Expected Confidence Dependency: A Novel Rough Set-Based Approach to Feature Selection

2025-12-03

AI Summary

Traditional rough set dependency measures rely on binary equivalence-class membership, resulting in insufficient precision and poor stability. To address this, we propose the Expected Confidence Dependency (ECD) measure for feature selection. ECD incorporates confidence-weighted contributions of equivalence classes and aggregates them via a normalized expectation operator, thereby overcoming the limitations of binary classification. This work establishes, for the first time, an expectation-based dependency framework that rigorously satisfies normalization, monotonicity, compatibility with classical dependency, and structural invariance. Theoretical analysis and extensive experiments across multiple benchmark datasets demonstrate that ECD significantly improves both the accuracy and robustness of feature importance assessment. Feature subsets selected by ECD consistently achieve higher classification accuracy and greater stability compared to those obtained by state-of-the-art rough set methods.

Technology Category

• Machine Learning: Dimensionality Reduction/Feature Selection
• Reasoning under Uncertainty: Other Foundations of Reasoning under Uncertainty
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Security and Privacy: Large-scale security measurements
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

This paper proposes Expected Confidence Dependency (ECD), a novel, soft computing-oriented, accuracy driven dependency measure for feature selection within the rough set theory framework. Unlike traditional rough set dependency measures that rely on binary characterizations of conditional blocks, ECD assigns confidence-based contributions to individual equivalence blocks and aggregates them through a normalized expectation operator. We formally establish several desirable properties of ECD, including normalization, compatibility with classical dependency, monotonicity, and invariance under structural and label-preserving transformations.

Problem

• Develops a new feature selection measure using rough set theory
• Introduces confidence-based contributions for equivalence blocks
• Ensures properties like normalization and monotonicity in selection

Innovation

• ECD assigns confidence-based contributions to equivalence blocks
• ECD aggregates contributions via normalized expectation operator
• ECD is accuracy-driven dependency measure for feature selection