Well-classified Examples are Underestimated in Classification with Deep Neural Networks

TL;DR

This paper challenges the traditional focus on poorly classified examples in deep learning, proposing a method to reward well-classified examples to improve representation learning, with empirical validation across multiple tasks.

Contribution

It introduces a novel approach that rewards well-classified examples, addressing limitations in current training practices and enhancing performance in diverse scenarios.

Findings

Improved classification accuracy across tasks

Enhanced representation learning and margin growth

Effective in imbalanced, OOD, and adversarial settings

Abstract

The conventional wisdom behind learning deep classification models is to focus on bad-classified examples and ignore well-classified examples that are far from the decision boundary. For instance, when training with cross-entropy loss, examples with higher likelihoods (i.e., well-classified examples) contribute smaller gradients in back-propagation. However, we theoretically show that this common practice hinders representation learning, energy optimization, and margin growth. To counteract this deficiency, we propose to reward well-classified examples with additive bonuses to revive their contribution to the learning process. This counterexample theoretically addresses these three issues. We empirically support this claim by directly verifying the theoretical results or significant performance improvement with our counterexample on diverse tasks, including image classification, graph classification, and machine translation. Furthermore, this paper shows that we can deal with complex scenarios, such as imbalanced classification, OOD detection, and applications under adversarial attacks because our idea can solve these three issues. Code is available at: https://github.com/lancopku/well-classified-examples-are-underestimated.