dedup-isc-ft-v107-score
float64 0.32
1
| uid
stringlengths 32
32
| text
stringlengths 0
32.5k
| paper_id
stringlengths 1
14
| original_image_filename
stringlengths 5
222
|
|---|---|---|---|---|
0.542963 | 039d714f827746a6a2b7ab5ef155fb3f | Initial spectral dist. | 2012.13975 | dist1.jpg |
0.422007 | b131128b7c904694a9b751cebbec93de | Push-forward (MaxExp) | 2012.13975 | dist2.jpg |
0.426654 | 4f088b3a98ad4e8d98ede30ba2b0c44f | Push-forward (Gamma) | 2012.13975 | dist3.jpg |
0.415654 | 6fea734a24064641ae8dcd4e5bb8b995 | Push-forward (HDP) | 2012.13975 | dist4.jpg |
0.485277 | cf6ae0356acc459281c9ffbd6bde832d | 2012.13975 | eval1c.jpg |
|
0.455795 | 2b3b8ab3748d4477836b3d3200cb0655 | 2012.13975 | eval2c.jpg |
|
0.477585 | 06a15b8ea1454f688a60ac0df74bca41 | 2012.13975 | eval3c.jpg |
|
0.493939 | 1c108721c2fd48f6b2caa11b594ce753 | 2012.13975 | eval5c.jpg |
|
0.432203 | 0433ab57b1f04a8ebe97430abf8568d1 | Grassmann | 2012.13975 | grassmann1.jpg |
0.482373 | d6a3324d21b049f4a4817c805f003929 | HDP | 2012.13975 | hdp1n.jpg |
0.486781 | 7a6ffab907ab40f5af00a6b40ec3bf15 | Protocol I | 2012.13975 | openmic1.jpg |
0.409297 | 73d2b39e7a274a1bb6cdfc9c95b6e46e | Protocol II | 2012.13975 | openmic2.jpg |
0.459933 | 37f050647db54b4185c852daf21395a6 | 2012.13975 | poolunc.jpg |
|
0.557088 | 449ad392b69c4fff8f8843deb48e7ed4 | 2012.13975 | pow1n.jpg |
|
0.47062 | 5782a514a0aa4163850ecc422553eb82 | 2012.13975 | pow2n.jpg |
|
0.470327 | 44ceceb739da456cbb41f978d6f8ee15 | 2012.13975 | pow3n.jpg |
|
0.43089 | 357c8c2d514b48c8827b9ea2d0aba59b | 2012.13975 | pow4n.jpg |
|
0.442192 | a1e186b2ccc247399ad3766eb93271cc | Our end-to-end pipeline. We pass an image (or patches) to CNN and extract feature vectors from its last conv. layer and augment them by encoded spatial coordinates . We perform pooling on second-order matrix by the Power Normalization function . | 2012.13975 | principle.jpg |
0.492312 | 7110a67677664fc19354ce42706a884d | 2012.13975 | principleb3.jpg |
|
0.481848 | 2ca37b7df26b4412bb028f193d68859c | 2012.13975 | principled.jpg |
|
0.416175 | c4571c1d5e5644aba311202c6e6e66c3 | 2012.13975 | sosn2b.jpg |
|
0.45201 | c9e2448252f84584912124cc2f2ff6b5 | 2012.13975 | timing1.jpg |
|
0.442873 | 08e73e623e614c8fb31e7ba32ea1b5cc | 2012.13975 | timing2b.jpg |
|
0.4212 | 69fde36d76ce4fd6983e1a906dc430ba | 2012.13975 | timing3b.jpg |
|
0.444566 | 2db0238783b84116894581d1e6f67868 | 2012.13975 | upperbound3n.jpg |
|
0.535701 | 9fbc65520f974796ab99370ebbe8a200 | 2012.13975 | var1.jpg |
|
0.498473 | fad1fc2f7fb740968f5bce7a3a122b00 | 2012.13975 | y_eta.jpg |
|
0.485217 | 08f666fd1ab84f53b93c6002591a6077 | Validation accuracy on abbreviation disambiguation. `SA' stands for self attention layer. | 2012.13978 | abb_accmay29.jpg |
0.463482 | 40cbe54d7426420daab5470a1924cc9b | Abbreviation count distribution | 2012.13978 | abb_count_hist_cr.jpg |
0.47623 | b55d40646c834ec08818fcb000a6bee4 | Top-10 recall on diagnosis prediction validation set. `SA' stands for self attention layer. `max' represents max-pooling output layer. `(s)' and `(p)' indicates whether the model is trained from scratch or pre-trained, respectively. | 2012.13978 | diagnosis_top_10_line.jpg |
0.457292 | a30777d32eed40e6bc636fae07017e88 | Top-30 recall on diagnosis prediction validation set. `SA' stands for self attention layer. `max' represents max-pooling output layer. `(s)' and `(p)' indicates whether the model is trained from scratch or pre-trained, respectively. | 2012.13978 | diagnosis_top_30_line.jpg |
0.508396 | 6a75740301e14ec396db6a47b8bfe303 | Top-5 recall on diagnosis prediction validation set. `SA' stands for self attention layer. `max' represents max-pooling output layer. `(s)' and `(p)' indicates whether the model is trained from scratch or pre-trained, respectively. | 2012.13978 | diagnosis_top_5_line.jpg |
0.473899 | 8aaf7442c1004e07a8952ed7742e9691 | Diagram of using MeDAL for pre-training NLU models in medical domain. | 2012.13978 | diagram-June1.jpg |
0.413897 | 491a0a7d6dbb403c8308a8a1f18b8a06 | Attention output layer for mortality and diagnosis prediction. | 2012.13978 | downstream-output.jpg |
0.481414 | 41ff78809d9749dc92d8e7625fe8681d | A sample in the MeDAL dataset. | 2012.13978 | rs_illustration.jpg |
0.493364 | 59557c258dbf42e08647d5358a6b63fc | Word count distribution | 2012.13978 | word_count_hist_cr.jpg |
0.424244 | 6e7c9b40cea64ff39d7366198742b25a | Semantic network map comprising all codes from 35 interviews at the level of p < 0.01. | 2012.13981 | Network_Map_p01.jpg |
0.463968 | 19b7667489dc4999b807e1fd7a1477be | The front cover of the SPS Observer features the DEEP student showing his demonstration at Texas A&M Physics & Engineering Festival <cit.>. Reproduced with permission from the SPS. Photo by Igor Kraguljac. | 2012.13981 | SPS_Observer_spring_small.jpg |
0.453476 | c82b5d2336514fe99c256c6e54565033 | Frequency of soft (transdisciplinary) skill codes for all interview participants. | 2012.13981 | Survey_Frequency.jpg |
0.401021 | efe3fa87715347528fe801abaa1df410 | Survey responses from all students on the impact of participating in outreach programs on recognizing connections between topics, their depth of conceptual physics understanding, networking within the department, and development of teamwork skills. | 2012.13981 | Survey_Results.jpg |
0.450917 | 9fc5fecf4ffc459199076b6a828a4aee | Schematic of the program principles and their associated informal physics programs at Texas A&M. | 2012.13981 | TAMU_OutreachPrograms.jpg |
0.465385 | 7cdf81b1402f49cc8a0d8eaa35aff544 | DEEP team members with their superconducting train track. This demonstration experiment was part of Nobel Laureate David Lee's public talk at the 2013 Physics & Engineering Festival. Photo by Natasha Sheffield. | 2012.13981 | superconductivity_small.jpg |
0.442065 | 2e31bd0d508a4ea0a8152a57e973d5bd | Scaling factor α controls NN behavior: NTK versus MF | 2012.13982 | ntk-mf.jpg |
0.474598 | 31c5b555ae594911ae372592f7e8be55 | Solution neighborhood size versus NN width m | 2012.13982 | ntk.jpg |
0.438359 | b8883bf624a84eb7bc033b911f336c49 | Feynman diagrams for the abelian theory on a cylinder in holomorphic-to-holomorphic polarization. | 2012.13983 | hol-hol.jpg |
0.509964 | 9c9b770a41a04a48ae5d6620f2b88ae3 | Distribution of the number of 132 recoils produced (i.e., the number of scatterings suffered) by monoenergetic neutrons of various energies. | 2012.13986 | Fig_NRProb_Mono.jpg |
0.47141 | d8652d950dbc4d43b156494c2428da36 | 132 recoil spectra generated by monoenergetic neutrons of different energies. | 2012.13986 | Fig_NRSpec_Mono.jpg |
0.449938 | 9a9e87e40528436784aead9a3e078394 | 132 differential recoil spectrum due to supernova neutrino induced (curve marked “") and the spectrum due to neutrons produced by inelastic CC interaction of supernova s with 132 nuclei (curve marked “") for the case of normal ordering of neutrino mass hierarchy shown in Fig. <ref>. Both the curves correspond to SN neutrino flux given by the Basel-Darmstadt simulations <cit.> of a 18 supernova progenitor at a distance of 1 kpc. | 2012.13986 | Fig_Xe132_NRspects_cenns_nuin_NO.jpg |
0.424243 | 8d778ab20f68420683bcf9a65491fe18 | Energy spectrum of electrons produced due to CC interaction of supernova s with 132 nuclei, for the case of normal ordering (NO) of neutrino mass hierarchy, for SN neutrino flux given by the Basel-Darmstadt simulations <cit.> of a 18 progenitor supernova at a distance of 1 kpc. | 2012.13986 | Fig_espect_NO.jpg |
0.443139 | 0779dc731574480eb25c3b81cfc63028 | Energy spectrum of deexcitation γ-rays emitted by excited 132 nuclei produced due to inelastic CC interaction of supernova s with 132 nuclei, for the case of normal ordering (NO) of neutrino mass hierarchy, for SN neutrino flux given by the Basel-Darmstadt simulations <cit.> of a 18 progenitor supernova at a distance of 1 kpc. | 2012.13986 | Fig_gspect_NO.jpg |
0.39368 | d0cc75ec1714432abf80cb17aaaaf196 | Energy spectrum of neutrons emitted by excited 132 nuclei produced due to inelastic CC interaction of supernova s with 132 nuclei, for the case of normal ordering (NO) of neutrino mass hierarchy, for SN neutrino flux given by the Basel-Darmstadt simulations <cit.> of a 18 progenitor supernova at a distance of 1 kpc. | 2012.13986 | Fig_nspect_NO.jpg |
0.479543 | e170462f804b498585c0ce547e330718 | Total cross section (solid curve) and neutron emission cross section (dashed curve) for the CC interaction of with 132 as functions of the incoming neutrino energy. | 2012.13986 | Fig_total-and-n-xsecs_linear.jpg |
0.50496 | cf33cfb46cf149adb5fcd6b8cacbb34d | Graph of the interactions between layers. | 2012.13987 | DBM_int.jpg |
0.42281 | 4974bf6c0cf745f290b46ab0fe158b3b | How to count the number of RTs and Likes: the RT count and Like count at the time of the last retweet in the aggregation period are the latest RT and Like counts in the dataset period. | 2012.13990 | decide_rt_like.jpg |
0.472188 | cca0a16404924685bb1e0da8a67149fa | Example of reference tweets: The tweets have URLs included in the news dataset. A single tweet may refer to multiple news articles. | 2012.13990 | reference_tweet.jpg |
0.408377 | d800b7420f5a46b180b5859cca16adec | Energy expectation values E_n(R) vs. R for n=1 and n=2, obtained from a fit to a single exponential. | 2012.13991 | ER12.jpg |
0.444104 | f535722a91b24a93999594bbeb01137e | Energy expectation values E_n(R) vs. R for n=1, 2, 3 together with the one photon threshold. The energy E_3 is obtained from a fit to two exponentials, as explained in the text. | 2012.13991 | ER123.jpg |
0.51539 | 367a50afea774e03be3923c266c94252 | Same as Fig. <ref>, except E_2 is obtained from fits to data points in the range T=3-6, rather than T=2-5. | 2012.13991 | ER12_3to6.jpg |
0.438808 | c7b35c4b14e24e78852c04a461cf7ef0 | Energy expectation values E_1, E_2 vs. R at R>3, obtained on a 12^4 lattice. | 2012.13991 | ER12v12.jpg |
0.441263 | 66ceacb9938141b98f9e9d89a6a19fc5 | Comparison of E_1,E_2 obtained using n_ev=2 and n_ev=4 Laplacian eigenstates. | 2012.13991 | compare.jpg |
0.444248 | a4d26b149f9d4fd89f262f7105d2b2f4 | _nn(R,T) vs. T for n=1,2 at fixed R=6.93 on a 16^4 lattice | 2012.13991 | fit12.jpg |
0.438468 | 5862768cf1194172a8b3efbc8db6d99f | T_22(R) vs. T at R=3.16. The reason for a discrepancy between a fit of data points at T=2-5, and T=3-6 is apparent. | 2012.13991 | fit12_n10.jpg |
0.417557 | 7d99ecf6d5c3480db837d3be43f646ea | Same as Fig. <ref>, but on a 12^4 lattice. | 2012.13991 | fit12v12.jpg |
0.395704 | cffbba268a6d4849957d303b78778168 | _33(R,T) vs. T at fixed R=6.93. The fit shown is to the sum of exponentials in eq. T33fit. | 2012.13991 | fit3.jpg |
0.438165 | 62ba186ff7474abfa63f357e9d1b223e | The plaquette-plaquette correlator computed for =̱3, =0.5. The photon mass is obtained from the slope of the line shown. | 2012.13991 | photon.jpg |
0.409584 | e304f1e722ad4cbab141d1d32b8488e6 | Phase diagram of the q=2 abelian Higgs model computed in <cit.> ( “conf” denotes the confinment phase). | 2012.13991 | thermo.jpg |
0.424595 | c2ab89b8dea74dd2a85f56ba49284cc2 | Distribution of the article vector and the article height. The article height tends to be long at the top (indicated by the circle). | 2012.13992 | article_vector_article_length_edit.jpg |
0.443517 | 68fba7e59e064d8bbacdbce632302f4b | Distribution of dwell time in one browsing session. The dwell time peaks near 0.2s and 2s. | 2012.13992 | dwelltime.jpg |
0.402731 | 7b7b43be4ed0483a84f40961bcb610ea | The user vector for the Nov-19 dataset (k-means, k=4). Unlike in Fig. <ref>, it is divided into only three directions. | 2012.13992 | user_vector.jpg |
0.457179 | 3c5f41b73b774e74ab8600fcc9bad5f6 | The user vector for the Nov-19 dataset (k-means, k=4). Unlike in Fig. <ref>, it is divided into only three directions. | 2012.13992 | user_vector.jpg |
0.410857 | 8e3a418dcf3c44eebdc96b3c68532c73 | Distribution of the user vector and dwell time in Nov-2020. | 2012.13992 | user_vector_dwelltime.jpg |
0.439014 | 7f28197bbdc44c3689ac3c860f8c6356 | Distribution of the user vector and dwell time. The dwell time tends to be short in the left and top parts (indicated by the red dashed circle) and long in the right and bottom parts (indicated by the green circle). | 2012.13992 | user_vector_dwelltime_edit.jpg |
0.388579 | 49bc3e0cc5994b2f8210a682da32224a | Long to short distance cumulative histogram. The x-axis indicates the distance. The average is 0.2 and the median is 0.18. | 2012.13992 | user_vector_long_to_short_distance_distribution_cumulative.jpg |
0.422349 | 42ae426c846d467191dc5270075f39ce | Changes in the user vector when browsing sessions with a dwell time of 5s or less are excluded. Arrows show user vectors after excluding clicks with short dwell time. 77.85% of user vectors moved to the right. | 2012.13992 | user_vector_move_long.jpg |
0.53041 | 0b9da57efe0d45aea7cc9e635c0b3c13 | Plot of f_ρ(α, θ) (full blue curve) and of f_σ(α, θ) (full red curve) as a function of θ (in units of π) for fixed α=1.1. Vertical dashed lines are placed at θ=1/2arccos(α-1/α+1)≈ 0.242 π ("moderately interacting" regime in blue) and θ=π/4 ("strongly interacting" regime in gray) as references. | 2012.14223 | Fig1.jpg |
0.435425 | da5a257dcda2439ba823a17774a48368 | Model for a QH edge state at filling factor ν=2. According to the chirality one can identify the incoming (injection) region (1), the interacting region (2) (shaded area of length L) and the outgoing (detection) region (3). In regions (1) and (3) the dynamics of the bosonic fields is well described in terms of free equations of motion (u=0). Moreover, the outgoing fields, written in the Fourier space (ϕ̃_i,out(x,ω), with i=1,2), are connected to the incoming ones (ϕ̃_i,in(x,ω), with i=1,2) through the edge-magnetoplasmon scattering matrix Ŝ(L, ω) which encodes the information of the inter-channel interaction acting over a length L and at a given frequency (energy) ω. | 2012.14223 | Fig2.jpg |
0.526684 | b4281355e89d41149046f01e6dba129d | Relative peak height as a function of the injection energy (measured in μeV) for two different sample lengths: L=0.75 μm (left panel) and L=0.48 μm (right panel). The non-dissipative case (green dash-dotted curve); the linear dissipative case (blue full curve) with γ_1=0.13 for the left panel andγ_1=0.43 for the right panel; quadratic dissipation with γ_2ε_0=0.03 for the both panel (red dotted curve) and with γ_2ε_0=0.13 for the left panel and γ_2ε_0=0.23 for the right one (bronze dashed curve). Other parameters are: α=2.1, θ=0.17π (left panel) and α=1.6, θ=0.16π (right panel). Light blue diamonds indicate the experimental data taken from Ref. <cit.>. | 2012.14223 | Fig3.jpg |
0.493904 | 90d8adc6155c41f3be8bff5aa6072dfa | Absolute value of various terms of Eq. (<ref>) as α varies (see legend for labeling of the terms). All terms except the one containing δ_7 are independent of α and therefore horizontal in this plot. See text for the vertical lines. Q=Q_eq is used and other parameters are the same as in Fig. <ref>. | 2012.14226 | Test3.jpg |
0.473612 | bfdc847a81b64711a9c896bdc993e1a9 | Dependence of δ_g, δ_c and δ on 1-v. The vertical lines from left to right correspond to the proton energy of 10^20, 10^19, 10^18, 10^15 [eV] respectively. Q=100Q_mag is used and other parameters are the same as in Fig. <ref>. | 2012.14226 | Test4.jpg |
0.527484 | b0f7b3bb619f41029239baa839cb4864 | A region 𝒟⊂(ℳ,J_ij). Notice that β_S=π-Ψ_S and β_O=Ψ_O. The radial and angular coordinates of O and S are (r_O,ϕ_O) and (r_S,ϕ_S). | 2012.14226 | figure1.jpg |
0.495544 | 2a17eafe8298441b8368be6296ab94d4 | The deflection of charged signal by the charged lens L from source S to observer O. The gravitational interaction always attracts the signal particle (dot-dashed line) while the electrostatic interaction might attract or repel (dotted line) the CMP depending on the sign of Q and q. The true trajectory is the combined one plotted using solid curve. | 2012.14226 | lensing_fig_zh_v2.jpg |
0.451885 | 9445dd826ca345c48a46a0cc84bee16a | Backscattered signal segmentation and signature extraction. The red segment represents a backscattered signal, and the green bars indicate extracted backscatter signatures. | 2012.14227 | Backscattered_signal_extraction.jpg |
0.45709 | 3af5a66ebf1043e6a7111ed6ab461224 | Floor plans of experimental environments. We evaluate our system in the indoor and outdoor environments. | 2012.14227 | Environment_setting.jpg |
0.452761 | 174ab4c08531454e9abee747fa79dedb | Experimental platform. We implement our system using USRP nodes, backscatter tags and iRobot Create robots. All tags are controlled by an Altera STEP-MAX10 FPGA. | 2012.14227 | Experimental_platform.jpg |
0.505665 | fd9176617afe406bb38bd51d0ddb47db | Overall detection accuracy and testing time for each sample using different machine learning models. | 2012.14227 | ML_performance.jpg |
0.509582 | b3a0ba3909ab4c4daf2b8b859de7e108 | Impact of the sort operation and number of decision trees. The dashed vertical line represents the number of trees used in our RF model. | 2012.14227 | RF_performance.jpg |
0.474526 | c925d2675ad64db99b66519eb0f5e527 | ROC curve of our system. Its AUROC reaches to 0.987 and the reference line represents the random guessing model. | 2012.14227 | ROC_curve.jpg |
0.436557 | 03477f3b209547b1badbeaf222155530 | System flow of ScatterID. It contains three core components, i.e., Signal Profile Construction, Profile Similarity Measurement and Fake Robot Detection. | 2012.14227 | System_overview.jpg |
0.416004 | 6820908de9da4b2dbc2eb3da0258ea41 | Backscattered signal traces reflected by two tags from one legitimate and two fake robots. | 2012.14227 | Traces.jpg |
0.385795 | e6e8ccf2bc7a4b87ab6d0a0e02aaab93 | Illustration of power-scaling attacks and Cosine distance. After power-scaling attacks, Cosine distance remains the same, but Euclidean distance is distorted. | 2012.14227 | cosine_distance.jpg |
0.444665 | ece80d8aac9d42f79793ba7bde0ce092 | CDFs of positive and negative samples using Cosine, Euclidean, Chebyshev and Manhattan distances. | 2012.14227 | distance_cdf.jpg |
0.459853 | f31d94c31ff541d1b64be50568a1f4ae | Performance using different distance metrics. Ps stands for power-scaling attacks. We only annotate TPRs and FPRs of Cosine and Euclidean distance metrics for brevity. | 2012.14227 | distance_performance.jpg |
0.420659 | 74454bb8f15b49beae691735b323dfd1 | Illustration of Sybil attacks in a robotic network. Robotic attackers can launch various types of Sybil attacks. | 2012.14227 | observation.jpg |
0.532812 | 166f9f85d57e48e5b2c9d6ed44c92f7b | Performance of our and baseline models under various Sybil attacks. Ps stands for power-scaling attacks. | 2012.14227 | ours_baseline_performance_comparison.jpg |
0.52173 | 475967d54d05462d9d5eac49ef4d493c | System performance in different environments. | 2012.14227 | overall_performance.jpg |
0.443066 | 3782ea6673b44211adaae1177eafc45f | Work flow of the profile similarity measurement. N is the number of surrounding robots, K the number of tags and L the number of multipath signatures in a signal profile. | 2012.14227 | profile_similarity_measurement.jpg |
0.440088 | 15f936367bfc4d4097f93939ede130f1 | AUROCs in terms of different profile sizes. The red mark corresponds to the selected size with K=4 and L=10. | 2012.14227 | profile_size.jpg |
0.462463 | 13b7992995244a24b48c98a97f80ad01 | Our RF model for fake robot detection. "0" stands for a real robot and "1" for a fake one. | 2012.14227 | random_forest.jpg |
0.490934 | 069f638753e04c32b5c6733a5de3c693 | CRM-CWMs: 0.025±0.01 | 2012.14228 | crm_slot2_k0_vis9_plot0_017001.jpg |
0.475613 | 1f454a6619ab421cbfb8c65a8760252d | CRM-CWMs: 0.033±0.005 | 2012.14228 | crm_slot3_k1_vis9_plot0_028403.jpg |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.