Long-tail learning via logit adjustment

Aditya Krishna Menon, Sadeep Jayasumana, Ankit Singh Rawat, Himanshu Jain, Andreas Veit, Sanjiv Kumar

Keywords: long-tail learning, class imbalance

Abstract: Real-world classification problems typically exhibit an imbalanced or long-tailed label distribution, wherein many labels have only a few associated samples. This poses a challenge for generalisation on such labels, and also makes naive learning biased towards dominant labels. In this paper, we present a statistical framework that unifies and generalises several recent proposals to cope with these challenges. Our framework revisits the classic idea of logit adjustment based on the label frequencies, which encourages a large relative margin between logits of rare positive versus dominant negative labels. This yields two techniques for long-tail learning, where such adjustment is either applied post-hoc to a trained model, or enforced in the loss during training. These techniques are statistically grounded, and practically effective on four real-world datasets with long-tailed label distributions.

One-sentence Summary: Adjusting classifier logits based on class priors, either post-hoc or during training, can improve performance on rare classes.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Code: [ google-research/google-research](https://github.com/google-research/google-research/tree/master/logit_adjustment) + [ 2 community implementations](https://paperswithcode.com/paper/?openreview=37nvvqkCo5)

Data: [CIFAR100-LT](https://paperswithcode.com/dataset/cifar100-lt), [ImageNet-LT](https://paperswithcode.com/dataset/imagenet-lt)

Community Implementations: [ 1 code implementation](https://www.catalyzex.com/paper/arxiv:2007.07314/code)