Robust Semi-supervised Learning via $f$-Divergence and $\alpha$-R\'enyi Divergence

TL;DR

This paper explores divergence-based empirical risk functions and regularization methods for semi-supervised learning, demonstrating robustness to noisy pseudo-labels and improving performance over traditional self-training approaches.

Contribution

It introduces divergence-inspired empirical risk functions and regularization techniques that enhance robustness and performance in semi-supervised self-training.

Findings

Empirical risk functions are robust to noisy pseudo-labels.

Proposed methods outperform traditional self-training in certain conditions.

Insights into divergence measures improve understanding of semi-supervised learning.

Abstract

This paper investigates a range of empirical risk functions and regularization methods suitable for self-training methods in semi-supervised learning. These approaches draw inspiration from various divergence measures, such as $f$-divergences and $\alpha$-R\'enyi divergences. Inspired by the theoretical foundations rooted in divergences, i.e., $f$-divergences and $\alpha$-R\'enyi divergence, we also provide valuable insights to enhance the understanding of our empirical risk functions and regularization techniques. In the pseudo-labeling and entropy minimization techniques as self-training methods for effective semi-supervised learning, the self-training process has some inherent mismatch between the true label and pseudo-label (noisy pseudo-labels) and some of our empirical risk functions are robust, concerning noisy pseudo-labels. Under some conditions, our empirical risk functions demonstrate better performance when compared to traditional self-training methods.