Weak-supervision for Deep Representation Learning under Class Imbalance

TL;DR

This paper introduces a weak-supervision framework that leverages automatically-generated labels to improve deep learning performance on imbalanced multi-class image classification tasks.

Contribution

It extends deep over-sampling methods by using side-information to guide feature learning, addressing class imbalance among many classes.

Findings

Significant improvement on imbalanced image classification benchmarks.

Effective use of abstract-labels to guide feature separation.

Enhanced deep representation learning under class imbalance.

Abstract

Class imbalance is a pervasive issue among classification models including deep learning, whose capacity to extract task-specific features is affected in imbalanced settings. However, the challenges of handling imbalance among a large number of classes, commonly addressed by deep learning, have not received a significant amount of attention in previous studies. In this paper, we propose an extension of the deep over-sampling framework, to exploit automatically-generated abstract-labels, i.e., a type of side-information used in weak-label learning, to enhance deep representation learning against class imbalance. We attempt to exploit the labels to guide the deep representation of instances towards different subspaces, to induce a soft-separation of inherent subtasks of the classification problem. Our empirical study shows that the proposed framework achieves a substantial improvement on image classification benchmarks with imbalanced among large and small numbers of classes.