Masking: A New Perspective of Noisy Supervision

TL;DR

This paper introduces Masking, a human-assisted method for robust classifier training with noisy labels by leveraging human cognition to specify invalid class transitions, reducing the complexity of estimating noise transition matrices.

Contribution

It proposes a structure-aware probabilistic model that incorporates human knowledge to simplify noise transition estimation, improving classifier robustness under noisy supervision.

Findings

Masking significantly improves classifier robustness on CIFAR-10 and CIFAR-100.

It effectively handles various noise structures, including industrial-level noise.

Experimental results demonstrate substantial performance gains over existing methods.

Abstract

It is important to learn various types of classifiers given training data with noisy labels. Noisy labels, in the most popular noise model hitherto, are corrupted from ground-truth labels by an unknown noise transition matrix. Thus, by estimating this matrix, classifiers can escape from overfitting those noisy labels. However, such estimation is practically difficult, due to either the indirect nature of two-step approaches, or not big enough data to afford end-to-end approaches. In this paper, we propose a human-assisted approach called Masking that conveys human cognition of invalid class transitions and naturally speculates the structure of the noise transition matrix. To this end, we derive a structure-aware probabilistic model incorporating a structure prior, and solve the challenges from structure extraction and structure alignment. Thanks to Masking, we only estimate unmasked noise transition probabilities and the burden of estimation is tremendously reduced. We conduct extensive experiments on CIFAR-10 and CIFAR-100 with three noise structures as well as the industrial-level Clothing1M with agnostic noise structure, and the results show that Masking can improve the robustness of classifiers significantly.