion", "**", "2", "/", "2.", "-", "$", "maxDistortion", "**", "3", "/", "3.", ")", ")", ";", "}" ]
Calculate the minimum number of dimensions for n total samples with a given maximum distortion using the Johnson-Lindenstrauss lemma. @param int $n @param float $maxDistortion @return int
[ "Calculate", "the", "minimum", "number", "of", "dimensions", "for", "n", "total", "samples", "with", "a", "given", "maximum", "distortion", "using", "the", "Johnson", "-", "Lindenstrauss", "lemma", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/GaussianRandomProjector.php#L49-L53
RubixML/RubixML
src/Transformers/GaussianRandomProjector.php
GaussianRandomProjector.fit
public function fit(Dataset $dataset) : void { if (!$dataset->homogeneous() or $dataset->columnType(0) !== DataType::CONTINUOUS) { throw new InvalidArgumentException('This transformer only works' . ' with continuous features.'); } $this->r = Matrix::gaussian($dataset->numColumns(), $this->dimensions); }
php
public function fit(Dataset $dataset) : void { if (!$dataset->homogeneous() or $dataset->columnType(0) !== DataType::CONTINUOUS) { throw new InvalidArgumentException('This transformer only works' . ' with continuous features.'); } $this->r = Matrix::gaussian($dataset->numColumns(), $this->dimensions); }
[ "public", "function", "fit", "(", "Dataset", "$", "dataset", ")", ":", "void", "{", "if", "(", "!", "$", "dataset", "->", "homogeneous", "(", ")", "or", "$", "dataset", "->", "columnType", "(", "0", ")", "!==", "DataType", "::", "CONTINUOUS", ")", "{", "throw", "new", "InvalidArgumentException", "(", "'This transformer only works'", ".", "' with continuous features.'", ")", ";", "}", "$", "this", "->", "r", "=", "Matrix", "::", "gaussian", "(", "$", "dataset", "->", "numColumns", "(", ")", ",", "$", "this", "->", "dimensions", ")", ";", "}" ]
Fit the transformer to the dataset. @param \Rubix\ML\Datasets\Dataset $dataset @throws \InvalidArgumentException
[ "Fit", "the", "transformer", "to", "the", "dataset", "." ]
train
https://github.com/RubixML/RubixML/blob/837385e2eb9f05c76c9522fc1673bdb87d11742d/src/Transformers/GaussianRandomProjector.php#L85-L93