How does Disagreement Help Generalization against Label Corruption?

Abstract

Learning with noisy labels is one of the hottest problems in weakly-supervised learning. Based on memorization effects of deep neural networks, training on small-loss instances becomes very promising for handling noisy labels. This fosters the state-of-the-art approach "Co-teaching" that cross-trains two deep neural networks using the small-loss trick. However, with the increase of epochs, two networks converge to a consensus and Co-teaching reduces to the self-training MentorNet. To tackle this issue, we propose a robust learning paradigm called Co-teaching+, which bridges the "Update by Disagreement" strategy with the original Co-teaching. First, two networks feed forward and predict all data, but keep prediction disagreement data only. Then, among such disagreement data, each network selects its small-loss data, but back propagates the small-loss data from its peer network and updates its own parameters. Empirical results on benchmark datasets demonstrate that Co-teaching+ is much superior to many state-of-the-art methods in the robustness of trained models.

References in corpus (3)

• A Closer Look at Memorization in Deep Networks
• Robust Inference via Generative Classifiers for Handling Noisy Labels
• Theoretical Foundation of Co-Training and Disagreement-Based Algorithms

Cited by in corpus (6)

• Distillation $\approx$ Early Stopping? Harvesting Dark Knowledge Utilizing Anisotropic Information Retrieval For Overparameterized Neural Network
• Meta Transition Adaptation for Robust Deep Learning with Noisy Labels
• Towards Noise-resistant Object Detection with Noisy Annotations
• Cross-denoising Network against Corrupted Labels in Medical Image Segmentation with Domain Shift
• More Information Supervised Probabilistic Deep Face Embedding Learning
• P-DIFF: Learning Classifier with Noisy Labels based on Probability Difference Distributions