Peer Loss Functions: Learning from Noisy Labels without Knowing Noise Rates

TL;DR

This paper introduces peer loss functions, a new family of loss functions that enable learning from noisy labels without needing to know noise rates, simplifying model training in noisy label scenarios.

Contribution

The paper proposes peer loss functions that work within the ERM framework and do not require prior knowledge of noise rates, providing a robust alternative for learning with noisy labels.

Findings

Peer loss functions achieve near-optimal classifiers on noisy datasets.

The approach simplifies model development in noisy label environments.

Extensive experiments validate the effectiveness of peer loss functions.

Abstract

Learning with noisy labels is a common challenge in supervised learning. Existing approaches often require practitioners to specify noise rates, i.e., a set of parameters controlling the severity of label noises in the problem, and the specifications are either assumed to be given or estimated using additional steps. In this work, we introduce a new family of loss functions that we name as peer loss functions, which enables learning from noisy labels and does not require a priori specification of the noise rates. Peer loss functions work within the standard empirical risk minimization (ERM) framework. We show that, under mild conditions, performing ERM with peer loss functions on the noisy dataset leads to the optimal or a near-optimal classifier as if performing ERM over the clean training data, which we do not have access to. We pair our results with an extensive set of experiments. Peer loss provides a way to simplify model development when facing potentially noisy training labels, and can be promoted as a robust candidate loss function in such situations.