Datasets:

CCRss

/

arxiv_papers_cs

like 2

on the communication complexity of xor functions

0909.3392

an xor function is a function of the form g(x,y) = f(x + y), for some booleanfunction f on n bits. we study the quantum and classical communicationcomplexity of xor functions. in the case of exact protocols, we completelycharacterise one-way communication complexity for all f. we also show that,when f is monotone, g's quantum and classical complexities are quadraticallyrelated, and that when f is a linear threshold function, g's quantum complexityis theta(n). more generally, we make a structural conjecture about the fourierspectra of boolean functions which, if true, would imply that the quantum andclassical exact communication complexities of all xor functions areasymptotically equivalent. we give two randomised classical protocols forgeneral xor functions which are efficient for certain functions, and a thirdprotocol for linear threshold functions with high margin. these protocolsoperate in the symmetric message passing model with shared randomness.

cs.cc quant-ph

nan

['montanaro', 'osborne']

https://arxiv.org/abs/0909.3392

the universal recommender

0909.3472

we describe the universal recommender, a recommender system for semanticdatasets that generalizes domain-specific recommenders such as content-based,collaborative, social, bibliographic, lexicographic, hybrid and otherrecommenders. in contrast to existing recommender systems, the universalrecommender applies to any dataset that allows a semantic representation. wedescribe the scalable three-stage architecture of the universal recommender andits application to internet protocol television (iptv). to achieve goodrecommendation accuracy, several novel machine learning and optimizationproblems are identified. we finally give a brief argument supporting the needfor machine learning recommenders.

cs.ir

nan

['kunegis', 'said', 'umbrath']

https://arxiv.org/abs/0909.3472

exploiting unlabeled data to enhance ensemble diversity

0909.3593

ensemble learning aims to improve generalization ability by using multiplebase learners. it is well-known that to construct a good ensemble, the baselearners should be accurate as well as diverse. in this paper, unlabeled datais exploited to facilitate ensemble learning by helping augment the diversityamong the base learners. specifically, a semi-supervised ensemble method namedudeed is proposed. unlike existing semi-supervised ensemble methods whereerror-prone pseudo-labels are estimated for unlabeled data to enlarge thelabeled data to improve accuracy, udeed works by maximizing accuracies of baselearners on labeled data while maximizing diversity among them on unlabeleddata. experiments show that udeed can effectively utilize unlabeled data forensemble learning and is highly competitive to well-established semi-supervisedensemble methods.

cs.lg cs.ai

nan

['zhang', 'zhou']

https://arxiv.org/abs/0909.3593

packet scheduling in a size-bounded buffer

0909.3637

we consider algorithms to schedule packets with values and deadlines in asize-bounded buffer. at any time, the buffer can store at most b packets.packets arrive over time. each packet has a non-negative value and an integerdeadline. in each time step, at most one packet can be sent. packets can bedropped at any time before they are sent. the objective is to maximize thetotal value gained by delivering packets no later than their respectivedeadlines. this model generalizes the well-studied bounded-delay model (hajek.ciss 2001. kesselman et al. stoc 2001). we first provide an optimal offlinealgorithm for this model. then we present an alternative proof of the2-competitive deterministic online algorithm (fung. arxiv july 2009). we alsoprove that the lower bound of competitive ratio of a family of (deterministicand randomized) algorithms is 2 - 1 / b.

cs.ds

nan

['li']

https://arxiv.org/abs/0909.3637

random scattering of bits by prediction

0909.3648

we investigate a population of binary mistake sequences that result fromlearning with parametric models of different order. we obtain estimates oftheir error, algorithmic complexity and divergence from a purely randombernoulli sequence. we study the relationship of these variables to thelearner's information density parameter which is defined as the ratio betweenthe lengths of the compressed to uncompressed files that contain the learner'sdecision rule. the results indicate that good learners have a low informationdensity$\rho$ while bad learners have a high $\rho$. bad learners generatemistake sequences that are atypically complex or diverge stochastically from apurely random bernoulli sequence. good learners generate typically complexsequences with low divergence from bernoulli sequences and they include mistakesequences generated by the bayes optimal predictor. based on the staticalgorithmic interference model of \cite{ratsaby_entropy} the learner here actsas a static structure which "scatters" the bits of an input sequence (to bepredicted) in proportion to its information density $\rho$ thereby deformingits randomness characteristics.

cs.ai cs.it math.it

nan

['ratsaby']

https://arxiv.org/abs/0909.3648

efficient and error-correcting data structures for membership and polynomial evaluation

0909.3696

we construct efficient data structures that are resilient against a constantfraction of adversarial noise. our model requires that the decoder answers mostqueries correctly with high probability and for the remaining queries, thedecoder with high probability either answers correctly or declares "don'tknow." furthermore, if there is no noise on the data structure, it answers allqueries correctly with high probability. our model is the common generalizationof a model proposed recently by de wolf and the notion of "relaxed locallydecodable codes" developed in the pcp literature. we measure the efficiency of a data structure in terms of its length,measured by the number of bits in its representation, and query-answering time,measured by the number of bit-probes to the (possibly corrupted)representation. in this work, we study two data structure problems: membershipand polynomial evaluation. we show that these two problems have constructionsthat are simultaneously efficient and error-correcting.

cs.ds

nan

['chen', 'grigorescu', 'de wolf']

https://arxiv.org/abs/0909.3696

maximal pivots on graphs with an application to gene assembly

0909.3789

we consider principal pivot transform (pivot) on graphs. we define a naturalvariant of this operation, called dual pivot, and show that both the kernel andthe set of maximally applicable pivots of a graph are invariant under thisoperation. the result is motivated by and applicable to the theory of geneassembly in ciliates.

math.co cs.dm

10.1016/j.dam.2010.08.030

['brijder', 'hoogeboom']

https://arxiv.org/abs/0909.3789

on carpi and alessandro conjecture

0909.3790

the well known open \v{c}ern\'y conjecture states that each \san with $n$states has a \sw of length at most $(n-1)^2$. on the other hand, the best knownupper bound is cubic of $n$. recently, in the paper \cite{carpi1} of alessandroand carpi, the authors introduced the new notion of strongly transitivity forautomata and conjectured that this property with a help of \emph{extension}method allows to get a quadratic upper bound for the length of the shortest\sws. they also confirmed this conjecture for circular automata. we disprovethis conjecture and the long-standing \emph{extension} conjecture too. we alsoconsider the widely used extension method and its perspectives.

cs.fl

nan

['berlinkov']

https://arxiv.org/abs/0909.3790

method of resolution of 3sat in polynomial time

0909.3868

presentation of a method for determining whether a problem 3sat has solution,and if yes to find one, in time max o(n^15). is thus proved that the problem3sat is fully resolved in polynomial time and therefore that it is in p, by thework of cook and levin, and can transform a sat problem in a 3sat in polynomialtime (ref. karp), it follows that p = np. open source program is available athttp://www.visainformatica.it/3sat

cs.cc

nan

['salemi']

https://arxiv.org/abs/0909.3868

automatic modular abstractions for template numerical constraints

0909.4013

we propose a method for automatically generating abstract transformers forstatic analysis by abstract interpretation. the method focuses on linearconstraints on programs operating on rational, real or floating-point variablesand containing linear assignments and tests. in addition to loop-free code, thesame method also applies for obtaining least fixed points as functions of theprecondition, which permits the analysis of loops and recursive functions. ouralgorithms are based on new quantifier elimination and symbolic manipulationtechniques. given the specification of an abstract domain, and a program block,our method automatically outputs an implementation of the correspondingabstract transformer. it is thus a form of program transformation. themotivation of our work is data-flow synchronous programming languages, used forbuilding control-command embedded systems, but it also applies to imperativeand functional programming.

cs.pl cs.lo

nan

['monniaux']

https://arxiv.org/abs/0909.4013

long non-crossing configurations in the plane

0909.4094

we revisit several maximization problems for geometric networks design underthe non-crossing constraint, first studied by alon, rajagopalan and suri (acmsymposium on computational geometry, 1993). given a set of $n$ points in theplane in general position (no three points collinear), compute a longestnon-crossing configuration composed of straight line segments that is: (a) amatching (b) a hamiltonian path (c) a spanning tree. here we obtain new resultsfor (b) and (c), as well as for the hamiltonian cycle problem: (i) for the longest non-crossing hamiltonian path problem, we give anapproximation algorithm with ratio $\frac{2}{\pi+1} \approx 0.4829$. theprevious best ratio, due to alon et al., was $1/\pi \approx 0.3183$. moreover,the ratio of our algorithm is close to $2/\pi$ on a relatively broad class ofinstances: for point sets whose perimeter (or diameter) is much shorter thanthe maximum length matching. the algorithm runs in $o(n^{7/3}\log{n})$ time. (ii) for the longest non-crossing spanning tree problem, we give anapproximation algorithm with ratio 0.502 which runs in $o(n \log{n})$ time. theprevious ratio, 1/2, due to alon et al., was achieved by a quadratic timealgorithm. along the way, we first re-derive the result of alon et al. with afaster $o(n \log{n})$-time algorithm and a very simple analysis. (iii) for the longest non-crossing hamiltonian cycle problem, we give anapproximation algorithm whose ratio is close to $2/\pi$ on a relatively broadclass of instances: for point sets with the product $\bf{<}$diameter$\times$convex hull size $\bf{>}$ much smaller than the maximum length matching. thealgorithm runs in $o(n^{7/3}\log{n})$ time. no previous approximation resultswere known for this problem.

cs.cg

nan

['dumitrescu', 'tóth']

https://arxiv.org/abs/0909.4094

a numerical algorithm for zero counting. ii: distance to ill-posedness and smoothed analysis

0909.4101

we show a condition number theorem for the condition number of zero countingfor real polynomial systems. that is, we show that this condition number equalsthe inverse of the normalized distance to the set of ill-posed systems (i.e.,those having multiple real zeros). as a consequence, a smoothed analysis ofthis condition number follows.

cs.na

10.1007/s11784-009-0127-4

['cucker', 'krick', 'malajovich', 'wschebor']

https://arxiv.org/abs/0909.4101

discrete mdl predicts in total variation

0909.4588

the minimum description length (mdl) principle selects the model that has theshortest code for data plus model. we show that for a countable class ofmodels, mdl predictions are close to the true distribution in a strong sense.the result is completely general. no independence, ergodicity, stationarity,identifiability, or other assumption on the model class need to be made. moreformally, we show that for any countable class of models, the distributionsselected by mdl (or map) asymptotically predict (merge with) the true measurein the class in total variation distance. implications for non-i.i.d. domainslike time-series forecasting, discriminative learning, and reinforcementlearning are discussed.

math.pr cs.it cs.lg math.it math.st stat.ml stat.th

nan

['hutter']

https://arxiv.org/abs/0909.4588

optimal feedback communication via posterior matching

0909.4828

in this paper we introduce a fundamental principle for optimal communicationover general memoryless channels in the presence of noiseless feedback, termedposterior matching. using this principle, we devise a (simple, sequential)generic feedback transmission scheme suitable for a large class of memorylesschannels and input distributions, achieving any rate below the correspondingmutual information. this provides a unified framework for optimal feedbackcommunication in which the horstein scheme (bsc) and the schalkwijk-kailathscheme (awgn channel) are special cases. thus, as a corollary, we prove thatthe horstein scheme indeed attains the bsc capacity, settling a longstandingconjecture. we further provide closed form expressions for the errorprobability of the scheme over a range of rates, and derive the achievablerates in a mismatch setting where the scheme is designed according to the wrongchannel model. several illustrative examples of the posterior matching schemefor specific channels are given, and the corresponding error probabilityexpressions are evaluated. the proof techniques employed utilize novelrelations between information rates and contraction properties of iteratedfunction systems.

cs.it math.it

nan

['shayevitz', 'feder']

https://arxiv.org/abs/0909.4828

range non-overlapping indexing

0909.4893

we study the non-overlapping indexing problem: given a text t, preprocess itso that you can answer queries of the form: given a pattern p, report themaximal set of non-overlapping occurrences of p in t. a generalization of thisproblem is the range non-overlapping indexing where in addition we are giventwo indexes i,j to report the maximal set of non-overlapping occurrencesbetween these two indexes. we suggest new solutions for these problems. for thenon-overlapping problem our solution uses o(n) space with query time of o(m +occ_{no}). for the range non-overlapping problem we propose a solution witho(n\log^\epsilon n) space for some 0<\epsilon<1 and o(m + \log\log n +occ_{ij,no}) query time.

cs.ds

nan

['cohen', 'porat']

https://arxiv.org/abs/0909.4893

fast set intersection and two patterns matching

0909.5146

in this paper we present a new problem, the fast set intersection problem,which is to preprocess a collection of sets in order to efficiently report theintersection of any two sets in the collection. in addition we suggest newsolutions for the two-dimensional substring indexing problem and the documentlisting problem for two patterns by reduction to the fast set intersectionproblem.

cs.ds

nan

['cohen', 'porat']

https://arxiv.org/abs/0909.5146

learning gaussian tree models: analysis of error exponents and extremal structures

0909.5216

the problem of learning tree-structured gaussian graphical models fromindependent and identically distributed (i.i.d.) samples is considered. theinfluence of the tree structure and the parameters of the gaussian distributionon the learning rate as the number of samples increases is discussed.specifically, the error exponent corresponding to the event that the estimatedtree structure differs from the actual unknown tree structure of thedistribution is analyzed. finding the error exponent reduces to a least-squaresproblem in the very noisy learning regime. in this regime, it is shown that theextremal tree structure that minimizes the error exponent is the star for anyfixed set of correlation coefficients on the edges of the tree. if themagnitudes of all the correlation coefficients are less than 0.63, it is alsoshown that the tree structure that maximizes the error exponent is the markovchain. in other words, the star and the chain graphs represent the hardest andthe easiest structures to learn in the class of tree-structured gaussiangraphical models. this result can also be intuitively explained by correlationdecay: pairs of nodes which are far apart, in terms of graph distance, areunlikely to be mistaken as edges by the maximum-likelihood estimator in theasymptotic regime.

stat.ml cs.it math.it math.st stat.th

10.1109/tsp.2010.2042478

['tan', 'anandkumar', 'willsky']

https://arxiv.org/abs/0909.5216

practical rate and route adaptation with efficient link quality estimation for ieee 802.11b/g multi-hop networks

0909.5263

accurate and fast packet delivery rate (pdr) estimation, used in evaluatingwireless link quality, is a prerequisite to increase the performance of mobile,multi-hop and multi-rate wireless ad hoc networks. unfortunately, contemporarypdr estimation methods, i.e. beacon-based packet counting in estimatedtransmission time and expected transmission count metrics, have unsatisfactoryperformance. therefore, in this paper we propose a novel pdr estimation methodbased on snr profiles. we classify all possible link quality estimation methodsand compare them analytically against our design. results show that it leads toa more efficient link quality estimation. further investigations with theprototype implementation of our method in ieee 802.11b/g testbeds reveal thatthe accuracy of pdr estimation in mobile scenarios can be improved up to 50% incomparison to generic packet-based pdr. experiments with the same prototype onlink and routing layers for different measurement scenarios show that it leadsto a better rate adaptation and route selection in the form of end-to-endthroughput increase compared to traditional packet counting methods.

cs.ni

nan

['zhou', 'rao', 'pawełczak', 'wu', 'mohapatra']

https://arxiv.org/abs/0909.5263

a simple necessary and sufficient condition for the double unicast problem

0909.5268

we consider a directed acyclic network where there are two source-terminalpairs and the terminals need to receive the symbols generated at the respectivesources. each source independently generates an i.i.d. random process over thesame alphabet. each edge in the network is error-free, delay-free, and cancarry one symbol from the alphabet per use. we give a simple necessary andsufficient condition for being able to simultaneously satisfy the unicastrequirements of the two source-terminal pairs at rate-pair $(1,1)$ using vectornetwork coding. the condition is also sufficient for doing this using only"xor" network coding and is much simpler compared to the necessary andsufficient conditions known from previous work. our condition also yields asimple characterization of the capacity region of a double-unicast networkwhich does not support the rate-pair $(1,1)$.

cs.it math.it

nan

['shenvi', 'dey']

https://arxiv.org/abs/0909.5268

the remote point problem, small bias space, and expanding generator sets

0909.5313

using $\epsilon$-bias spaces over $f_2$, we show that the remote pointproblem (rpp), introduced by alon et al [apy09], has an $nc^2$ algorithm(achieving the same parameters as [apy09]). we study a generalization of theremote point problem to groups: we replace $f^n$ by $g^n$ for an arbitraryfixed group $g$. when $g$ is abelian, we give an $nc^2$ algorithm for rpp,again using $\epsilon$-bias spaces. for nonabelian $g$, we give a deterministicpolynomial-time algorithm for rpp. we also show the connection to constructionof expanding generator sets for the group $g^n$. all our algorithms for the rppachieve essentially the same parameters as [apy09].

cs.cc cs.ds

nan

['arvind', 'srinivasan']

https://arxiv.org/abs/0909.5313

iterative shrinkage approach to restoration of optical imagery

0909.5460

the problem of reconstruction of digital images from their degradedmeasurements is regarded as a problem of central importance in various fieldsof engineering and imaging sciences. in such cases, the degradation istypically caused by the resolution limitations of an imaging device in useand/or by the destructive influence of measurement noise. specifically, whenthe noise obeys a poisson probability law, standard approaches to the problemof image reconstruction are based on using fixed-point algorithms which followthe methodology first proposed by richardson and lucy. the practice of usingthese methods, however, shows that their convergence properties tend todeteriorate at relatively high noise levels. accordingly, in the present paper,a novel method for de-noising and/or de-blurring of digital images corrupted bypoisson noise is introduced. the proposed method is derived under theassumption that the image of interest can be sparsely represented in the domainof a linear transform. consequently, a shrinkage-based iterative procedure isproposed, which guarantees the solution to converge to the global maximizer ofan associated maximum-a-posteriori criterion. it is shown in a series of bothcomputer-simulated and real-life experiments that the proposed methodoutperforms a number of existing alternatives in terms of stability, precision,and computational efficiency.

cs.cv

nan

['shaked', 'michailovich']

https://arxiv.org/abs/0909.5460

two-phase flow complexity in heterogeneous media

0909.5583

in this study, we investigate the appeared complexity of two-phase flow(air/water) in a heterogeneous soil where the supposed porous media isnon-deformable media which is under the timedependent gas pressure. afterobtaining of governing equations and considering the capillarypressuresaturation and permeability functions, the evolution of the modelunknown parameters were obtained. in this way, using comsol (femlab) and fluidflow/script module, the role of heterogeneity in intrinsic permeability wasanalysed. also, the evolution of relative permeability of wetting andnon-wetting fluid, capillary pressure and other parameters were elicited. inthe last part, a complex network approach to analysis of emerged patterns willbe employed.

cs.ce physics.flu-dyn

nan

['ghaffari']

https://arxiv.org/abs/0909.5583

quantum interpolation of polynomials

0909.5683

we consider quantum interpolation of polynomials. we imagine a quantumcomputer with black-box access to input/output pairs (x_i, f(x_i)), where f isa degree-d polynomial, and we wish to compute f(0). we give asymptoticallytight quantum lower bounds for this problem, even in the case where 0 is amongthe possible values of x_i.

quant-ph cs.cc

nan

['kane', 'kutin']

https://arxiv.org/abs/0909.5683

finding associations and computing similarity via biased pair sampling

0910.0112

this version is ***superseded*** by a full version that can be found athttp://www.itu.dk/people/pagh/papers/mining-jour.pdf, which contains strongertheoretical results and fixes a mistake in the reporting of experiments. abstract: sampling-based methods have previously been proposed for theproblem of finding interesting associations in data, even for low-supportitems. while these methods do not guarantee precise results, they can be vastlymore efficient than approaches that rely on exact counting. however, for manysimilarity measures no such methods have been known. in this paper we show howa wide variety of measures can be supported by a simple biased sampling method.the method also extends to find high-confidence association rules. wedemonstrate theoretically that our method is superior to exact methods when thethreshold for "interesting similarity/confidence" is above the average pairwisesimilarity/confidence, and the average support is not too low. our method isparticularly good when transactions contain many items. we confirm inexperiments on standard association mining benchmarks that this gives asignificant speedup on real data sets (sometimes much larger than thetheoretical guarantees). reductions in computation time of over an order ofmagnitude, and significant savings in space, are observed.

cs.ds cs.db cs.lg

nan

['campagna', 'pagh']

https://arxiv.org/abs/0910.0112

linear rank inequalities on five or more variables

0910.0284

ranks of subspaces of vector spaces satisfy all linear inequalities satisfiedby entropies (including the standard shannon inequalities) and an additionalinequality due to ingleton. it is known that the shannon and ingletoninequalities generate all such linear rank inequalities on up to fourvariables, but it has been an open question whether additional inequalitieshold for the case of five or more variables. here we give a list of 24inequalities which, together with the shannon and ingleton inequalities,generate all linear rank inequalities on five variables. we also give a partiallist of linear rank inequalities on six variables and general results whichproduce such inequalities on an arbitrary number of variables; we prove thatthere are essentially new inequalities at each number of variables beyond four(a result also proved recently by kinser).

cs.it math.it

nan

['dougherty', 'freiling', 'zeger']

https://arxiv.org/abs/0910.0284

the averaging trick and the cerny conjecture

0910.0410

the results of several papers concerning the \v{c}ern\'y conjecture arededuced as consequences of a simple idea that i call the averaging trick. thisidea is implicitly used in the literature, but no attempt was made to formalizethe proof scheme axiomatically. instead, authors axiomatized classes ofautomata to which it applies.

cs.fl

nan

['steinberg']

https://arxiv.org/abs/0910.0410

exact covers via determinants

0910.0460

given a k-uniform hypergraph on n vertices, partitioned in k equal parts suchthat every hyperedge includes one vertex from each part, the k-dimensionalmatching problem asks whether there is a disjoint collection of the hyperedgeswhich covers all vertices. we show it can be solved by a randomized polynomialspace algorithm in time o*(2^(n(k-2)/k)). the o*() notation hides factorspolynomial in n and k. when we drop the partition constraint and permit arbitrary hyperedges ofcardinality k, we obtain the exact cover by k-sets problem. we show it can besolved by a randomized polynomial space algorithm in time o*(c_k^n), wherec_3=1.496, c_4=1.642, c_5=1.721, and provide a general bound for larger k. both results substantially improve on the previous best algorithms for theseproblems, especially for small k, and follow from the new observation thatlovasz' perfect matching detection via determinants (1979) admits an embeddingin the recently proposed inclusion-exclusion counting scheme for set covers,despite its inability to count the perfect matchings.

cs.ds

nan

['björklund']

https://arxiv.org/abs/0910.0460

regularization techniques for learning with matrices

0910.0610

there is growing body of learning problems for which it is natural toorganize the parameters into matrix, so as to appropriately regularize theparameters under some matrix norm (in order to impose some more sophisticatedprior knowledge). this work describes and analyzes a systematic method forconstructing such matrix-based, regularization methods. in particular, we focuson how the underlying statistical properties of a given problem can help usdecide which regularization function is appropriate. our methodology is based on the known duality fact: that a function isstrongly convex with respect to some norm if and only if its conjugate functionis strongly smooth with respect to the dual norm. this result has already beenfound to be a key component in deriving and analyzing several learningalgorithms. we demonstrate the potential of this framework by deriving novelgeneralization and regret bounds for multi-task learning, multi-class learning,and kernel learning.

cs.lg stat.ml

nan

['kakade', 'shalev-shwartz', 'tewari']

https://arxiv.org/abs/0910.0610

optimal testing of reed-muller codes

0910.0641

we consider the problem of testing if a given function f : f_2^n -> f_2 isclose to any degree d polynomial in n variables, also known as the reed-mullertesting problem. the gowers norm is based on a natural 2^{d+1}-query test forthis property. alon et al. [akklr05] rediscovered this test and showed that itaccepts every degree d polynomial with probability 1, while it rejectsfunctions that are omega(1)-far with probability omega(1/(d 2^{d})). we give anasymptotically optimal analysis of this test, and show that it rejectsfunctions that are (even only) omega(2^{-d})-far with omega(1)-probability (sothe rejection probability is a universal constant independent of d and n). thisimplies a tight relationship between the (d+1)st gowers norm of a function andits maximal correlation with degree d polynomials, when the correlation isclose to 1. our proof works by induction on n and yields a new analysis of eventhe classical blum-luby-rubinfeld [blr93] linearity test, for the setting offunctions mapping f_2^n to f_2. the optimality follows from a tighter analysisof counterexamples to the "inverse conjecture for the gowers norm" constructedby [gt09,lms08]. our result has several implications. first, it shows that thegowers norm test is tolerant, in that it also accepts close codewords. second,it improves the parameters of an xor lemma for polynomials given by viola andwigderson [vw07]. third, it implies a "query hierarchy" result for propertytesting of affine-invariant properties. that is, for every function q(n), itgives an affine-invariant property that is testable with o(q(n))-queries, butnot with o(q(n))-queries, complementing an analogous result of [gknr09] forgraph properties.

math.co cs.cc cs.it math.it

nan

['bhattacharyya', 'kopparty', 'schoenebeck', 'sudan', 'zuckerman']

https://arxiv.org/abs/0910.0641

transmission line inspires a new distributed algorithm to solve linear system of circuit

0910.0663

transmission line, or wire, is always troublesome to integrated circuitsdesigners, but it could be helpful to parallel computing researchers. thispaper proposes the virtual transmission method (vtm), which is a newdistributed and stationary iterative algorithm to solve the linear systemextracted from circuit. it tears the circuit by virtual transmission lines toachieve distributed computing. for the symmetric positive definite (spd) linearsystem, vtm is proved to be convergent. for the unsymmetrical linear system,numerical experiments show that vtm is possible to achieve better convergenceproperty than the traditional stationary algorithms. vtm could be acceleratedby some preconditioning techniques, and the convergence speed of vtm is fastwhen its preconditioner is properly chosen.

cs.ce cs.dc cs.ms cs.na

nan

['wei', 'yang']

https://arxiv.org/abs/0910.0663

cracking the liu key exchange protocol in its most secure state with lorentzian spectra

0910.0665

we have found a security risk in the liu's cypher based on random signals andfeedback, when it utilizes a large class of noises for communication in itsmost secure state, the steady state. for the vulnerability to exist, the noisemust have a spectrum which can be transformed to white-like noise by linearfiltering. for the cracking, we utilize the natural properties of power densityspectra and autocorrelation functions. we introduce and demonstrate the methodfor lorentzian spectra. some of the implications of the results concern thetransient operation during changing bits, where the modulation products ofnoise cannot be band-limited therefore the cypher is vulnerable. we propose theapplication of line filters to provide a proper spectral shape and to improvethe security.

physics.data-an cs.cr quant-ph

10.1142/s0219477510000058

['kish', 'zhang', 'kish']

https://arxiv.org/abs/0910.0665

variable sigma gaussian processes: an expectation propagation perspective

0910.0668

gaussian processes (gps) provide a probabilistic nonparametric representationof functions in regression, classification, and other problems. unfortunately,exact learning with gps is intractable for large datasets. a variety ofapproximate gp methods have been proposed that essentially map the largedataset into a small set of basis points. the most advanced of these, thevariable-sigma gp (vsgp) (walder et al., 2008), allows each basis point to haveits own length scale. however, vsgp was only derived for regression. wedescribe how vsgp can be applied to classification and other problems, byderiving it as an expectation propagation algorithm. in this view, sparse gpapproximations correspond to a kl-projection of the true posterior onto acompact exponential family of gps. vsgp constitutes one such family, and weshow how to enlarge this family to get additional accuracy. in particular, weshow that endowing each basis point with its own full covariance matrixprovides a significant increase in approximation power.

cs.lg

nan

['qi', 'abdel-gawad', 'minka']

https://arxiv.org/abs/0910.0668

statistics on graphs, exponential formula and combinatorial physics

0910.0695

the concern of this paper is a famous combinatorial formula known under thename "exponential formula". it occurs quite naturally in many contexts(physics, mathematics, computer science). roughly speaking, it expresses thatthe exponential generating function of a whole structure is equal to theexponential of those of connected substructures. keeping this descriptivestatement as a guideline, we develop a general framework to handle manydifferent situations in which the exponential formula can be applied.

cs.dm cs.ce math.co quant-ph

nan

['poinsot', 'duchamp', 'goodenough', 'penson']

https://arxiv.org/abs/0910.0695

performance of statistical tests for single source detection using random matrix theory

0910.0827

this paper introduces a unified framework for the detection of a source witha sensor array in the context where the noise variance and the channel betweenthe source and the sensors are unknown at the receiver. the generalized maximumlikelihood test is studied and yields the analysis of the ratio between themaximum eigenvalue of the sampled covariance matrix and its normalized trace.using recent results of random matrix theory, a practical way to evaluate thethreshold and the $p$-value of the test is provided in the asymptotic regimewhere the number $k$ of sensors and the number $n$ of observations per sensorare large but have the same order of magnitude. the theoretical performance ofthe test is then analyzed in terms of receiver operating characteristic (roc)curve. it is in particular proved that both type i and type ii errorprobabilities converge to zero exponentially as the dimensions increase at thesame rate, and closed-form expressions are provided for the error exponents.these theoretical results rely on a precise description of the large deviationsof the largest eigenvalue of spiked random matrix models, and establish thatthe presented test asymptotically outperforms the popular test based on thecondition number of the sampled covariance matrix.

math.pr cs.it math.it math.st stat.th

nan

['bianchi', 'debbah', 'maïda', 'najim']

https://arxiv.org/abs/0910.0827

use of l-system mathematics for making new subfamily members of olfactory receptor full length genes, or1d2, or1d4 and or1d5

0910.1416

ligands for only two human olfactory receptors are known. one of them, or1d2,binds to bourgeonal [malnic b, godfrey p-a, buck l-b (2004) the human olfactoryreceptor gene family. proc. natl. acad. sci u. s. a. 101: 2584-2589 and erratumin: proc natl acad sci u. s. a. (2004) 101: 7205]. or1d2, or1d4 and or1d5 arethree full length olfactory receptors present in an olfactory locus in humangenome. these receptors are more than 80% identical in dna sequences and have108 base pair mismatches among them. we have used l-system mathematics and havebeen able to show a closely related subfamily of or1d2, or1d4 and or1d5.

cs.oh cs.fl

nan

['hassan', 'choudhury', 'pal', 'brahmachary', 'goswami']

https://arxiv.org/abs/0910.1416

distributed computing with adaptive heuristics

0910.1585

we use ideas from distributed computing to study dynamic environments inwhich computational nodes, or decision makers, follow adaptive heuristics (hart2005), i.e., simple and unsophisticated rules of behavior, e.g., repeatedly"best replying" to others' actions, and minimizing "regret", that have beenextensively studied in game theory and economics. we explore when convergenceof such simple dynamics to an equilibrium is guaranteed in asynchronouscomputational environments, where nodes can act at any time. our researchagenda, distributed computing with adaptive heuristics, lies on the borderlineof computer science (including distributed computing and learning) and gametheory (including game dynamics and adaptive heuristics). we exhibit a generalnon-termination result for a broad class of heuristics with boundedrecall---that is, simple rules of behavior that depend only on recent historyof interaction between nodes. we consider implications of our result across awide variety of interesting and timely applications: game theory, circuitdesign, social networks, routing and congestion control. we also study thecomputational and communication complexity of asynchronous dynamics and presentsome basic observations regarding the effects of asynchrony on no-regretdynamics. we believe that our work opens a new avenue for research in bothdistributed computing and game theory.

cs.dc cs.gt

nan

['jaggard', 'schapira', 'wright']

https://arxiv.org/abs/0910.1585

covering points by disjoint boxes with outliers

0910.1643

for a set of n points in the plane, we consider the axis--aligned (p,k)-boxcovering problem: find p axis-aligned, pairwise-disjoint boxes that togethercontain n-k points. in this paper, we consider the boxes to be either squaresor rectangles, and we want to minimize the area of the largest box. for generalp we show that the problem is np-hard for both squares and rectangles. for asmall, fixed number p, we give algorithms that find the solution in thefollowing running times: for squares we have o(n+k log k) time for p=1, and o(n log n+k^p log^p k timefor p = 2,3. for rectangles we get o(n + k^3) for p = 1 and o(n log n+k^{2+p}log^{p-1} k) time for p = 2,3. in all cases, our algorithms use o(n) space.

cs.cg cs.ds

nan

['ahn', 'bae', 'demaine', 'demaine', 'kim', 'korman', 'reinbacher', 'son']

https://arxiv.org/abs/0910.1643

balancing egoism and altruism on mimo interference channel

0910.1688

this paper considers the so-called multiple-input-multiple-outputinterference channel (mimo-ic) which has relevance in applications such asmulti-cell coordination in cellular networks as well as spectrum sharing incognitive radio networks among others. we consider a beamforming designframework based on striking a compromise between beamforming gain at theintended receiver (egoism) and the mitigation of interference created towardsother receivers (altruism). combining egoistic and altruistic beamforming hasbeen shown previously in several papers to be instrumental to optimizing therates in a multiple-input-single-output interference channel miso-ic (i.e.where receivers have no interference canceling capability). here, by using theframework of bayesian games, we shed more light on these game-theoreticconcepts in the more general context of mimo channels and more particularlywhen coordinating parties only have channel state information (csi) of channelsthat they can measure directly. this allows us to derive distributedbeamforming techniques. we draw parallels with existing work on the mimo-ic,including rate-optimizing and interference-alignment precoding techniques,showing how such techniques may be improved or re-interpreted through a commonprism based on balancing egoistic and altruistic beamforming. our analysis andsimulations currently limited to single stream transmission per user attest theimprovements over known interference alignment based methods in terms of sumrate performance in the case of so-called asymmetric networks.

cs.it math.it

nan

['ho', 'gesbert']

https://arxiv.org/abs/0910.1688

sensitive white space detection with spectral covariance sensing

0910.1787

this paper proposes a novel, highly effective spectrum sensing algorithm forcognitive radio and whitespace applications. the proposed spectral covariancesensing (scs) algorithm exploits the different statistical correlations of thereceived signal and noise in the frequency domain. test statistics are computedfrom the covariance matrix of a partial spectrogram and compared with adecision threshold to determine whether a primary signal or arbitrary type ispresent or not. this detector is analyzed theoretically and verified throughrealistic open-source simulations using actual digital television signalscaptured in the us. compared to the state of the art in the literature, scsimproves sensitivity by 3 db for the same dwell time, which is a verysignificant improvement for this application. further, it is shown that scs ishighly robust to noise uncertainty, whereas many other spectrum sensors arenot.

cs.ni

nan

['kim', 'andrews']

https://arxiv.org/abs/0910.1787

engineering a scalable high quality graph partitioner

0910.2004

we describe an approach to parallel graph partitioning that scales tohundreds of processors and produces a high solution quality. for example, formany instances from walshaw's benchmark collection we improve the best knownpartitioning. we use the well known framework of multi-level graphpartitioning. all components are implemented by scalable parallel algorithms.quality improvements compared to previous systems are due to betterprioritization of edges to be contracted, better approximation algorithms foridentifying matchings, better local search heuristics, and perhaps mostnotably, a parallelization of the fm local search algorithm that works morelocally than previous approaches.

cs.dc cs.ds

nan

['holtgrewe', 'sanders', 'schulz']

https://arxiv.org/abs/0910.2004

strategies for online inference of model-based clustering in large and growing networks

0910.2034

in this paper we adapt online estimation strategies to perform model-basedclustering on large networks. our work focuses on two algorithms, the firstbased on the saem algorithm, and the second on variational methods. these twostrategies are compared with existing approaches on simulated and real data. weuse the method to decipher the connexion structure of the political websphereduring the us political campaign in 2008. we show that our online em-basedalgorithms offer a good trade-off between precision and speed, when estimatingparameters for mixture distributions in the context of random graphs.

stat.ap cs.lg

10.1214/10-aoas359

['zanghi', 'picard', 'miele', 'ambroise']

https://arxiv.org/abs/0910.2034

higher coordination with less control - a result of information maximization in the sensorimotor loop

0910.2039

this work presents a novel learning method in the context of embodiedartificial intelligence and self-organization, which has as few assumptions andrestrictions as possible about the world and the underlying model. the learningrule is derived from the principle of maximizing the predictive information inthe sensorimotor loop. it is evaluated on robot chains of varying length withindividually controlled, non-communicating segments. the comparison of theresults shows that maximizing the predictive information per wheel leads to ahigher coordinated behavior of the physically connected robots compared to amaximization per robot. another focus of this paper is the analysis of theeffect of the robot chain length on the overall behavior of the robots. it willbe shown that longer chains with less capable controllers outperform those ofshorter length and more complex controllers. the reason is found and discussedin the information-geometric interpretation of the learning process.

cs.ai cs.it cs.ro math.it

nan

['zahedi', 'ay', 'der']

https://arxiv.org/abs/0910.2039

minimax rates of estimation for high-dimensional linear regression over $\ell_q$-balls

0910.2042

consider the standard linear regression model $\y = \xmat \betastar + w$,where $\y \in \real^\numobs$ is an observation vector, $\xmat \in\real^{\numobs \times \pdim}$ is a design matrix, $\betastar \in \real^\pdim$is the unknown regression vector, and $w \sim \mathcal{n}(0, \sigma^2 i)$ isadditive gaussian noise. this paper studies the minimax rates of convergencefor estimation of $\betastar$ for $\ell_\rpar$-losses and in the$\ell_2$-prediction loss, assuming that $\betastar$ belongs to an$\ell_{\qpar}$-ball $\ballq(\myrad)$ for some $\qpar \in [0,1]$. we show thatunder suitable regularity conditions on the design matrix $\xmat$, the minimaxerror in $\ell_2$-loss and $\ell_2$-prediction loss scales as $\rq\big(\frac{\log \pdim}{n}\big)^{1-\frac{\qpar}{2}}$. in addition, we providelower bounds on minimax risks in $\ell_{\rpar}$-norms, for all $\rpar \in [1,+\infty], \rpar \neq \qpar$. our proofs of the lower bounds areinformation-theoretic in nature, based on fano's inequality and results on themetric entropy of the balls $\ballq(\myrad)$, whereas our proofs of the upperbounds are direct and constructive, involving direct analysis of least-squaresover $\ell_{\qpar}$-balls. for the special case $q = 0$, a comparison with$\ell_2$-risks achieved by computationally efficient $\ell_1$-relaxationsreveals that although such methods can achieve the minimax rates up to constantfactors, they require slightly stronger assumptions on the design matrix$\xmat$ than algorithms involving least-squares over the $\ell_0$-ball.

math.st cs.it math.it stat.th

nan

['raskutti', 'wainwright', 'yu']

https://arxiv.org/abs/0910.2042

on product, generic and random generic quantum satisfiability

0910.2058

we report a cluster of results on k-qsat, the problem of quantumsatisfiability for k-qubit projectors which generalizes classicalsatisfiability with k-bit clauses to the quantum setting. first we define thenp-complete problem of product satisfiability and give a geometrical criterionfor deciding when a qsat interaction graph is product satisfiable with positiveprobability. we show that the same criterion suffices to establish quantumsatisfiability for all projectors. second, we apply these results to the randomgraph ensemble with generic projectors and obtain improved lower bounds on thelocation of the sat--unsat transition. third, we present numerical results onrandom, generic satisfiability which provide estimates for the location of thetransition for k=3 and k=4 and mild evidence for the existence of a phase whichis satisfiable by entangled states alone.

quant-ph cond-mat.stat-mech cs.cc

10.1103/physreva.81.062345

['laumann', 'läuchli', 'moessner', 'scardicchio', 'sondhi']

https://arxiv.org/abs/0910.2058

cooperative multi-cell block diagonalization with per-base-station power constraints

0910.2304

block diagonalization (bd) is a practical linear precoding technique thateliminates the inter-user interference in downlink multiuser multiple-inputmultiple-output (mimo) systems. in this paper, we apply bd to the downlinktransmission in a cooperative multi-cell mimo system, where the signals fromdifferent base stations (bss) to all the mobile stations (mss) are jointlydesigned with the perfect knowledge of the downlink channels and transmitmessages. specifically, we study the optimal bd precoder design to maximize theweighted sum-rate of all the mss subject to a set of per-bs power constraints.this design problem is formulated in an auxiliary mimo broadcast channel (bc)with a set of transmit power constraints corresponding to those for individualbss in the multi-cell system. by applying convex optimization techniques, thispaper develops an efficient algorithm to solve this problem, and derives theclosed-form expression for the optimal bd precoding matrix. it is revealed thatthe optimal bd precoding vectors for each ms in the per-bs power constraintcase are in general non-orthogonal, which differs from the conventionalorthogonal bd precoder design for the mimo-bc under one single sum-powerconstraint. moreover, for the special case of single-antenna bss and mss, theproposed solution reduces to the optimal zero-forcing beamforming (zf-bf)precoder design for the weighted sum-rate maximization in the multiple-inputsingle-output (miso) bc with per-antenna power constraints. suboptimal andlow-complexity bd/zf-bf precoding schemes are also presented, and theirachievable rates are compared against those with the optimal schemes.

cs.it math.it

nan

['zhang']

https://arxiv.org/abs/0910.2304

a component based heuristic search method with evolutionary eliminations

0910.2593

nurse rostering is a complex scheduling problem that affects hospitalpersonnel on a daily basis all over the world. this paper presents a newcomponent-based approach with evolutionary eliminations, for a nurse schedulingproblem arising at a major uk hospital. the main idea behind this technique isto decompose a schedule into its components (i.e. the allocated shift patternof each nurse), and then to implement two evolutionary elimination strategiesmimicking natural selection and natural mutation process on these componentsrespectively to iteratively deliver better schedules. the worthiness of allcomponents in the schedule has to be continuously demonstrated in order forthem to remain there. this demonstration employs an evaluation function whichevaluates how well each component contributes towards the final objective. twoelimination steps are then applied: the first elimination eliminates a numberof components that are deemed not worthy to stay in the current schedule; thesecond elimination may also throw out, with a low level of probability, someworthy components. the eliminated components are replenished with new onesusing a set of constructive heuristics using local optimality criteria.computational results using 52 data instances demonstrate the applicability ofthe proposed approach in solving real-world problems.

cs.ai cs.ne

nan

['li', 'aickelin', 'burke']

https://arxiv.org/abs/0910.2593

physical layer network coding with multiple antennas

0910.2603

the two-phase mimo nc (network coding) scheme can be used to boost thethroughput in a two-way relay channel in which nodes are equipped with multipleantennas. the obvious strategy is for the relay node to extract the individualpackets from the two end nodes and mix the two packets to form a network-codedpacket. in this paper, we propose a new scheme called mimo pnc (physicalnetwork coding), in which the relay extracts the summation and difference ofthe two end packets and then converts them to the network-coded form. mimo pncis a natural combination of the single-antenna pnc scheme and the linear mimodetection scheme. the advantages of mimo pnc are many. first, it removes thestringent carrier-phase requirement in single-antenna pnc. second, it is linearin complexity with respect to the constellation size and the number ofsimultaneous data streams in mimo. simulation shows that mimo pnc outperformsthe straightforward mimo nc significantly under random rayleigh fading channel.based on our analysis, we further conjecture that mimo pnc outperforms mimo ncunder all possible realizations of the channel.

cs.ni

nan

['zhang', 'liew']

https://arxiv.org/abs/0910.2603

an agent based classification model

0910.2874

the major function of this model is to access the uci wisconsin breast can-cer data-set[1] and classify the data items into two categories, which arenormal and anomalous. this kind of classifi cation can be referred as anomalydetection, which discriminates anomalous behaviour from normal behaviour incomputer systems. one popular solution for anomaly detection is artifi cialimmune sys- tems (ais). ais are adaptive systems inspired by theoreticalimmunology and observed immune functions, principles and models which areapplied to prob- lem solving. the dendritic cell algorithm (dca)[2] is an aisalgorithm that is developed specifi cally for anomaly detection. it has beensuccessfully applied to intrusion detection in computer security. it isbelieved that agent-based mod- elling is an ideal approach for implementingais, as intelligent agents could be the perfect representations of immuneentities in ais. this model evaluates the feasibility of re-implementing thedca in an agent-based simulation environ- ment called anylogic, where theimmune entities in the dca are represented by intelligent agents. if this modelcan be successfully implemented, it makes it possible to implement morecomplicated and adaptive ais models in the agent-based simulation environment.

cs.ai cs.ma

nan

['gu', 'aickelin', 'greensmith']

https://arxiv.org/abs/0910.2874

which digraphs with ring structure are essentially cyclic?

0910.3113

we say that a digraph is essentially cyclic if its laplacian spectrum is notcompletely real. the essential cyclicity implies the presence of directedcycles, but not vice versa. the problem of characterizing essential cyclicityin terms of graph topology is difficult and yet unsolved. its solution isimportant for some applications of graph theory, including that indecentralized control. in the present paper, this problem is solved withrespect to the class of digraphs with ring structure, which models some typicalcommunication networks. it is shown that the digraphs in this class areessentially cyclic, except for certain specified digraphs. the main technicaltool we employ is the chebyshev polynomials of the second kind. a by-product ofthis study is a theorem on the zeros of polynomials that differ by one from theproducts of chebyshev polynomials of the second kind. we also consider theproblem of essential cyclicity for weighted digraphs and enumerate the spanningtrees in some digraphs with ring structure.

math.co cs.dm cs.ma math.sp

10.1016/j.aam.2010.01.005

['agaev', 'chebotarev']

https://arxiv.org/abs/0910.3113

an idiotypic immune network as a short term learning architecture for mobile robots

0910.3115

a combined short-term learning (stl) and long-term learning (ltl) approach tosolving mobile robot navigation problems is presented and tested in both realand simulated environments. the ltl consists of rapid simulations that use agenetic algorithm to derive diverse sets of behaviours. these sets are thentransferred to an idiotypic artificial immune system (ais), which forms the stlphase, and the system is said to be seeded. the combined ltl-stl approach iscompared with using stl only, and with using a handdesigned controller. inaddition, the stl phase is tested when the idiotypic mechanism is turned off.the results provide substantial evidence that the best option is the seededidiotypic system, i.e. the architecture that merges ltl with an idiotypic aisfor the stl. they also show that structurally different environments can beused for the two phases without compromising transferability

cs.ai cs.ne cs.ro

nan

['whitbrook', 'aickelin', 'garibaldi']

https://arxiv.org/abs/0910.3115

an immune inspired approach to anomaly detection

0910.3117

the immune system provides a rich metaphor for computer security: anomalydetection that works in nature should work for machines. however, earlyartificial immune system approaches for computer security had only limitedsuccess. arguably, this was due to these artificial systems being based on toosimplistic a view of the immune system. we present here a second generationartificial immune system for process anomaly detection. it improves on earliersystems by having different artificial cell types that process information.following detailed information about how to build such second generationsystems, we find that communication between cells types is key to performance.through realistic testing and validation we show that second generationartificial immune systems are capable of anomaly detection beyond genericsystem policies. the paper concludes with a discussion and outline of the nextsteps in this exciting area of computer security.

cs.ai cs.cr cs.ne

nan

['twycross', 'aickelin']

https://arxiv.org/abs/0910.3117

wee lcp

0910.3123

we prove that longest common prefix (lcp) information can be stored in muchless space than previously known. more precisely, we show that in the presenceof the text and the suffix array, o(n) additional bits are sufficient to answerlcp-queries asymptotically in the same time that is needed to retrieve an entryfrom the suffix array. this yields the smallest compressed suffix tree withsub-logarithmic navigation time.

cs.ds

nan

['fischer']

https://arxiv.org/abs/0910.3123

an immune inspired network intrusion detection system utilising correlation context

0910.3124

network intrusion detection systems (nids) are computer systems which monitora network with the aim of discerning malicious from benign activity on thatnetwork. while a wide range of approaches have met varying levels of success,most idss rely on having access to a database of known attack signatures whichare written by security experts. nowadays, in order to solve problems withfalse positive alerts, correlation algorithms are used to add additionalstructure to sequences of ids alerts. however, such techniques are of no helpin discovering novel attacks or variations of known attacks, something thehuman immune system (his) is capable of doing in its own specialised domain.this paper presents a novel immune algorithm for application to the idsproblem. the goal is to discover packets containing novel variations of attackscovered by an existing signature base.

cs.ai cs.cr cs.ne

nan

['tedesco', 'aickelin']

https://arxiv.org/abs/0910.3124

faster algorithms for max-product message-passing

0910.3301

maximum a posteriori inference in graphical models is often solved viamessage-passing algorithms, such as the junction-tree algorithm, or loopybelief-propagation. the exact solution to this problem is well known to beexponential in the size of the model's maximal cliques after it istriangulated, while approximate inference is typically exponential in the sizeof the model's factors. in this paper, we take advantage of the fact that manymodels have maximal cliques that are larger than their constituent factors, andalso of the fact that many factors consist entirely of latent variables (i.e.,they do not depend on an observation). this is a common case in a wide varietyof applications, including grids, trees, and ring-structured models. in suchcases, we are able to decrease the exponent of complexity for message-passingby 0.5 for both exact and approximate inference.

cs.ai cs.ds

nan

['mcauley', 'caetano']

https://arxiv.org/abs/0910.3301

composition and inversion of schema mappings

0910.3372

in the recent years, a lot of attention has been paid to the development ofsolid foundations for the composition and inversion of schema mappings. in thispaper, we review the proposals for the semantics of these crucial operators.for each of these proposals, we concentrate on the three following problems:the definition of the semantics of the operator, the language needed to expressthe operator, and the algorithmic issues associated to the problem of computingthe operator. it should be pointed out that we primarily consider theformalization of schema mappings introduced in the work on data exchange. inparticular, when studying the problem of computing the composition and inverseof a schema mapping, we will be mostly interested in computing these operatorsfor mappings specified by source-to-target tuple-generating dependencies.

cs.db

nan

['arenas', 'perez', 'reutter', 'riveros']

https://arxiv.org/abs/0910.3372

least and greatest fixed points in linear logic

0910.3383

the first-order theory of mall (multiplicative, additive linear logic) overonly equalities is an interesting but weak logic since it cannot captureunbounded (infinite) behavior. instead of accounting for unbounded behavior viathe addition of the exponentials (! and ?), we add least and greatest fixedpoint operators. the resulting logic, which we call mumall, satisfies twofundamental proof theoretic properties: we establish weak normalization for it,and we design a focused proof system that we prove complete. that second resultprovides a strong normal form for cut-free proof structures that can be used,for example, to help automate proof search. we show how these foundations canbe applied to intuitionistic logic.

cs.lo

nan

['baelde']

https://arxiv.org/abs/0910.3383

a complete solution to blackwell's unique ergodicity problem for hidden markov chains

0910.3603

we develop necessary and sufficient conditions for uniqueness of theinvariant measure of the filtering process associated to an ergodic hiddenmarkov model in a finite or countable state space. these results provide acomplete solution to a problem posed by blackwell (1957), and subsume earlierpartial results due to kaijser, kochman and reeds. the proofs of our mainresults are based on the stability theory of nonlinear filters.

math.pr cs.it math.it

10.1214/10-aap688

['chigansky', 'van handel']

https://arxiv.org/abs/0910.3603

optimal bounds for sign-representing the intersection of two halfspaces by polynomials

0910.4224

the threshold degree of a function f:{0,1}^n->{-1,+1} is the least degree ofa real polynomial p with f(x)=sgn p(x). we prove that the intersection of twohalfspaces on {0,1}^n has threshold degree omega(n), which matches the trivialupper bound and completely answers a question due to klivans (2002). the bestprevious lower bound was omega(sqrt n). our result shows that the intersectionof two halfspaces on {0,1}^n only admits a trivial 2^{theta(n)}-time learningalgorithm based on sign-representation by polynomials, unlike the advancesachieved in pac learning dnf formulas and read-once boolean formulas. the proofintroduces a new technique of independent interest, based on fourier analysisand matrix theory.

cs.cc

nan

['sherstov']

https://arxiv.org/abs/0910.4224

putting dots in triangles

0910.4325

given a right-angled triangle of squares in a grid whose horizontal andvertical sides are $n$ squares long, let n(n) denote the maximum number of dotsthat can be placed into the cells of the triangle such that each row, eachcolumn, and each diagonal parallel to the long side of the triangle contains atmost one dot. it has been proven that $n(n) = \lfloor \frac{2n+1}{3} \rfloor$. in this note, we give a new proof of this result using linear programmingtechniques.

cs.dm

nan

['blackburn', 'paterson', 'stinson']

https://arxiv.org/abs/0910.4325

minimal realizations of linear systems: the "shortest basis" approach

0910.4336

given a controllable discrete-time linear system c, a shortest basis for c isa set of linearly independent generators for c with the least possible lengths.a basis b is a shortest basis if and only if it has the predictable spanproperty (i.e., has the predictable delay and degree properties, and isnon-catastrophic), or alternatively if and only if it has the subsystem basisproperty (for any interval j, the generators in b whose span is in j is a basisfor the subsystem c_j). the dimensions of the minimal state spaces and minimaltransition spaces of c are simply the numbers of generators in a shortest basisb that are active at any given state or symbol time, respectively. a minimallinear realization for c in controller canonical form follows directly from ashortest basis for c, and a minimal linear realization for c in observercanonical form follows directly from a shortest basis for the orthogonal systemc^\perp. this approach seems conceptually simpler than that of classicalminimal realization theory.

cs.it cs.sy math.it math.oc

nan

['forney']

https://arxiv.org/abs/0910.4336

scotgrid: providing an effective distributed tier-2 in the lhc era

0910.4507

scotgrid is a distributed tier-2 centre in the uk with sites in durham,edinburgh and glasgow. scotgrid has undergone a huge expansion in hardware inanticipation of the lhc and now provides more than 4msi2k and 500tb to the lhcvos. scaling up to this level of provision has brought many challenges to thetier-2 and we show in this paper how we have adopted new methods of organisingthe centres, from fabric management and monitoring to remote management ofsites to management and operational procedures, to meet these challenges. wedescribe how we have coped with different operational models at the sites,where glagsow and durham sites are managed "in house" but resources atedinburgh are managed as a central university resource. this required theadoption of a different fabric management model at edinburgh and a specialengagement with the cluster managers. challenges arose from the different jobmodels of local and grid submission that required special attention to resolve.we show how scotgrid has successfully provided an infrastructure for atlas andlhcb monte carlo production. special attention has been paid to ensuring thatuser analysis functions efficiently, which has required optimisation of localstorage and networking to cope with the demands of user analysis. finally,although these tier-2 resources are pledged to the whole vo, we haveestablished close links with our local physics user communities as being thebest way to ensure that the tier-2 functions effectively as a part of the lhcgrid computing framework..

cs.dc

10.1088/1742-6596/219/5/052014

['skipsey', 'ambrose-griffith', 'cowan', 'kenyon', 'richards', 'roffe', 'stewart']

https://arxiv.org/abs/0910.4507

optimised access to user analysis data using the glite dpm

0910.4510

the scotgrid distributed tier-2 now provides more that 4msi2k and 500tb forlhc computing, which is spread across three sites at durham, edinburgh andglasgow. tier-2 sites have a dual role to play in the computing models of thelhc vos. firstly, their cpu resources are used for the generation of montecarlo event data. secondly, the end user analysis data is distributed acrossthe grid to the site's storage system and held on disk ready for processing byphysicists' analysis jobs. in this paper we show how we have designed thescotgrid storage and data management resources in order to optimise access byphysicists to lhc data. within scotgrid, all sites use the glite dpm storagemanager middleware. using the egee grid to submit real atlas analysis code toprocess vo data stored on the scotgrid sites, we present an analysis of theperformance of the architecture at one site, and procedures that may beundertaken to improve such. the results will be presented from the point ofview of the end user (in terms of number of events processed/second) and fromthe point of view of the site, which wishes to minimise load and the impactthat analysis activity has on other users of the system.

cs.dc

10.1088/1742-6596/219/6/062066

['skipsey', 'cowan', 'kenyon', 'purdie', 'stewart']

https://arxiv.org/abs/0910.4510

classification with tarskian system executions (bakery algorithms as an example)

0910.4565

we argue that predicate languages and their tarskian structures have animportant place for the study of concurrency. the argument in our paper isbased on an example: we show that two seemingly dissimilar algorithms have acommon set of high-level properties, which reveals their affinity. thealgorithms are a variant of lamport's bakery algorithm and the ricart andagrawala algorithm. they seem different because one uses shared memory and theother message passing for communication. yet it is intuitively obvious thatthey are in some sense very similar, and they belong to the same "family ofbakery algorithms". the aim of this paper is to express in a formal way thisintuition that classifies the two algorithms together. for this aim ofexpressing the abstract high level properties that are shared by the twoalgorithms we use predicate languages and their taskian structures. we find aset of properties expressed in quantification language which are satisfied byevery tarskian system execution that models a run by either one of theprotocols, and which is strong enough to ensure that the mutual exclusionproperty holds in these runs.

cs.lo

nan

['abraham']

https://arxiv.org/abs/0910.4565

a game theoretic analysis of incentives in content production and sharing over peer-to-peer networks

0910.4618

user-generated content can be distributed at a low cost using peer-to-peer(p2p) networks, but the free-rider problem hinders the utilization of p2pnetworks. in order to achieve an efficient use of p2p networks, we investigatefundamental issues on incentives in content production and sharing using gametheory. we build a basic model to analyze non-cooperative outcomes without anincentive scheme and then use different game formulations derived from thebasic model to examine five incentive schemes: cooperative, payment, repeatedinteraction, intervention, and enforced full sharing. the results of this papershow that 1) cooperative peers share all produced content while non-cooperativepeers do not share at all without an incentive scheme; 2) a cooperative schemeallows peers to consume more content than non-cooperative outcomes do; 3) acooperative outcome can be achieved among non-cooperative peers by introducingan incentive scheme based on payment, repeated interaction, or intervention;and 4) enforced full sharing has ambiguous welfare effects on peers. inaddition to describing the solutions of different formulations, we discussenforcement and informational requirements to implement each solution, aimingto offer a guideline for protocol designers when designing incentive schemesfor p2p networks.

cs.ni cs.gt

10.1109/jstsp.2010.2048609

['park', 'van der schaar']

https://arxiv.org/abs/0910.4618

moderate deviations of the random riccati equation

0910.4686

we characterize the invariant filtering measures resulting from kalmanfiltering with intermittent observations (\cite{bruno}), where the observationarrival is modeled as a bernoulli process. in \cite{riccati-weakconv}, it wasshown that there exists a $\overline{\gamma}^{\{\scriptsize{sb}}}>0$ such thatfor every observation packet arrival probability $\overline{\gamma}$,$\overline{\gamma}>\overline{\gamma}^{\{\scriptsize{sb}}}>0$, the sequence ofrandom conditional error covariance matrices converges in distribution to aunique invariant distribution $\mathbb{\mu}^{\overline{\gamma}}$ (independentof the filter initialization.) in this paper, we prove that, for controllableand observable systems, $\overline{\gamma}^{\{\scriptsize{sb}}}=0$ and that, as$\overline{\gamma}\uparrow 1$, the family$\{\mathbb{\mu}^{\overline{\gamma}}\}_{\overline{\gamma}>0}$ of invariantdistributions satisfies a moderate deviations principle (mdp) with a good ratefunction $i$. the rate function $i$ is explicitly identified. in particular,our results show:

math.pr cs.it math.ds math.it math.oc

nan

['kar', 'moura']

https://arxiv.org/abs/0910.4686

a $p$-adic ransac algorithm for stereo vision using hensel lifting

0910.4839

a $p$-adic variation of the ran(dom) sa(mple) c(onsensus) method for solvingthe relative pose problem in stereo vision is developped. from two 2-adicallyencoded images a random sample of five pairs of corresponding points is taken,and the equations for the essential matrix are solved by lifting solutionsmodulo 2 to the 2-adic integers. a recently devised $p$-adic hierarchicalclassification algorithm imitating the known lbg quantisation method classifiesthe solutions for all the samples after having determined the number ofclusters using the known intra-inter validity of clusterings. in the successfulcase, a cluster ranking will determine the cluster containing a 2-adicapproximation to the "true" solution of the problem.

cs.cv

10.1134/s2070046610010048

['bradley']

https://arxiv.org/abs/0910.4839

artificial immune systems

0910.4899

the biological immune system is a robust, complex, adaptive system thatdefends the body from foreign pathogens. it is able to categorize all cells (ormolecules) within the body as self-cells or non-self cells. it does this withthe help of a distributed task force that has the intelligence to take actionfrom a local and also a global perspective using its network of chemicalmessengers for communication. there are two major branches of the immunesystem. the innate immune system is an unchanging mechanism that detects anddestroys certain invading organisms, whilst the adaptive immune system respondsto previously unknown foreign cells and builds a response to them that canremain in the body over a long period of time. this remarkable informationprocessing biological system has caught the attention of computer science inrecent years. a novel computational intelligence technique, inspired byimmunology, has emerged, called artificial immune systems. several conceptsfrom the immune have been extracted and applied for solution to real worldscience and engineering problems. in this tutorial, we briefly describe theimmune system metaphors that are relevant to existing artificial immune systemsmethods. we will then show illustrative real-world problems suitable forartificial immune systems and give a step-by-step algorithm walkthrough for onesuch problem. a comparison of the artificial immune systems to other well-knownalgorithms, areas for future work, tips & tricks and a list of resources willround this tutorial off. it should be noted that as artificial immune systemsis still a young and evolving field, there is not yet a fixed algorithmtemplate and hence actual implementations might differ somewhat from time totime and from those examples given here.

cs.ai cs.ne

nan

['aickelin', 'dasgupta']

https://arxiv.org/abs/0910.4899

articulation and clarification of the dendritic cell algorithm

0910.4903

the dendritic cell algorithm (dca) is inspired by recent work in innateimmunity. in this paper a formal description of the dca is given. the dca isdescribed in detail, and its use as an anomaly detector is illustrated withinthe context of computer security. a port scan detection task is performed tosubstantiate the influence of signal selection on the behaviour of thealgorithm. experimental results provide a comparison of differing input signalmappings.

cs.ai cs.ne

nan

['greensmith', 'aickelin', 'twycross']

https://arxiv.org/abs/0910.4903

a general upper bound on the size of constant-weight conflict-avoiding codes

0910.5073

conflict-avoiding codes are used in the multiple-access collision channelwithout feedback. the number of codewords in a conflict-avoiding code is thenumber of potential users that can be supported in the system. in this paper, anew upper bound on the size of conflict-avoiding codes is proved. this upperbound is general in the sense that it is applicable to all code lengths and allhamming weights. several existing constructions for conflict-avoiding codes,which are known to be optimal for hamming weights equal to four and five, areshown to be optimal for all hamming weights in general.

cs.it math.it

10.1109/tit.2010.2048508

['shum', 'wong', 'chen']

https://arxiv.org/abs/0910.5073

algorithmic randomness and monotone complexity on product space

0910.5076

we study algorithmic randomness and monotone complexity on product of the setof infinite binary sequences. we explore the following problems: monotonecomplexity on product space, lambalgen's theorem for correlated probability,classification of random sets by likelihood ratio tests, decomposition ofcomplexity and independence, bayesian statistics for individual randomsequences. formerly lambalgen's theorem for correlated probability is shownunder a uniform computability assumption in [h. takahashi inform. comp. 2008].in this paper we show the theorem without the assumption.

cs.it math.it

nan

['takahashi']

https://arxiv.org/abs/0910.5076

the complexity of iterated strategy elimination

0910.5107

we consider the computational complexity of the question whether a certainstrategy can be removed from a game by means of iterated elimination ofdominated strategies. in particular, we study the influence of differentdefinitions of domination and of the number of different payoff values. inaddition, the consequence of restriction to constant-sum games is shown.

cs.cc cs.gt

nan

['pauly']

https://arxiv.org/abs/0910.5107

artificial immune tissue using self-orgamizing networks

0910.5405

as introduced by bentley et al. (2005), artificial immune systems (ais) arelacking tissue, which is present in one form or another in all livingmulti-cellular organisms. some have argued that this concept in the context ofais brings little novelty to the already saturated field of the immune inspiredcomputational research. this article aims to show that such a component of anais has the potential to bring an advantage to a data processing algorithm interms of data pre-processing, clustering and extraction of features desired bythe immune inspired system. the proposed tissue algorithm is based onself-organizing networks, such as self-organizing maps (som) developed bykohonen (1996) and an analogy of the so called toll-like receptors (tlr)affecting the activation function of the clusters developed by the som.

cs.ai cs.ne

nan

['feyereisl', 'aickelin']

https://arxiv.org/abs/0910.5405

the cyborg astrobiologist: testing a novelty-detection algorithm on two mobile exploration systems at rivas vaciamadrid in spain and at the mars desert research station in utah

0910.5454

(abridged) in previous work, two platforms have been developed for testingcomputer-vision algorithms for robotic planetary exploration (mcguire et al.2004b,2005; bartolo et al. 2007). the wearable-computer platform has beentested at geological and astrobiological field sites in spain (rivasvaciamadrid and riba de santiuste), and the phone-camera has been tested at ageological field site in malta. in this work, we (i) apply a hopfieldneural-network algorithm for novelty detection based upon color, (ii) integratea field-capable digital microscope on the wearable computer platform, (iii)test this novelty detection with the digital microscope at rivas vaciamadrid,(iv) develop a bluetooth communication mode for the phone-camera platform, inorder to allow access to a mobile processing computer at the field sites, and(v) test the novelty detection on the bluetooth-enabled phone-camera connectedto a netbook computer at the mars desert research station in utah. this systemsengineering and field testing have together allowed us to develop a real-timecomputer-vision system that is capable, for example, of identifying lichens asnovel within a series of images acquired in semi-arid desert environments. weacquired sequences of images of geologic outcrops in utah and spain consistingof various rock types and colors to test this algorithm. the algorithm robustlyrecognized previously-observed units by their color, while requiring only asingle image or a few images to learn colors as familiar, demonstrating itsfast learning capability.

cs.cv astro-ph.ep astro-ph.im cs.lg stat.ml

10.1017/s1473550409990358

['mcguire', 'gross', 'wendt', 'bonnici', 'souza-egipsy', 'ormo', 'diaz-martinez', 'foing', 'bose', 'walter', 'oesker', 'ontrup', 'haschke', 'ritter']

https://arxiv.org/abs/0910.5454

finding overlapping communities in networks by label propagation

0910.5516

we propose an algorithm for finding overlapping community structure in verylarge networks. the algorithm is based on the label propagation technique ofraghavan, albert, and kumara, but is able to detect communities that overlap.like the original algorithm, vertices have labels that propagate betweenneighbouring vertices so that members of a community reach a consensus on theircommunity membership. our main contribution is to extend the label andpropagation step to include information about more than one community: eachvertex can now belong to up to v communities, where v is the parameter of thealgorithm. our algorithm can also handle weighted and bipartite networks. testson an independently designed set of benchmarks, and on real networks, show thealgorithm to be highly effective in recovering overlapping communities. it isalso very fast and can process very large and dense networks in a short time.

physics.soc-ph cs.si

10.1088/1367-2630/12/10/103018

['gregory']

https://arxiv.org/abs/0910.5516

high dimensional error-correcting codes

0910.5697

in this paper we construct multidimensional codes with high dimension. thecodes can correct high dimensional errors which have the form of either smallclusters, or confined to an area with a small radius. we also consider smallnumber of errors in a small area. the clusters which are discussed are mainlyspheres such as semi-crosses and crosses. also considered are clusters withsmall number of errors such as 2-bursts, two errors in various clusters, andthree errors on a line. our main focus is on the redundancy of the codes whenthe most dominant parameter is the dimension of the code.

cs.it math.it

nan

['yaakobi', 'etzion']

https://arxiv.org/abs/0910.5697

approximate privacy: foundations and quantification

0910.5714

increasing use of computers and networks in business, government, recreation,and almost all aspects of daily life has led to a proliferation of onlinesensitive data about individuals and organizations. consequently, concern aboutthe privacy of these data has become a top priority, particularly those datathat are created and used in electronic commerce. there have been manyformulations of privacy and, unfortunately, many negative results about thefeasibility of maintaining privacy of sensitive data in realistic networkedenvironments. we formulate communication-complexity-based definitions, bothworst-case and average-case, of a problem's privacy-approximation ratio. we useour definitions to investigate the extent to which approximate privacy isachievable in two standard problems: the second-price vickrey auction and themillionaires problem of yao. for both the second-price vickrey auction and the millionaires problem, weshow that not only is perfect privacy impossible or infeasibly costly toachieve, but even close approximations of perfect privacy suffer from the samelower bounds. by contrast, we show that, if the values of the parties are drawnuniformly at random from {0,...,2^k-1}, then, for both problems, simple andnatural communication protocols have privacy-approximation ratios that arelinear in k (i.e., logarithmic in the size of the space of possible inputs). weconjecture that this improved privacy-approximation ratio is achievable for anyprobability distribution.

cs.cr cs.gt

nan

['feigenbaum', 'jaggard', 'schapira']

https://arxiv.org/abs/0910.5714

quantifying pervasive authentication: the case of the hancke-kuhn protocol

0910.5745

as mobile devices pervade physical space, the familiar authenticationpatterns are becoming insufficient: besides entity authentication, manyapplications require, e.g., location authentication. many interesting protocolshave been proposed and implemented to provide such strengthened forms ofauthentication, but there are very few proofs that such protocols satisfy therequired security properties. the logical formalisms, devised for reasoningabout security protocols on standard computer networks, turn out to bedifficult to adapt for reasoning about hybrid protocols, used in pervasive andheterogenous networks. <p> we refine the dolev-yao-style algebraic method for protocol analysis by aprobabilistic model of guessing, needed to analyze protocols that mix weakcryptography with physical properties of nonstandard communication channels.applying this model, we provide a precise security proof for a proximityauthentication protocol, due to hancke and kuhn, that uses a subtle form ofprobabilistic reasoning to achieve its goals.

cs.cr

nan

['pavlovic', 'meadows']

https://arxiv.org/abs/0910.5745

the positive semidefinite grothendieck problem with rank constraint

0910.5765

given a positive integer n and a positive semidefinite matrix a = (a_{ij}) ofsize m x m, the positive semidefinite grothendieck problem withrank-n-constraint (sdp_n) is maximize \sum_{i=1}^m \sum_{j=1}^m a_{ij} x_i \cdot x_j, where x_1, ..., x_m\in s^{n-1}. in this paper we design a polynomial time approximation algorithm for sdp_nachieving an approximation ratio of \gamma(n) = \frac{2}{n}(\frac{\gamma((n+1)/2)}{\gamma(n/2)})^2 = 1 -\theta(1/n). we show that under the assumption of the unique games conjecture the achievedapproximation ratio is optimal: there is no polynomial time algorithm whichapproximates sdp_n with a ratio greater than \gamma(n). we improve theapproximation ratio of the best known polynomial time algorithm for sdp_1 from2/\pi to 2/(\pi\gamma(m)) = 2/\pi + \theta(1/m), and we show a tighterapproximation ratio for sdp_n when a is the laplacian matrix of a graph withnonnegative edge weights.

math.oc cs.ds math.co math.fa

10.1007/978-3-642-14165-2_4

['briet', 'filho', 'vallentin']

https://arxiv.org/abs/0910.5765

entropy sensitivity of languages defined by infinite automata, via markov chains with forbidden transitions

0911.0142

a language l over a finite alphabet is growth-sensitive (or entropysensitive) if forbidding any set of subwords f yields a sub-language l^f whoseexponential growth rate (entropy) is smaller than that of l. let (x, e, l) bean infinite, oriented, labelled graph. considering the graph as an (infinite)automaton, we associate with any pair of vertices x,y in x the languageconsisting of all words that can be read as the labels along some path from xto y. under suitable, general assumptions we prove that these languages aregrowth-sensitive. this is based on using markov chains with forbiddentransitions.

cs.fl math.pr

10.1016/j.tcs.2010.07.020

['huss', 'sava', 'woess']

https://arxiv.org/abs/0911.0142

fast arc-annotated subsequence matching in linear space

0911.0577

an arc-annotated string is a string of characters, called bases, augmentedwith a set of pairs, called arcs, each connecting two bases. givenarc-annotated strings $p$ and $q$ the arc-preserving subsequence problem is todetermine if $p$ can be obtained from $q$ by deleting bases from $q$. whenevera base is deleted any arc with an endpoint in that base is also deleted.arc-annotated strings where the arcs are ``nested'' are a natural model of rnamolecules that captures both the primary and secondary structure of these. thearc-preserving subsequence problem for nested arc-annotated strings is basicprimitive for investigating the function of rna molecules. gramm et al. [acmtrans. algorithms 2006] gave an algorithm for this problem using $o(nm)$ timeand space, where $m$ and $n$ are the lengths of $p$ and $q$, respectively. inthis paper we present a new algorithm using $o(nm)$ time and $o(n + m)$ space,thereby matching the previous time bound while significantly reducing the spacefrom a quadratic term to linear. this is essential to process large rnamolecules where the space is likely to be a bottleneck. to obtain our result weintroduce several novel ideas which may be of independent interest for relatedproblems on arc-annotated strings.

cs.ds

nan

['bille', 'goertz']

https://arxiv.org/abs/0911.0577

research report : collaborative peer 2 peer edition: avoiding conflicts is better than solving conflicts

0911.0838

collaborative edition is achieved by distinct sites that work independentlyon (a copy of) a shared document. conflicts may arise during this process andmust be solved by the collaborative editor. in pure peer to peer collaborativeediting, no centralization nor locks nor time-stamps are used which makeconflict resolution difficult. we propose an algorithm which relies on thenotion or semantics dependence and avoids the need of any integrationtransformation to solve conflicts. furthermore, it doesn't use any history filerecording operations performed since starting the edition process. we show howto define editing operations for semi-structured documents i.e. xml-like trees,that are enriched with informations derived for free from the editing process.then we define the semantics dependence relation required by the algorithm andwe present preliminary results obtained by a prototype implementation.

cs.hc

nan

['martin', 'lugiez']

https://arxiv.org/abs/0911.0838

sampling and reconstruction of signals in a reproducing kernel subspace of $l^p({\bbb r}^d)$

0911.0844

in this paper, we consider sampling and reconstruction of signals in areproducing kernel subspace of $l^p(\rd), 1\le p\le \infty$, associated with anidempotent integral operator whose kernel has certain off-diagonal decay andregularity. the space of $p$-integrable non-uniform splines and theshift-invariant spaces generated by finitely many localized functions are ourmodel examples of such reproducing kernel subspaces of $l^p(\rd)$. we show thata signal in such reproducing kernel subspaces can be reconstructed in a stableway from its samples taken on a relatively-separated set with sufficientlysmall gap. we also study the exponential convergence, consistency, and theasymptotic pointwise error estimate of the iterative approximation-projectionalgorithm and the iterative frame algorithm for reconstructing a signal inthose reproducing kernel spaces from its samples with sufficiently small gap.

cs.it math.fa math.it

10.1016/j.jfa.2009.12.012

['nashed', 'sun']

https://arxiv.org/abs/0911.0844

waveform transmission method, a new waveform-relaxation based algorithm to solve ordinary differential equations in parallel

0911.1166

waveform relaxation method (wr) is a beautiful algorithm to solve ordinarydifferential equations (odes). however, because of its poor convergencecapability, it was rarely used. in this paper, we propose a new distributedalgorithm, named waveform transmission method (wtm), by virtually insertingwaveform transmission lines into the dynamical system to achieve distributedcomputing of extremely large odes. wtm has better convergence capability thanthe traditional wr algorithms.

math.na cs.dc math.ca

nan

['wei', 'yang']

https://arxiv.org/abs/0911.1166

affine grassmann codes

0911.1298

we consider a new class of linear codes, called affine grassmann codes. thesecan be viewed as a variant of generalized reed-muller codes and are closelyrelated to grassmann codes. we determine the length, dimension, and the minimumdistance of any affine grassmann code. moreover, we show that affine grassmanncodes have a large automorphism group and determine the number of minimumweight codewords.

cs.it math.ag math.it

10.1109/tit.2010.2048470

['beelen', 'ghorpade', 'hoeholdt']

https://arxiv.org/abs/0911.1298

on improving the representation of a region achieved by a sensor network

0911.1379

this report considers the class of applications of sensor networks in whicheach sensor node makes measurements, such as temperature or humidity, at theprecise location of the node. such spot-sensing applications approximate thephysical condition of the entire region of interest by the measurements made atonly the points where the sensor nodes are located. given a certain density ofnodes in a region, a more spatially uniform distribution of the nodes leads toa better approximation of the physical condition of the region. this reportconsiders the error in this approximation and seeks to improve the quality ofrepresentation of the physical condition of the points in the region in thedata collected by the sensor network. we develop two essential metrics whichtogether allow a rigorous quantitative assessment of the quality ofrepresentation achieved: the average representation error and the unevenness ofrepresentation error, the latter based on a well-accepted measure of inequalityused in economics. we present the rationale behind the use of these metrics andderive relevant theoretical bounds on them in the common scenario of a planarregion of arbitrary shape covered by a sensor network deployment. a simple newheuristic algorithm is presented for each node to determine if and when itshould sense or sleep to conserve energy while also preserving the quality ofrepresentation. simulation results show that it achieves a significantimprovement in the quality of representation compared to other relateddistributed algorithms. interestingly, our results also show that improvedspatial uniformity has the welcome side-effect of a significant increase in thenetwork lifetime.

cs.ni

nan

['chu', 'sethu']

https://arxiv.org/abs/0911.1379

mimo cluster cooperative assignment cross layer scheme for hybrid wireless sensor networks

0911.1511

the dual-cross scenario of the hybrid wireless sensor networks (wsns) isstudied and a novel mimo cluster cooperative assignment cross layer schedulingscheme (mcca-clss) is proposed in this paper. the comparison and thepredominance of the proposed scheme are demonstrated, the clusters areoptimized. with the help of the simulations, the relative energy consumptionand the end-to-end blocking probability are all improved. the addressing ratioof success in the condition of the unchanged parameters and externalinformation can be increased and the network can tolerate more hops to supportreliable transportation by the proposed scheme.

cs.ni

10.1109/iscit.2009.5341234

['zhong', 'kwak', 'ullah']

https://arxiv.org/abs/0911.1511

a discourse-based approach in text-based machine translation

0911.1516

this paper presents a theoretical research based approach to ellipsisresolution in machine translation. the formula of discourse is applied in orderto resolve ellipses. the validity of the discourse formula is analyzed byapplying it to the real world text, i.e., newspaper fragments. the source textis converted into mono-sentential discourses where complex discourses requirefurther dissection either directly into primitive discourses or first intocompound discourses and later into primitive ones. the procedure of dissectionneeds further improvement, i.e., discovering as many primitive discourse formsas possible. an attempt has been made to investigate new primitive discoursesor patterns from the given text.

cs.cl

nan

['ullah', 'khan', 'kwak']

https://arxiv.org/abs/0911.1516

resolution of unidentified words in machine translation

0911.1517

this paper presents a mechanism of resolving unidentified lexical units intext-based machine translation (tbmt). in a machine translation (mt) system itis unlikely to have a complete lexicon and hence there is intense need of a newmechanism to handle the problem of unidentified words. these unknown wordscould be abbreviations, names, acronyms and newly introduced terms. we haveproposed an algorithm for the resolution of the unidentified words. thisalgorithm takes discourse unit (primitive discourse) as a unit of analysis andprovides real time updates to the lexicon. we have manually applied thealgorithm to news paper fragments. along with anaphora and cataphoraresolution, many unknown words especially names and abbreviations were updatedto the lexicon.

cs.cl

nan

['ullah', 'hussain', 'kwak']

https://arxiv.org/abs/0911.1517

fast approximation in subspaces by doubling metric decomposition

0911.1626

in this paper we propose and study a new complexity model for approximationalgorithms. the main motivation are practical problems over large data setsthat need to be solved many times for different scenarios, e.g., many multicasttrees that need to be constructed for different groups of users. in our modelwe allow a preprocessing phase, when some information of the input graph$g=(v,e)$ is stored in a limited size data structure. next, the data structureenables processing queries of the form ``solve problem a for an input$s\subseteq v$''. we consider problems like {\sc steiner forest}, {\sc facilitylocation}, {\sc $k$-median}, {\sc $k$-center} and {\sc tsp} in the case whenthe graph induces a doubling metric. our main results are data structures ofnear-linear size that are able to answer queries in time close to linear in$|s|$. this improves over typical worst case reuniting time of approximationalgorithms in the classical setting which is $\omega(|e|)$ independently of thequery size. in most cases, our approximation guarantees are arbitrarily closeto those in the classical setting. additionally, we present the first fullydynamic algorithm for the steiner tree problem.

cs.ds

nan

['cygan', 'kowalik', 'mucha', 'pilipczuk', 'sankowski']

https://arxiv.org/abs/0911.1626

vertical partitioning of relational oltp databases using integer programming

0911.1691

a way to optimize performance of relational row store databases is to reducethe row widths by vertically partitioning tables into table fractions in orderto minimize the number of irrelevant columns/attributes read by eachtransaction. this paper considers vertical partitioning algorithms forrelational row-store oltp databases with an h-store-like architecture, meaningthat we would like to maximize the number of single-sited transactions. wepresent a model for the vertical partitioning problem that, given a schematogether with a vertical partitioning and a workload, estimates the costs(bytes read/written by storage layer access methods and bytes transferredbetween sites) of evaluating the workload on the given partitioning. the costmodel allows for arbitrarily prioritizing load balancing of sites vs. totalcost minimization. we show that finding a minimum-cost vertical partitioning inthis model is np-hard and present two algorithms returning solutions in whichsingle-sitedness of read queries is preserved while allowing column replication(which may allow a drastically reduced cost compared to disjoint partitioning).the first algorithm is a quadratic integer program that finds optimalminimum-cost solutions with respect to the model, and the second algorithm is amore scalable heuristic based on simulated annealing. experiments show that thealgorithms can reduce the cost of the model objective by 37% when applied tothe tpc-c benchmark and the heuristic is shown to obtain solutions with costclose to the ones found using the quadratic program.

cs.db cs.pf

nan

['amossen']

https://arxiv.org/abs/0911.1691

the replicator equation as an inference dynamic

0911.1763

the replicator equation is interpreted as a continuous inference equation anda formal similarity between the discrete replicator equation and bayesianinference is described. further connections between inference and thereplicator equation are given including a discussion of information divergencesand exponential families as solutions for the replicator dynamic, using fisherinformation and information geometry.

math.ds cs.it math.it

nan

['harper']

https://arxiv.org/abs/0911.1763

a natural dynamics for bargaining on exchange networks

0911.1767

bargaining networks model the behavior of a set of players that need to reachpairwise agreements for making profits. nash bargaining solutions are specialoutcomes of such games that are both stable and balanced. kleinberg and tardosproved a sharp algorithmic characterization of such outcomes, but left open theproblem of how the actual bargaining process converges to them. a partialanswer was provided by azar et al. who proposed a distributed algorithm forconstructing nash bargaining solutions, but without polynomial bounds on itsconvergence rate. in this paper, we introduce a simple and natural model forthis process, and study its convergence rate to nash bargaining solutions. ateach time step, each player proposes a deal to each of her neighbors. theproposal consists of a share of the potential profit in case of agreement. theshare is chosen to be balanced in nash's sense as far as this is feasible (withrespect to the current best alternatives for both players). we prove that,whenever the nash bargaining solution is unique (and satisfies a positive gapcondition) this dynamics converges to it in polynomial time. our analysis isbased on an approximate decoupling phenomenon between the dynamics on differentsubstructures of the network. this approach may be of general interest for theanalysis of local algorithms on networks.

cs.gt

nan

['kanoria', 'bayati', 'borgs', 'chayes', 'montanari']

https://arxiv.org/abs/0911.1767

big macs and eigenfactor scores: don't let correlation coefficients fool you

0911.1807

the eigenfactor metrics provide an alternative way of evaluating scholarlyjournals based on an iterative ranking procedure analogous to google's pagerankalgorithm. these metrics have recently been adopted by thomson-reuters and arelisted alongside the impact factor in the journal citation reports. but dothese metrics differ sufficiently so as to be a useful addition to thebibliometric toolbox? davis (2008) has argued otherwise, based on his findingof a 0.95 correlation coefficient between eigenfactor score and total citationsfor a sample of journals in the field of medicine. this conclusion is mistaken;here we illustrate the basic statistical fallacy to which davis succumbed. weprovide a complete analysis of the 2006 journal citation reports anddemonstrate that there are statistically and economically significantdifferences between the information provided by the eigenfactor metrics andthat provided by impact factor and total citations.

cs.dl

nan

['west', 'bergstrom', 'bergstrom']

https://arxiv.org/abs/0911.1807

functional units for natural numbers

0911.1851

interaction with services provided by an execution environment forms part ofthe behaviours exhibited by instruction sequences under execution. mechanismsrelated to the kind of interaction in question have been proposed in thesetting of thread algebra. like thread, service is an abstract behaviouralconcept. the concept of a functional unit is similar to the concept of aservice, but more concrete. a state space is inherent in the concept of afunctional unit, whereas it is not inherent in the concept of a service. inthis paper, we establish the existence of a universal computable functionalunit for natural numbers and related results.

cs.pl cs.lo

nan

['bergstra', 'middelburg']

https://arxiv.org/abs/0911.1851

quantifying transversality by measuring the robustness of intersections

0911.2142

by definition, transverse intersections are stable under infinitesimalperturbations. using persistent homology, we extend this notion to a measure.given a space of perturbations, we assign to each homology class of theintersection its robustness, the magnitude of a perturbations in this spacenecessary to kill it, and prove that robustness is stable. among theapplications of this result is a stable notion of robustness for fixed pointsof continuous mappings and a statement of stability for contours of smoothmappings.

cs.cg math.gm

nan

['edelsbrunner', 'morozov', 'patel']

https://arxiv.org/abs/0911.2142

qpace -- a qcd parallel computer based on cell processors

0911.2174

qpace is a novel parallel computer which has been developed to be primarilyused for lattice qcd simulations. the compute power is provided by the ibmpowerxcell 8i processor, an enhanced version of the cell processor that is usedin the playstation 3. the qpace nodes are interconnected by a custom,application optimized 3-dimensional torus network implemented on an fpga. toachieve the very high packaging density of 26 tflops per rack a new watercooling concept has been developed and successfully realized. in this paper wegive an overview of the architecture and highlight some important technicaldetails of the system. furthermore, we provide initial performance results andreport on the installation of 8 qpace racks providing an aggregate peakperformance of 200 tflops.

hep-lat cs.ar

nan

['baier', 'boettiger', 'drochner', 'eicker', 'fischer', 'fodor', 'frommer', 'gomez', 'goldrian', 'heybrock', 'hierl', 'hüsken', 'huth', 'krill', 'lauritsen', 'lippert', 'maurer', 'mendl', 'meyer', 'nobile', 'ouda', 'pivanti', 'pleiter', 'ries', 'schäfer', 'schick', 'schifano', 'simma', 'solbrig', 'streuer', 'sulanke', 'tripiccione', 'vogt', 'wettig', 'winter']

https://arxiv.org/abs/0911.2174

approximation schemes for the betweenness problem in tournaments and related ranking problems

0911.2214

we design the first polynomial time approximation schemes (ptass) for theminimum betweenness problem in tournaments and some related higher arityranking problems. this settles the approximation status of the betweennessproblem in tournaments along with other ranking problems which were open forsome time now. the results depend on a new technique of dealing with fragileranking constraints and could be of independent interest.

cs.ds cs.dm

nan

['karpinski', 'schudy']

https://arxiv.org/abs/0911.2214

tile packing tomography is np-hard

0911.2567

discrete tomography deals with reconstructing finite spatial objects fromlower dimensional projections and has applications for example in timetabledesign. in this paper we consider the problem of reconstructing a tile packingfrom its row and column projections. it consists of disjoint copies of a fixedtile, all contained in some rectangular grid. the projections tell how manycells are covered by a tile in each row and column. how difficult is it toconstruct a tile packing satisfying given projections? it was known to besolvable by a greedy algorithm for bars (tiles of width or height 1), andnp-hardness results were known for some specific tiles. this paper shows thatthe problem is np-hard whenever the tile is not a bar.

cs.cc cs.ds

nan

['chrobak', 'durr', 'guinez', 'lozano', 'thang']

https://arxiv.org/abs/0911.2567

neural networks for dynamic shortest path routing problems - a survey

0911.2865

this paper reviews the overview of the dynamic shortest path routing problemand the various neural networks to solve it. different shortest pathoptimization problems can be solved by using various neural networksalgorithms. the routing in packet switched multi-hop networks can be describedas a classical combinatorial optimization problem i.e. a shortest path routingproblem in graphs. the survey shows that the neural networks are the bestcandidates for the optimization of dynamic shortest path routing problems dueto their fastness in computation comparing to other softcomputing andmetaheuristics algorithms

cs.ne cs.ai

nan

['nallusamy', 'duraiswamy']

https://arxiv.org/abs/0911.2865

computation speed of the f.a.s.t. model

0911.2900

the f.a.s.t. model for microscopic simulation of pedestrians was formulatedwith the idea of parallelizability and small computation times in general inmind, but so far it was never demonstrated, if it can in fact be implementedefficiently for execution on a multi-core or multi-cpu system. in thiscontribution results are given on computation times for the f.a.s.t. model onan eight-core pc.

cs.ma physics.soc-ph

nan

['kretz']

https://arxiv.org/abs/0911.2900