Abstract:We propose a novel discriminative feature learning method via Max-Min Ratio Analysis (MMRA) for exclusively dealing with the long-standing "worst-case class separation" problem. Existing technologies simply consider maximizing the minimal pairwise distance on all class pairs in the low-dimensional subspace, which is unable to separate overlapped classes entirely especially when the distribution of samples within same class is diverging. We propose a new criterion, i.e., Max-Min Ratio Analysis (MMRA) that focuses on maximizing the minimal ratio value of between-class and within-class scatter to extremely enlarge the separability on the overlapped pairwise classes. Furthermore, we develop two novel discriminative feature learning models for dimensionality reduction and metric learning based on our MMRA criterion. However, solving such a non-smooth non-convex max-min ratio problem is challenging. As an important theoretical contribution in this paper, we systematically derive an alternative iterative algorithm based on a general max-min ratio optimization framework to solve a general max-min ratio problem with rigorous proofs of convergence. More importantly, we also present another solver based on bisection search strategy to solve the SDP problem efficiently. To evaluate the effectiveness of proposed methods, we conduct extensive pattern classification and image retrieval experiments on several artificial datasets and real-world ScRNA-seq datasets, and experimental results demonstrate the effectiveness of proposed methods.

Self/Non-self Discrimination Based on Fractional Distance

A Preliminary Study on Why Using the Nonself Detector Set for Anomaly Detection in Artificial Immune Systems

A New Fault Detection Method Based on Artificial Immune Systems

A Kernel Fractional-Step Nonlinear Discriminant Analysis for Pattern Recognition

Discriminative Concept Factorization for Data Representation

Face Recognition Using Novel LDA-Based Algorithms.

Adaptive Quasiconformal Kernel Fisher Discriminant Analysis Via Weighted Maximum Margin Criterion

Manifold distance based artificial immune incremental data clustering algorithm

Nearest Feature Space Analysis for Classification

A Novel Intrinsic Measure of Data Separability

Fuzzy discrimination analysis method based on RBFNN and its application in soft measurement

Worst-Case Discriminative Feature Learning via Max-Min Ratio Analysis

Analysis on fisher discriminant criterion and linear separability of feature space

Optimal Dimensionality of Metric Space for Classification

Feature Discriminability for Pattern Classification Based on Neural Incremental Attribute Learning

Discriminant analysis for perceptionally comparable classes.

Subspace Distance Analysis with Application to Adaptive Bayesian Algorithm for Face Recognition.

Two-Dimensional fisher discriminant analysis and its application to face recognition

A Variant Fisher Discriminant Analysis Algorithm and Its Applicationto Fault Diagnosis.

Fault Detection Based on Real-Value Negative Selection Algorithm of Artificial Immune System

Approximating Discrimination Within Models When Faced With Several Non-Binary Sensitive Attributes