Structural, Syntactic, and Statistical Pattern Recognition, Joint IAPRs, SSPR 2016

Preface......Page 6
Organization......Page 7
Contents......Page 9
Dimensionality Reduction, Manifold Learning and Embedding Methods......Page 14
 1 Introduction......Page 15
 2 Literature Review......Page 17
  3.2 Jensen-Shannon Divergence......Page 19
  4.1 Construction of Graph-Based Features......Page 20
  5.1 The Most Influential Features for Credit Risks (Continuous Target Features)......Page 21
  5.2 Classification for the Credit Rating (Discrete Target Features)......Page 23
 6 Conclusion......Page 24
 References......Page 25
 1 Introduction......Page 27
  2.1 Objective Function......Page 28
  2.2 Embedding Test Instances......Page 31
 3 Related Work......Page 32
  4.1 Setting......Page 33
  4.2 Results......Page 35
 References......Page 36
 1 Introduction......Page 38
 2 Stiefel Manifolds and Least Squares......Page 39
  3.1 Optimisation and Initialisation......Page 41
  3.2 Classification......Page 42
  3.3 Relation to Neural Networks and Support Vector Machines......Page 43
  4.1 Synthetic Data......Page 44
  4.2 Network Behaviour Classification......Page 45
 5 Conclusions......Page 46
 References......Page 47
 1 Introduction......Page 49
 2 Pattern Recognition in Vector Space......Page 50
 3 Pattern Recognition in Multi-linear Forms......Page 52
 4 Mutual Tensor Subspace Method......Page 56
 References......Page 59
 1 Introduction......Page 61
 2 Quantum Thermodynamics of the Network......Page 62
  3.1 Datasets......Page 66
  3.2 Experiments......Page 67
 References......Page 70
Dissimilarity Representations......Page 72
 1 Introduction......Page 73
 2 Related Work......Page 74
 3 Proposed Approach......Page 76
  4.1 Experimental Setting......Page 78
  4.2 Experimental Results......Page 79
 5 Conclusion......Page 81
 References......Page 82
 1 Introduction......Page 84
  2.2 Non Linear Scaling......Page 86
 3 The Proposed Criterion......Page 87
 4 Experimental Results......Page 88
 5 Conclusions......Page 92
 References......Page 93
 1 Introduction......Page 94
 2 Dissimilarity Matrix Representation......Page 95
 3 Illustrative Examples......Page 96
  4.1 Supervised Experiments: Multiple Instance Learning......Page 99
  4.2 Unsupervised Experiments: Analysis of Teams of People......Page 100
 5 Conclusions......Page 102
 References......Page 103
 1 Introduction......Page 105
 2 Related Work......Page 106
  3.1 Continuous Vector-Space Representation of Words......Page 107
  3.3 DS Using Vector Space Representations of Sentences......Page 108
 4 Cross-Entropy Difference Method......Page 109
  5.1 Experimental Setup......Page 110
  5.2 Comparative with Cross-Entropy Selection......Page 111
  5.3 Comparative with Bilingual Cross-Entropy Selection......Page 112
  5.4 Summary of the Results......Page 113
 6 Conclusion and Future Work......Page 114
 References......Page 115
 1 Introduction......Page 117
 2 Prototype Embedding Based Generalized Median Computation......Page 118
  2.2 Computation of Geometric Median by Weiszfeld Algorithm......Page 119
  2.3 Reconstruction Methods......Page 120
  3.1 Linear Recursive......Page 121
 4 Experimental Evaluation......Page 122
  4.2 Quality of Computed Generalized Median......Page 123
  4.3 Computational Time......Page 125
 References......Page 126
Graph-Theoretic Methods......Page 128
 1 Introduction......Page 129
  2.1 Graph and Neighbourhood......Page 130
  2.2 Graph Edit Distance......Page 131
 3 Distance Models for Neighbourhoods......Page 133
  3.1 The Cost of Processing Neighbourhoods......Page 134
 4 Experimental Evaluation......Page 135
 References......Page 137
 1 Introduction......Page 140
 2 Related Work......Page 141
 3 Grid Mining......Page 142
 4 Experiments......Page 144
 5 Conclusion......Page 148
 References......Page 149
 1 Introduction......Page 151
  2.1 Quantum States and Von Neumann Entropy......Page 152
  2.3 Holevo Quantity of a Graph Decomposition......Page 153
  3.1 Relation with Degree Centrality......Page 154
  4.2 Edge Centrality in Real-World Networks......Page 156
  4.3 Robustness Analysis......Page 158
 5 Conclusion......Page 159
 References......Page 160
 1 Introduction......Page 161
  2.1 Density Matrix......Page 163
 3 Quantum Statistics in Networks......Page 164
  3.2 Bose-Einstein Statistics......Page 165
  3.3 Fermi-Dirac Statistics......Page 166
  4.1 Numerical Results......Page 167
  4.2 Experimental Results......Page 168
 References......Page 170
 1 Introduction......Page 171
  1.1 Related Literature......Page 172
  2.1 Mean Reversion Autoregressive Model......Page 173
  2.2 Generative Model Learning......Page 174
 3 Experiments......Page 177
 References......Page 180
 1 Introduction......Page 182
 2 Graph Entropy and Information Functionals......Page 183
 3 Substructure Based Approach for Graph Entropy......Page 184
  4.1 Random Graphs......Page 187
  4.3 Time-Varying Networks......Page 189
 References......Page 191
  1.1 Motivation......Page 193
 2 Return Random Walks......Page 195
  3.1 The Line Graph......Page 196
  3.2 The Dirichlet Functional for the Line Graph......Page 197
 4 Experiments and Conclusions......Page 199
 References......Page 202
 1 Introduction......Page 204
  1.2 Outline......Page 205
  2.1 Jensen-Shannon Divergence......Page 206
  2.2 von Neumann Entropy for Directed Graphs......Page 207
  2.3 Jensen-Shannon Divergence Kernel for Directed Graphs......Page 209
  3.1 Datasets......Page 210
  3.2 Graph Classification......Page 211
 4 Conclusion......Page 213
 References......Page 214
 1 Introduction......Page 215
 2 Neighborhood Graphs......Page 216
 3 X-Nearest Neighborh Graph......Page 217
  3.1 Full XNN......Page 218
  3.2 Properties......Page 219
  3.3 Hierarchical XNN......Page 220
 4 Applications......Page 222
 5 Conclusion......Page 224
 References......Page 225
 1 Introduction......Page 226
  2.3 von Neumann Entropy......Page 228
  3.1 Thermal Quantities......Page 229
  3.2 Bose-Einstein Statistics......Page 230
  4.1 Jensen-Shannon Divergence......Page 231
  5.1 Dataset......Page 232
  5.2 Experimental Results......Page 233
  5.3 Discussion......Page 234
 6 Conclusion......Page 235
 References......Page 236
Model Selection, Classification and Clustering......Page 237
 1 Introduction......Page 238
  2.1 Tensor Representation for N-way Arrays......Page 239
  2.3 Principal Component Analysis for Third-Order Tensors......Page 240
  2.4 Reduction by Three-Dimensional Discrete Cosine Transform......Page 242
 4 Numerical Examples......Page 243
 5 Conclusions......Page 246
 References......Page 247
  1.1 Motivation......Page 248
  2.1 Combinatorial Formulation......Page 250
  2.2 Spectral Formulation......Page 253
 3 Discussion and Conclusions......Page 256
 References......Page 257
 1 Introduction......Page 259
  2.1 General Distance K-means......Page 260
  2.2 Density Sensitive Geodesic Distance......Page 261
  3.1 Distance Metric Based on SNN......Page 262
  3.2 Geodesic Based Local Outlier Factor......Page 263
  3.3 Weighting Transform Model......Page 264
  4.2 Results and Discussion......Page 265
 5 Conclusion......Page 268
 References......Page 269
 1 Introduction......Page 270
  2.1 Models......Page 271
 3 Variational Inference for Model Selection......Page 273
  3.1 Variational Inference and Proposed Optimization......Page 274
  3.2 Discussion......Page 275
  4.1 Artificial Data......Page 276
  4.2 Real Data......Page 278
 References......Page 279
 1 Introduction......Page 281
  2.1 Multiple Structure Recovery......Page 283
  2.2 Clustering......Page 284
 3 Biclustering......Page 285
  3.1 FABIA......Page 286
  4.1 Adelaide Dataset......Page 287
  4.2 Reproducibility......Page 288
 5 Conclusion and Discussion......Page 289
 References......Page 290
 1 Introduction......Page 292
 2 External Cluster Validity......Page 294
  3.1 Prototype vs. Partition Similarity......Page 297
  3.2 Model Similarity......Page 298
  3.3 Arbitrary Partitions......Page 299
 4 Experiments......Page 300
 References......Page 302
Semi and Fully Supervised Learning Methods......Page 304
 1 Introduction......Page 305
 2 Supervised Peaking......Page 307
 3 Semi-supervised Classifier......Page 309
  4.2 Simulation of Contributions......Page 310
 5 Convergence to a Better Solution than the Base Learner?......Page 311
 7 Discussion and Conclusion......Page 313
 References......Page 314
 1 Introduction......Page 316
 2 Learning Gait Features......Page 317
  3.1 Database......Page 321
  3.2 Evaluation Set-Ups and Metrics......Page 322
  3.3 Results......Page 323
 4 Conclusions......Page 325
 References......Page 326
 1 Introduction......Page 328
  2.1 Attacker\'s Model......Page 329
  2.2 Robustness and Regularization......Page 330
 4 Experimental Analysis......Page 332
 References......Page 337
 1 Introduction......Page 339
 2 Related Works......Page 340
 3 Proposed Approach......Page 341
 4 Experimental Results......Page 344
  4.1 Segmentation Results with Different Training Strategies......Page 345
  4.2 Detailed Analysis on Correlations Across Tasks......Page 347
 References......Page 348
 1 Introduction and Related Work......Page 350
  2.1 Average Precision......Page 351
  2.2 Sequential Labeling......Page 352
 3 Training and Testing Sequential Labeling with the AP Loss......Page 353
  3.1 Inference and Loss-Augmented Inference......Page 354
  4.1 Results on the TUM Kitchen Dataset......Page 355
  4.2 Results on the CMU Multimodal Activity Dataset......Page 358
 References......Page 359
Shape Analysis......Page 361
 1 Introduction......Page 362
  2.1 Textures and Graphometric Features......Page 364
  2.2 GLCM and Haralick’s Descriptors......Page 365
  2.3 Fractal Geometry......Page 366
  3.1 Acquisition and Preparation of Manuscript Database......Page 367
  3.4 Rule of Decision......Page 369
 4 Results......Page 370
 References......Page 371
 1 Introduction......Page 373
 2 State of the Art......Page 374
  3.2 Usual Selection Mechanisms......Page 376
  3.3 Analysis of Information Distribution......Page 378
 4 Experiments......Page 379
 References......Page 382
 1 Introduction......Page 384
 2 Related Work......Page 385
 3 Learning Markov--Gibbs Random Fields......Page 386
  3.1 Model Nesting and Sampling......Page 387
  3.3 Filter Pre-training......Page 388
 4 Experimental Results......Page 389
 5 Conclusions......Page 392
 References......Page 393
 1 Introduction......Page 395
 3 Shape Normalization......Page 397
  4.1 Representation of Worms......Page 398
  4.3 Handling Occlusions by Re-Identification......Page 399
  4.4 Trajectory Analysis for Head and Tail Definition......Page 401
  5.2 Tracking......Page 402
 References......Page 404
 1 Introduction......Page 406
 2 Proposed Method......Page 408
  2.2 Extraction Entity in Contiguous Structure......Page 409
  2.3 Extraction Entity in Noncontiguous Structure......Page 410
  3.1 Dataset......Page 413
  3.3 Structure Correction Evaluation......Page 414
 References......Page 416
 1 Introduction......Page 417
 2 Related Work......Page 418
 3 Thickness-Based Ellipse Detection (TED)......Page 419
  3.2 Shape Flattening......Page 420
  3.4 Ellipse Parameter Estimation......Page 421
  3.5 Complexity of the Algorithms......Page 423
 5 Experiments and Discussion......Page 424
 6 Conclusion and Future Work......Page 426
 References......Page 427
Spatio-temporal Pattern Recognition......Page 429
 1 Introduction......Page 430
 2 Related Work......Page 431
  3.1 Classical Convolutional Autoencoders......Page 433
  3.2 Interpretable Pattern Discovery with Autoencoders......Page 434
  4.2 Results on the Synthetic Dataset......Page 436
 5 Conclusion......Page 439
 References......Page 440
 1 Introduction......Page 442
  2.2 Map Handling......Page 444
  2.4 Map Projection and the Grid......Page 445
 3 Route Search......Page 446
 4 Experiments......Page 448
  4.1 Efficiency of the Search......Page 449
  4.2 Usability Evaluation......Page 450
 References......Page 452
 1 Introduction......Page 453
  2.2 Spatio-Temporal Similarity......Page 454
  3.1 Short-Term GPS Dynamics......Page 456
  3.2 Gaussian Mixture Model Classifier......Page 457
  4.1 Data Sets......Page 458
  4.2 Evaluation......Page 460
 5 Results......Page 461
 References......Page 462
Structural Matching......Page 464
 1 Introduction......Page 465
 3 The Proposed Approach......Page 467
  3.1 Computing the Enrichment Vectors......Page 469
 4 Experimental Evaluation......Page 470
  4.1 Results and Discussion......Page 471
 References......Page 474
 1 Introduction......Page 476
  2.1 Continuous-Time Quantum Walks......Page 477
  2.2 Average Mixing Matrix......Page 478
  2.3 Properties of the Average Mixing Matrix......Page 479
 3 The Average Mixing Matrix Signature......Page 480
  4.1 HKS and WKS......Page 481
  4.4 Sensitivity to Parameters......Page 482
  4.5 Robustness to Erdős-Rényi Structural Noise......Page 483
 5 Conclusion......Page 484
 References......Page 485
 1 Introduction......Page 487
 2 Definitions......Page 488
 3 BLP Formulation of GED......Page 489
  3.1 BLP-GED Variables......Page 490
  3.3 Constraints......Page 491
 4 Experiments......Page 493
 5 Conclusion......Page 495
 References......Page 496
 1 Introduction......Page 498
 2 GED as a Quadratic Assignment Problem......Page 499
 3 Resolution of the QAP......Page 503
 4 Experiments......Page 506
 References......Page 507
 1 Introduction......Page 509
 2 Basic Definitions......Page 510
 3 Minimisation Method......Page 511
 4 Evolutionary Method......Page 514
  4.2 Weighted Means of Pairs of Correspondences......Page 515
 5 Experimental Validation......Page 516
 6 Conclusions and Future Work......Page 518
 References......Page 519
 1 Introduction......Page 521
  2.1 General Structure......Page 523
   2.2.2 Palmprint......Page 525
   2.2.3 Letters......Page 526
   2.2.5 House-Hotel......Page 527
  2.3 Repository Summary......Page 528
 3 Conclusions......Page 529
 References......Page 530
 1 Introduction......Page 532
 2 Graphs and Graph Edit Distance......Page 533
 3 Restrictions on the Graph Edit Distance......Page 535
 4 Defining the Graph Edit Distance as a True Distance......Page 537
 5 Experimental Validation......Page 538
 6 Conclusions......Page 540
 References......Page 541
Text and Document Analysis......Page 543
 1 Introduction......Page 544
 2 A CNN Based Word Image Categorization Method......Page 546
  2.1 Convolutional Neural Networks......Page 547
  2.2 Fully Connected Layers......Page 548
  3.1 Esposalles Dataset......Page 549
  3.2 Experiments and Results......Page 550
 References......Page 552
 1 Introduction......Page 554
 2 Graph-Based Representation of Word Images......Page 555
  2.1 Graph Extraction Based on Keypoints......Page 556
  2.2 Graph Extraction Based on a Segmentation Grid......Page 557
  2.3 Graph Extraction Based on Projection Profiles......Page 558
  2.4 Graph Extraction Based on Splittings......Page 559
 3 Experimental Evaluation......Page 561
 4 Conclusion and Outlook......Page 562
 References......Page 563
 1 Introduction......Page 565
 2 Graph Edit Distance......Page 567
  3.1 Image Preprocessing......Page 568
  3.2 Graph Representation......Page 569
  3.4 Pairwise Matching......Page 570
 4 Experimental Evaluation......Page 571
 References......Page 573
 1 Introduction......Page 575
  2.1 Taxonomy of Methods......Page 576
  3.1 End-to-end Pipeline Overview......Page 578
  3.2 Local Binary Patterns......Page 579
  3.3 Spatial Sampling......Page 580
  4.1 Experimental Framework......Page 581
  4.3 Results on BHHMD Dataset......Page 582
 References......Page 583
Learning Robust Features for Gait Recognition by Maximum Margin Criterion......Page 585
Author Index......Page 587