Rethinking the Value of Labels for Improving Class-Imbalanced Learning

Yuzhe Yang         Zhi Xu
Computer Science & Artificial Intelligence Laboratory
Massachusetts Institute of Technology


Real-world data often exhibits long-tailed distributions with heavy class imbalance, posing great challenges for deep recognition models. We identify a persisting dilemma on the value of labels in the context of imbalanced learning: on the one hand, supervision from labels typically leads to better results than its unsupervised counterparts; on the other hand, heavily imbalanced data naturally incurs "label bias" in the classifier, where the decision boundary can be drastically altered by the majority classes. In this work, we systematically investigate these two facets of labels. We demonstrate, theoretically and empirically, that class-imbalanced learning can significantly benefit in both semi-supervised and self-supervised manners. Specifically, we confirm that (1) positively, imbalanced labels are valuable: given more unlabeled data, the original labels can be leveraged with the extra data to reduce label bias in a semi-supervised manner, which greatly improves the final classifier; (2) negatively however, we argue that imbalanced labels are not useful always: classifiers that are first pre-trained in a self-supervised manner consistently outperform their corresponding baselines. Extensive experiments on large-scale imbalanced datasets verify our theoretically grounded strategies, showing superior performance over the previous state-of-the-arts. Our intriguing findings highlight the need to rethink the usage of imbalanced labels in realistic long-tailed tasks.


Rethinking the Value of Labels for Improving Class-Imbalanced Learning
Yuzhe Yang, and Zhi Xu
Conference on Neural Information Processing Systems (NeurIPS 2020)
[Paper]  •  [Code]  •  [Blog Post]  •  [BibTeX]  




(1) Semi-Supervised Imbalanced Learning: Using unlabeled data helps to shape clearer class boundaries and results in better class separation, especially for the tail classes.

(2) Self-Supervised Imbalanced Learning: Self-supervised pre-training (SSP) helps mitigate the tail classes leakage during testing, which results in better learned boundaries and representations.

(3) State-of-the-art Results: Improved performance on large-scale (real-world) imbalanced / long-tailed benchmarks.


  title={Rethinking the Value of Labels for Improving Class-Imbalanced Learning},
  author={Yang, Yuzhe and Xu, Zhi},
  booktitle={Conference on Neural Information Processing Systems (NeurIPS)},

Related Publications

Delving into Deep Imbalanced Regression
Yuzhe Yang, Kaiwen Zha, Ying-Cong Chen, Hao Wang, and Dina Katabi
International Conference on Machine Learning (ICML 2021)
[Project Page]  •  [Paper]  •  [Code]