Asymmetric Co-teaching with Multi-view Consensus for Noisy Label Learning

TL;DR

This paper introduces Asymmetric Co-teaching with multi-view consensus, a novel approach for noisy-label learning that enhances robustness and reduces confirmation bias by using different training strategies and multi-view sample selection.

Contribution

The paper proposes a new asymmetric co-teaching method with multi-view consensus, improving noisy-label learning by addressing confirmation bias and eliminating the small-loss assumption.

Findings

Outperforms current SOTA methods on synthetic noisy datasets.

Demonstrates robustness to different types and levels of label noise.

Effective in real-world noisy-label datasets.

Abstract

Learning with noisy-labels has become an important research topic in computer vision where state-of-the-art (SOTA) methods explore: 1) prediction disagreement with co-teaching strategy that updates two models when they disagree on the prediction of training samples; and 2) sample selection to divide the training set into clean and noisy sets based on small training loss. However, the quick convergence of co-teaching models to select the same clean subsets combined with relatively fast overfitting of noisy labels may induce the wrong selection of noisy label samples as clean, leading to an inevitable confirmation bias that damages accuracy. In this paper, we introduce our noisy-label learning approach, called Asymmetric Co-teaching (AsyCo), which introduces novel prediction disagreement that produces more consistent divergent results of the co-teaching models, and a new sample selection approach that does not require small-loss assumption to enable a better robustness to confirmation bias than previous methods. More specifically, the new prediction disagreement is achieved with the use of different training strategies, where one model is trained with multi-class learning and the other with multi-label learning. Also, the new sample selection is based on multi-view consensus, which uses the label views from training labels and model predictions to divide the training set into clean and noisy for training the multi-class model and to re-label the training samples with multiple top-ranked labels for training the multi-label model. Extensive experiments on synthetic and real-world noisy-label datasets show that AsyCo improves over current SOTA methods.