Rethinking Class-Balanced Methods for Long-Tailed Visual Recognition from a Domain Adaptation Perspective

TL;DR

This paper reinterprets class-balanced methods for long-tailed visual recognition through the lens of domain adaptation, revealing their assumptions and proposing a meta-learning based augmentation to better handle tail classes.

Contribution

It connects class-balanced methods to target shift in domain adaptation and introduces a meta-learning approach to explicitly estimate distribution differences for improved long-tailed recognition.

Findings

Improved performance on six benchmark datasets

Meta-learning approach effectively estimates class distribution differences

Enhanced handling of tail classes in long-tailed recognition

Abstract

Object frequency in the real world often follows a power law, leading to a mismatch between datasets with long-tailed class distributions seen by a machine learning model and our expectation of the model to perform well on all classes. We analyze this mismatch from a domain adaptation point of view. First of all, we connect existing class-balanced methods for long-tailed classification to target shift, a well-studied scenario in domain adaptation. The connection reveals that these methods implicitly assume that the training data and test data share the same class-conditioned distribution, which does not hold in general and especially for the tail classes. While a head class could contain abundant and diverse training examples that well represent the expected data at inference time, the tail classes are often short of representative training data. To this end, we propose to augment the classic class-balanced learning by explicitly estimating the differences between the class-conditioned distributions with a meta-learning approach. We validate our approach with six benchmark datasets and three loss functions.