Entropy based fuzzy multi-view twin random vector functional link for class imbalanced data

doi:10.1016/j.engappai.2025.113451

Entropy based fuzzy multi-view twin random vector functional link for class imbalanced data

2025-12-15

0

PRE

AI

DOI：10.1016/j.engappai.2025.113451

Original

Original request for help

Save

Abstract

En 中文

Multi-view learning (MVL) is employed to improve generalization performance through the integration of information collected from multiple views. Following MVL principles, and in conjunction with the twin random vector Functional Link (TRVFL) network and fuzzy entropy, we introduce here a new method of Entropy based Fuzzy Multi-view Twin Random Vector Functional Link (EFMvTRVFL) for real-world dataset classification. By taking the fuzzy membership value for each sample and allocating it according to the entropy value, samples with higher class certainty are allocated relatively greater fuzzy membership. The proposed model is designed to address uncertainty and class imbalance in real-world datasets by assigning fuzzy membership values based on entropy, thereby giving more importance to samples with higher class certainty. EFMvTRVFL retains the structural benefits of TRVFL but employs multi-view learning to learn complementary information from various feature representations. The novelty of EFMvTRVFL is the integration of fuzzy entropy-based membership assignment and multi-view TRVFL into a single optimization framework, whereas EFTWSVM and MvTWSVM lack random feature mapping and clear entropy regularization. Experimental evaluation on 20 benchmark datasets demonstrates that EFMvTRVFL consistently outperforms existing models in terms of both accuracy and stability. The results confirm the model's effectiveness in handling high-dimensional, imbalanced, and noisy data.

Entropy based fuzzy multi-view twin random vector functional link for class imbalanced data

Abstract

Journal

Organization