Disentangling Label Distribution for Long-tailed Visual Recognition

TL;DR

This paper addresses long-tailed visual recognition by formulating it as a label shift problem and proposes a novel LADE loss to disentangle label distributions, achieving state-of-the-art results across multiple datasets.

Contribution

It introduces a new LADE loss that directly disentangles source label distribution from model prediction during training, improving long-tailed recognition performance.

Findings

LADE outperforms existing methods on benchmark datasets.

A simple post-processing baseline surpasses previous state-of-the-art.

LADE demonstrates robustness across various label shift scenarios.

Abstract

The current evaluation protocol of long-tailed visual recognition trains the classification model on the long-tailed source label distribution and evaluates its performance on the uniform target label distribution. Such protocol has questionable practicality since the target may also be long-tailed. Therefore, we formulate long-tailed visual recognition as a label shift problem where the target and source label distributions are different. One of the significant hurdles in dealing with the label shift problem is the entanglement between the source label distribution and the model prediction. In this paper, we focus on disentangling the source label distribution from the model prediction. We first introduce a simple but overlooked baseline method that matches the target label distribution by post-processing the model prediction trained by the cross-entropy loss and the Softmax function. Although this method surpasses state-of-the-art methods on benchmark datasets, it can be further improved by directly disentangling the source label distribution from the model prediction in the training phase. Thus, we propose a novel method, LAbel distribution DisEntangling (LADE) loss based on the optimal bound of Donsker-Varadhan representation. LADE achieves state-of-the-art performance on benchmark datasets such as CIFAR-100-LT, Places-LT, ImageNet-LT, and iNaturalist 2018. Moreover, LADE outperforms existing methods on various shifted target label distributions, showing the general adaptability of our proposed method.