Towards Calibrated Model for Long-Tailed Visual Recognition from Prior Perspective

TL;DR

This paper introduces two novel methods, UniMix and Bayias, to improve calibration and performance in long-tailed visual recognition by addressing class imbalance and prior bias, validated through extensive experiments.

Contribution

The paper proposes UniMix for balanced data augmentation and Bayias for bias correction, both ensuring better calibration in long-tailed recognition tasks.

Findings

Both methods improve model calibration theoretically and empirically.

Combined strategies achieve state-of-the-art results on multiple long-tailed datasets.

The approaches effectively address class imbalance and prior bias issues.

Abstract

Real-world data universally confronts a severe class-imbalance problem and exhibits a long-tailed distribution, i.e., most labels are associated with limited instances. The na\"ive models supervised by such datasets would prefer dominant labels, encounter a serious generalization challenge and become poorly calibrated. We propose two novel methods from the prior perspective to alleviate this dilemma. First, we deduce a balance-oriented data augmentation named Uniform Mixup (UniMix) to promote mixup in long-tailed scenarios, which adopts advanced mixing factor and sampler in favor of the minority. Second, motivated by the Bayesian theory, we figure out the Bayes Bias (Bayias), an inherent bias caused by the inconsistency of prior, and compensate it as a modification on standard cross-entropy loss. We further prove that both the proposed methods ensure the classification calibration theoretically and empirically. Extensive experiments verify that our strategies contribute to a better-calibrated model, and their combination achieves state-of-the-art performance on CIFAR-LT, ImageNet-LT, and iNaturalist 2018.