Uncertainty-Aware Ensemble Deep Randomized Neural Networks for Classification

2026-08-07

AI Summary

This work addresses the limitations of existing deep random vector functional-link networks (dRVFL), which treat all samples equally and thus struggle with real-world data contaminated by noise and outliers, leading to degraded discriminative performance due to error propagation through hidden layers. To overcome this, the study introduces an intuitionistic fuzzy mechanism into dRVFL for the first time, adaptively computing membership and non-membership degrees by jointly modeling each sample’s distance to its class center and the heterogeneity of its local neighborhood. This enables effective differentiation among clean, noisy, and anomalous samples, assigning them differentiated weights accordingly. Combined with kernel-space neighborhood analysis and ensemble learning, the proposed method significantly outperforms current state-of-the-art fuzzy and non-fuzzy approaches on UCI and KEEL benchmark datasets under both Gaussian-noise and noise-free settings, demonstrating superior robustness and generalization capability.

Abstract

The current state-of-the-art (SOTA) deep randomized neural networks, such as deep Random Vector Functional Link (dRVFL) and ensemble deep RVFL (edRVFL), treat all training samples uniformly, which limits their robustness and effectiveness when applied to real-world datasets containing noise and outliers. Furthermore, the propagation of contaminated features across hidden layers negatively influences the decision-making capability of these models. To overcome these limitations, we propose intuitionistic fuzzy dRVFL (IF-dRVFL) and intuitionistic fuzzy edRVFL (IF-edRVFL) frameworks that enhance model robustness. The proposed models unify intuitionistic fuzzy theory to exploit sample neighborhood information in the kernel space by jointly considering membership and non-membership degrees for each sample. Membership degrees are computed based on the distance of samples from their respective class centroids, while non-membership degrees quantify sample heterogeneity within local neighborhoods. These measures are employed to assign adaptive weights to training samples, enabling effective discrimination among clean, noisy, and outlier data points. Extensive experiments conducted on UCI and KEEL benchmark datasets, with and without the presence of Gaussian noise, demonstrate the superiority of the proposed IF-dRVFL and IF-edRVFL models over existing SOTA fuzzy and non-fuzzy approaches. The source code is available at https://github.com/mtanveer1/IF-edRVFL.

Problem

• deep randomized neural networks
• noise and outliers
• robustness
• uncertainty-aware classification
• sample weighting

Innovation

• Intuitionistic Fuzzy Theory
• Deep Randomized Neural Networks
• Uncertainty-Aware Learning
• Adaptive Sample Weighting
• Robust Classification