Are Anchor Points Really Indispensable in Label-Noise Learning?

TL;DR

This paper introduces a transition-revision method for label-noise learning that effectively learns transition matrices without relying on anchor points, improving classifier performance on noisy datasets.

Contribution

The paper proposes a novel transition-revision approach that eliminates the need for anchor points in learning transition matrices, enhancing robustness in label-noise learning.

Findings

Outperforms state-of-the-art methods on benchmark datasets

Effective in scenarios without anchor points

Improves classifier accuracy in noisy label settings

Abstract

In label-noise learning, \textit{noise transition matrix}, denoting the probabilities that clean labels flip into noisy labels, plays a central role in building \textit{statistically consistent classifiers}. Existing theories have shown that the transition matrix can be learned by exploiting \textit{anchor points} (i.e., data points that belong to a specific class almost surely). However, when there are no anchor points, the transition matrix will be poorly learned, and those current consistent classifiers will significantly degenerate. In this paper, without employing anchor points, we propose a \textit{transition-revision} ($T$-Revision) method to effectively learn transition matrices, leading to better classifiers. Specifically, to learn a transition matrix, we first initialize it by exploiting data points that are similar to anchor points, having high \textit{noisy class posterior probabilities}. Then, we modify the initialized matrix by adding a \textit{slack variable}, which can be learned and validated together with the classifier by using noisy data. Empirical results on benchmark-simulated and real-world label-noise datasets demonstrate that without using exact anchor points, the proposed method is superior to the state-of-the-art label-noise learning methods.