Affinity-based fuzzy twin random vector functional link network classifier

doi:10.1016/j.compeleceng.2025.110923

Affinity-based fuzzy twin random vector functional link network classifier

2026-01-08

0

PRE

AI

Barenya Bikash Hazarika

DOI：10.1016/j.compeleceng.2025.110923

Original

Original request for help

Save

Abstract

En 中文

In real-world, numerous leaf diseases are proliferating due to soil pollution and weather-related factors. Manual identification is slow and often ineffective. Identification hazards are created when noisy data and binary class imbalance problems are present. To address the noise and imbalanced data issue, several affinity and class probability-models were suggested, which reduce noise through regularization and handles class imbalance using affinity values from support vector data description (SVDD) and class probabilities from k-nearest neighbour (KNN). Minority samples with low affinity and probability receive less weight, while majority samples with higher values strongly influence the decision boundary. To enhance generalization an computational efficiency, an affinity and class probability-based fuzzy random vector functional link network (ACFRVFL) is introduced, combining fuzzy logic, SVDD, and KNN with RVFL. Moreover, an affinity and class probability-based fuzzy twin RVFL (ACFTRVFL) model is also suggested for improved performance. The study evaluates performance using various benchmark datasets.

Affinity-based fuzzy twin random vector functional link network classifier

Abstract

Journal

Organization