Similarity-Based Pattern Recognition: Third International Workshop, SIMBAD 2015, Copenhagen, Denmark, October 12-14, 2015. Proceedings (Image ... Vision, Pattern Recognition, and Graphics)

✍ Scribed by Aasa Feragen (editor), Marcello Pelillo (editor), Marco Loog (editor)

• Publisher: Springer
• Year: 2015
• Tongue: English
• Leaves: 238
• Edition: 1st ed. 2015
• Category: Library

✦ Synopsis

This book constitutes the proceedings of the Third International Workshop on Similarity Based Pattern Analysis and Recognition, SIMBAD 2015, which was held in Copenahgen, Denmark, in October 2015. The 15 full and 8 short papers presented were carefully reviewed and selected from 30 submissions.The workshop focus on problems, techniques, applications, and perspectives: from supervisedto unsupervised learning, from generative to discriminative models, and fromtheoretical issues to empirical validations.

✦ Table of Contents

Preface
Contents
A Novel Data Representation Based on Dissimilarity Increments
1 Introduction
2 Dissimilarity Representation
2.1 Dissimilarity Increments Space
3 Characterization of the Dissimilarity Spaces
4 Unsupervised Learning in Dissimilarity Spaces
5 Conclusions
References
Characterizing Multiple Instance Datasets
1 Introduction
2 Multiple Instance Learning
3 Datasets
3.1 Artificial Datasets
3.2 Real-Life Datasets
4 Proposed Approach
5 Experiments
6 Conclusions
References
Supervised Learning of Diffusion Distance to Improve Histogram Matching
1 Introduction
2 Proposed Method
2.1 Diffusion Distance for Histogram Comparison
2.2 Metric Learning
2.3 Learning Diffusion Distance
3 Experiments
3.1 Datasets
3.2 Methods Evaluated
3.3 Results
4 Conclusions
References
Similarity Analysis from Limiting Quantum Walks
1 Introduction
2 Quantum Walks for Analyzing Transport
2.1 Unitary Evolution and the Schrödinger Operator
2.2 Long-Time Averages from Magnitude
3 Edges from Quantum Walks
3.1 Local-to-Global Role of Quantum Walks
3.2 Filtering Texture
3.3 The Effect of Similarity Compression
4 Conclusions
References
Introducing Negative Evidence in Ensemble Clustering Application in Automatic ECG Analysis
1 Introduction
2 Clustering Ensembles
2.1 Generating the Clustering Ensembles
2.2 Combining the Data Partitions
2.3 Extracting the Final Data Partition
3 Application to Heartbeat Clustering
3.1 ECG Database
3.2 Preprocessing
3.3 Heartbeat Representation
4 Experimental Results
5 Discussion
6 Conclusions
References
Dissimilarity Representations for Low-Resolution Face Recognition
1 Introduction
2 Related Work
3 Proposed Approach: Reduced Dissimilarity Space
3.1 Dissimilarity Space and Prototype Selection
3.2 Considerations at Test Time
4 Experiments and Discussion
4.1 Databases Description
4.2 Experimental Setup
4.3 Results and Discussion
5 Conclusions
References
Deep Metric Learning Using Triplet Network
1 Introduction
2 The Triplet Network
2.1 Training
3 Tests and Results
3.1 Datasets
3.2 The Embedding Net
3.3 Results
3.4 2d Visualization of Features
3.5 Comparison with Performance of the Siamese Network
4 Future Work
5 Conclusions
References
Cluster Merging Based on Dominant Sets
1 Introduction
2 Dominant Sets
3 Cluster Merging
4 Experiments
5 Conclusion
References
An Adaptive Radial Basis Function Kernel for Support Vector Data Description
1 Introduction
2 Related Approaches
3 Support Vector Data Description
4 Proposed Method
5 Results
5.1 Artificial Dataset
5.2 Real Dataset
6 Conclusions
References
Robust Initialization for Learning Latent Dirichlet Allocation
1 Introduction
2 Background: Latent Dirichlet Allocation
3 The Proposed Approach
3.1 LDA Initialization by Pruning
3.2 LDA Initialization by Splitting
4 Experimental Evaluation
4.1 Log-Likelihood Evaluation
4.2 Coherence Evaluation
4.3 Science Magazine Dataset
5 Conclusions
References
Unsupervised Motion Segmentation Using Metric Embedding of Features
1 Introduction
2 Notations and Definitions
3 Optimizing Number of Segments
4 Application to Motion Segmentation
4.1 Adjusting Rigidity in Segmentation
4.2 Aggregating Motion Over Frames
5 Discussion and Future Work
References
Transitive Assignment Kernels for Structural Classification
1 Introduction
1.1 Graph Kernels
1.2 Assignment Kernels
1.3 Multi-graph Matching
1.4 Contribution
2 Projection on the Transitive Alignment Space
3 Transitive Assignment Kernel
4 Experimental Evaluation
5 Conclusion
References
Large Scale Indefinite Kernel Fisher Discriminant
1 Introduction
2 Indefinite Fisher and Kernel Quadratic Discriminant
3 Probabilistic Classification Vector Learning
4 Nyström Approximation
4.1 Pseudo Inverse, SVD and EVD of a Nyström Approximated Matrix
5 Nyström Based Indefinite Kernel Fisher Discriminant
6 Complexity Analysis
7 Experiments
8 Conclusions
References
Similarity-Based User Identification Across Social Networks
1 Introduction
2 Problem Description
3 Approach
3.1 Description of Profiles
3.2 Similarity Measures
3.3 Classification
4 Experimental Results
4.1 Data Collection
4.2 Experimental Setup
4.3 Results for Separate Measures
4.4 Baseline Classification Results
4.5 Results of the Trained Classifiers
5 Related Work
6 Conclusion and Future Work
References
Dominant-Set Clustering Using Multiple Affinity Matrices
1 Introduction
2 Dominant Sets as Equilibria of Evolutionary Games
3 Pareto Nash Equilibra in Multi-payoff Games
4 Dominant Sets with Multiple Affinities
4.1 Heuristics for Selecting Multi-objective Dominant Sets
4.2 Combination into a Single Affinity Matrix and Final Clustering
5 Experimental Results
5.1 Data Clustering
5.2 Application to Image Segmentation
6 Conclusions
References
Discovery of Salient Low-Dimensional Dynamical Structure in Neuronal Population Activity Using Hopfield Networks
1 Introduction
2 Background
3 Our Method
4 Application to Data
5 Conclusion
References
Author Index