Timezone: »
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 naïve 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.
Author Information
Zhengzhuo Xu (Tsinghua University)
Zenghao Chai (Shenzhen International Graduate School, Tsinghua University)
Chun Yuan (Tsinghua University)
More from the Same Authors
-
2022 : On the Sparsity of Image Super-resolution Network »
Chenyu Dong · Hailong Ma · Jinjin Gu · Ruofan Zhang · Jieming Li · Chun Yuan -
2022 Spotlight: One Model to Edit Them All: Free-Form Text-Driven Image Manipulation with Semantic Modulations »
Yiming Zhu · Hongyu Liu · Yibing Song · Ziyang Yuan · Xintong Han · Chun Yuan · Qifeng Chen · Jue Wang -
2022 Poster: One Model to Edit Them All: Free-Form Text-Driven Image Manipulation with Semantic Modulations »
Yiming Zhu · Hongyu Liu · Yibing Song · Ziyang Yuan · Xintong Han · Chun Yuan · Qifeng Chen · Jue Wang -
2017 : Competition II: Learning to Run »
Łukasz Kidziński · Carmichael Ong · Sharada Mohanty · Jason Fries · Jennifer Hicks · Zhuobin Zheng · Chun Yuan · Sergey Plis