Abstract:Information theory has been widely used to evaluate the relevance and redundancy of features in feature selection. The traditional feature selection methods based on information theory focus on classification-specific feature selection. Classification-specific feature selection considers all classes together and therefore may not be able to select a desirable feature subset for a particular class. In contrast, class-specific feature selection can choose an appropriate feature subset for each class. The main goal of this paper is to propose a class-specific feature selection method based on information theory. In this regard, we first develop a class-specific feature evaluation criterion called Class-Specific Maximal Dynamic Correlation Change and Minimal Redundancy (CSMDCCMR), which consists of three items. The first item characterizes the correlation between a candidate feature and a particular class. The second item quantifies the dynamic changes in the relevance of a candidate feature and selected features with respect to a particular class. The third item expresses the redundancy between a candidate feature and selected features with respect to a particular class. We then design a class-specific feature selection algorithm by combining the CSMDCCMR criterion with a sequential forward search strategy, which can be employed to select a suitable feature subset for each class. Finally, we carry out extensive experiments to compare our class-specific feature selection method with nine representative classification-specific feature selection methods based on information theory and six state-of-the-art class-specific feature selection methods. The experimental results show that our class-specific feature selection method achieves better classification performance than these comparison methods.

A Selective Bayesian Classifier Based on Change of Class Relevance Influence

An Improved Maximum Relevance and Minimum Redundancy Selective Bayesian Classifier

Selective Bayesian Classifier Based on Semi-supervised Clustering

Binary Classification Algorithm with Class-Dependent Reject Cost

Feature selection method based on conditional relevance

Class-specific feature selection via maximal dynamic correlation change and minimal redundancy

Feature Selection with Maximal Relevance and Minimal Supervised Redundancy

An Improved Classification Approach by Case-based Reasoning

Confidence-aware Contrastive Learning for Selective Classification

An Adaptive Feature Selection Method for Multi-Class Classification.

Multimodal Perturbation and Cluster Pruning Based Selective Ensemble Classifier and Its Iron Industrial Application

On Improving Reliability of Case-based Reasoning Classifier

An Optimal Bayes Classification Algorithm

An Optimally Selective Ensemble Classifier Based on Multimodal Perturbation and Its Application

Method of Feature Selection for Text Categorization with Bayesian Classifiers

A discriminative model selection approach and its application to text classification

Causal Feature Selection With Dual Correction

Research on the Multi-Source Causal Feature Selection Method Based on Multiple Causal Relevance

Towards Efficient Dimensionality Reduction for Evolving Bayesian Network Classifier

Distribution-Based Selective Classifiers for Incomplete Data

Class dependent feature scaling method via restrictive Bayesian network classifier combination