Elastic-InfoGAN: Unsupervised Disentangled Representation Learning in Imbalanced Data

Utkarsh Ojha, Krishna Kumar Singh, Cho-Jui Hsieh, Yong Jae Lee

TL;DR: Elastic-InfoGAN is a modification of InfoGAN that learns, without any supervision, disentangled representations in class imbalanced data

Abstract: We propose a novel unsupervised generative model, Elastic-InfoGAN, that learns to disentangle object identity from other low-level aspects in class-imbalanced datasets. We first investigate the issues surrounding the assumptions about uniformity made by InfoGAN, and demonstrate its ineffectiveness to properly disentangle object identity in imbalanced data. Our key idea is to make the discovery of the discrete latent factor of variation invariant to identity-preserving transformations in real images, and use that as the signal to learn the latent distribution's parameters. Experiments on both artificial (MNIST) and real-world (YouTube-Faces) datasets demonstrate the effectiveness of our approach in imbalanced data by: (i) better disentanglement of object identity as a latent factor of variation; and (ii) better approximation of class imbalance in the data, as reflected in the learned parameters of the latent distribution.

Keywords: Generative Adversarial Networks, Imbalanced data, Data Augmentation

Community Implementations: [ 1 code implementation](https://www.catalyzex.com/paper/arxiv:1910.01112/code)