Robust Conditional GAN from Uncertainty-Aware Pairwise Comparisons

Authors

  • Ligong Han Rutgers University
  • Ruijiang Gao The University of Texas at Austin
  • Mun Kim Rutgers University
  • Xin Tao Tencent YouTu Lab
  • Bo Liu JD Finance America Corporation
  • Dimitris Metaxas Rutgers University

DOI:

https://doi.org/10.1609/aaai.v34i07.6723

Abstract

Conditional generative adversarial networks have shown exceptional generation performance over the past few years. However, they require large numbers of annotations. To address this problem, we propose a novel generative adversarial network utilizing weak supervision in the form of pairwise comparisons (PC-GAN) for image attribute editing. In the light of Bayesian uncertainty estimation and noise-tolerant adversarial training, PC-GAN can estimate attribute rating efficiently and demonstrate robust performance in noise resistance. Through extensive experiments, we show both qualitatively and quantitatively that PC-GAN performs comparably with fully-supervised methods and outperforms unsupervised baselines. Code and Supplementary can be found on the project website*.

Downloads

Published

2020-04-03

How to Cite

Han, L., Gao, R., Kim, M., Tao, X., Liu, B., & Metaxas, D. (2020). Robust Conditional GAN from Uncertainty-Aware Pairwise Comparisons. Proceedings of the AAAI Conference on Artificial Intelligence, 34(07), 10909-10916. https://doi.org/10.1609/aaai.v34i07.6723

Issue

Section

AAAI Technical Track: Vision