Corpus ID: 195345228

Bias Correction of Learned Generative Models using Likelihood-Free Importance Weighting

  title={Bias Correction of Learned Generative Models using Likelihood-Free Importance Weighting},
  author={Aditya Grover and Jiaming Song and A. Agarwal and K. Tran and A. Kapoor and E. Horvitz and S. Ermon},
  • Aditya Grover, Jiaming Song, +4 authors S. Ermon
  • Published 2019
  • Mathematics, Computer Science
  • ArXiv
  • A learned generative model often produces biased statistics relative to the underlying data distribution. [...] Key Result Finally, we demonstrate its utility on representative applications in a) data augmentation for classification using generative adversarial networks, and b) model-based policy evaluation using off-policy data.Expand Abstract
    Fair Generative Modeling via Weak Supervision
    • 5
    • PDF
    Towards Fairness in Visual Recognition: Effective Strategies for Bias Mitigation
    • 5
    • PDF
    Energy-Based Models for Text
    • 5
    • PDF
    KALE: When Energy-Based Learning Meets Adversarial Training
    • 3
    Subsampling Generative Adversarial Networks: Density Ratio Estimation in Feature Space With Softplus Loss
    • 2
    • Highly Influenced
    • PDF
    Effectively Unbiased FID and Inception Score and Where to Find Them
    • 2
    • PDF


    Publications referenced by this paper.
    Generative Adversarial Nets
    • 17,776
    • PDF
    Auto-Encoding Variational Bayes
    • 9,147
    • PDF
    On sequential Monte Carlo sampling methods for Bayesian filtering
    • 4,378
    • PDF
    Improved Techniques for Training GANs
    • 3,585
    • PDF
    Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning
    • 2,222
    • PDF
    Automatic differentiation in PyTorch
    • 6,546
    An Introduction to the Bootstrap
    • 9,934
    • PDF
    GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
    • 1,748
    • PDF