Countering Noisy Labels By Learning From Auxiliary Clean Labels

TL;DR

This paper introduces RDCR, a unified framework that leverages auxiliary clean labels and self-supervised rotation tasks to improve learning from noisy labels, especially under high noise conditions.

Contribution

The paper proposes a novel RDCR framework combining consistency regularization with self-supervised rotation tasks to enhance noise robustness in label learning.

Findings

RDCR performs comparably to state-of-the-art under small noise.

RDCR significantly outperforms existing methods with large noise.

The approach effectively utilizes auxiliary clean labels and self-supervision.

Abstract

We consider the learning from noisy labels (NL) problem which emerges in many real-world applications. In addition to the widely-studied synthetic noise in the NL literature, we also consider the pseudo labels in semi-supervised learning (Semi-SL) as a special case of NL. For both types of noise, we argue that the generalization performance of existing methods is highly coupled with the quality of noisy labels. Therefore, we counter the problem from a novel and unified perspective: learning from the auxiliary clean labels. Specifically, we propose the Rotational-Decoupling Consistency Regularization (RDCR) framework that integrates the consistency-based methods with the self-supervised rotation task to learn noise-tolerant representations. The experiments show that RDCR achieves comparable or superior performance than the state-of-the-art methods under small noise, while outperforms the existing methods significantly when there is large noise.