Measuring Fairness in Generative Models
arXiv:2107.07754
Abstract
Deep generative models have made much progress in improving training stability and quality of generated data. Recently there has been increased interest in the fairness of deep-generated data. Fairness is important in many applications, e.g. law enforcement, as biases will affect efficacy. Central to fair data generation are the fairness metrics for the assessment and evaluation of different generative models. In this paper, we first review fairness metrics proposed in previous works and highlight potential weaknesses. We then discuss a performance benchmark framework along with the assessment of alternative metrics.
arXiv commentAccepted in ICML 2021 Workshop - Machine Learning for Data: Automated Creation, Privacy, Bias
References in corpus (12)
- Auto-Encoding Variational Bayes
- Deep Residual Learning for Image Recognition
- Large Scale GAN Training for High Fidelity Natural Image Synthesis
- Improved Training of Wasserstein GANs
- NIPS 2016 Tutorial: Generative Adversarial Networks
- Fairness in Machine Learning: A Survey
- Unrolled Generative Adversarial Networks
- Deep Learning Face Attributes in the Wild
- Bias Correction of Learned Generative Models using Likelihood-Free Importance Weighting
- FairGAN: Fairness-aware Generative Adversarial Networks
- Self-supervised GAN: Analysis and Improvement with Multi-class Minimax Game
- Improving the Fairness of Deep Generative Models without Retraining