Mitigating Dataset Imbalance via Joint Generation and Classification

TL;DR

This paper proposes a joint classification and generation approach using GANs to address dataset imbalance, improving robustness of classifiers and generators across diverse datasets.

Contribution

It introduces a novel joint training strategy combining classifiers with GANs to mitigate class imbalance effects in supervised learning.

Findings

Improved classifier accuracy on imbalanced datasets.

Enhanced generation quality for under-represented classes.

Robustness of the method across multiple datasets.

Abstract

Supervised deep learning methods are enjoying enormous success in many practical applications of computer vision and have the potential to revolutionize robotics. However, the marked performance degradation to biases and imbalanced data questions the reliability of these methods. In this work we address these questions from the perspective of dataset imbalance resulting out of severe under-representation of annotated training data for certain classes and its effect on both deep classification and generation methods. We introduce a joint dataset repairment strategy by combining a neural network classifier with Generative Adversarial Networks (GAN) that makes up for the deficit of training examples from the under-representated class by producing additional training examples. We show that the combined training helps to improve the robustness of both the classifier and the GAN against severe class imbalance. We show the effectiveness of our proposed approach on three very different datasets with different degrees of imbalance in them. The code is available at https://github.com/AadSah/ImbalanceCycleGAN .