Jo-SRC: A Contrastive Approach for Combating Noisy Labels

TL;DR

Jo-SRC is a novel contrastive learning method that improves training with noisy labels by jointly selecting samples and regularizing the model, leading to better robustness and performance.

Contribution

The paper introduces Jo-SRC, a contrastive learning framework that effectively handles noisy labels through joint sample selection and consistency regularization.

Findings

Outperforms existing methods on noisy label benchmarks

Effectively distinguishes clean and noisy samples using contrastive predictions

Enhances model robustness and generalization in noisy environments

Abstract

Due to the memorization effect in Deep Neural Networks (DNNs), training with noisy labels usually results in inferior model performance. Existing state-of-the-art methods primarily adopt a sample selection strategy, which selects small-loss samples for subsequent training. However, prior literature tends to perform sample selection within each mini-batch, neglecting the imbalance of noise ratios in different mini-batches. Moreover, valuable knowledge within high-loss samples is wasted. To this end, we propose a noise-robust approach named Jo-SRC (Joint Sample Selection and Model Regularization based on Consistency). Specifically, we train the network in a contrastive learning manner. Predictions from two different views of each sample are used to estimate its "likelihood" of being clean or out-of-distribution. Furthermore, we propose a joint loss to advance the model generalization performance by introducing consistency regularization. Extensive experiments have validated the superiority of our approach over existing state-of-the-art methods.