Similarity-Based Pattern Recognition Third International Workshop, SIMBAD 2015, Copenhagen, Denmark, October 12-14, 2015. Proceedings

A Novel Data Representation based on a Second-Order Dissimilarity Measure.- Characterizing Multiple Instance Datasets.- Supervised learning of diffiusion distance to improve histogram matching.- Similarity Analysis from Limiting Quantum Walks.- Introducing Negative Evidence in Ensemble Clustering.- Dissimilarity representations for low-resolution face recognition.- Deep metric learning using Triplet network.- Cluster Merging Based on Dominant Sets.- An Adaptive Radial Basis Function Kernel for Support Vector Data Description.- Robust initialization for learning Latent Dirichlet Allocation.- Unsupervised Motion Segmentation Using Metric Embedding of Features.- Transitive Assignment Kernels for Structural Classification.- Large scale Indefinite Kernel Fisher Discriminant.- Similarity-based User Identification across Social Networks.- Dominant-Set Clustering Using Multiple Affinity Matrices.- Distance-Based Network Recovery under Feature Correlation.- Discovery of salient low-dimensional dynamical structure using Hopfield Networks.- On Geodesic Exponential Kernels.- A Matrix Factorization Approach to Graph Compression.- A Geometrical Approach to Find Corresponding Patches in 3D Medical Surfaces.- Similarities, SDEs, and Most Probable Paths.- Can the optimum similarity matrix be selected before clustering for graph-based approaches?.- Approximate spectral clustering with utilized similarity information fusing geodesic based hybrid distance measures.