Rethinking Noisy Label Learning in Real-world Annotation Scenarios from the Noise-type Perspective

TL;DR

This paper introduces Proto-semi, a novel approach for learning from noisy labels in real-world scenarios by distinguishing noise types and refining labels through prototype-based classification and semi-supervised learning.

Contribution

The paper proposes a new sample selection method that differentiates factual and ambiguity noise, improving noisy label learning with prototype-based classification.

Findings

Proto-semi outperforms existing methods on real-world datasets.

Prototype-based repartitioning reduces the impact of label noise.

Semi-supervised learning enhances model robustness.

Abstract

In this paper, we investigate the problem of learning with noisy labels in real-world annotation scenarios, where noise can be categorized into two types: factual noise and ambiguity noise. To better distinguish these noise types and utilize their semantics, we propose a novel sample selection-based approach for noisy label learning, called Proto-semi. Proto-semi initially divides all samples into the confident and unconfident datasets via warm-up. By leveraging the confident dataset, prototype vectors are constructed to capture class characteristics. Subsequently, the distances between the unconfident samples and the prototype vectors are calculated to facilitate noise classification. Based on these distances, the labels are either corrected or retained, resulting in the refinement of the confident and unconfident datasets. Finally, we introduce a semi-supervised learning method to enhance training. Empirical evaluations on a real-world annotated dataset substantiate the robustness of Proto-semi in handling the problem of learning from noisy labels. Meanwhile, the prototype-based repartitioning strategy is shown to be effective in mitigating the adverse impact of label noise. Our code and data are available at https://github.com/fuxiAIlab/ProtoSemi.