Fair Generative Modeling via Weak Supervision
arXiv:1910.12008
Abstract
Real-world datasets are often biased with respect to key demographic factors such as race and gender. Due to the latent nature of the underlying factors, detecting and mitigating bias is especially challenging for unsupervised machine learning. We present a weakly supervised algorithm for overcoming dataset bias for deep generative models. Our approach requires access to an additional small, unlabeled reference dataset as the supervision signal, thus sidestepping the need for explicit labels on the underlying bias factors. Using this supplementary dataset, we detect the bias in existing datasets via a density ratio technique and learn generative models which efficiently achieve the twin goals of: 1) data efficiency by using training examples from both biased and reference datasets for learning; and 2) data generation close in distribution to the reference dataset at test time. Empirically, we demonstrate the efficacy of our approach which reduces bias w.r.t. latent factors by an average of up to 34.6% over baselines for comparable image generation using generative adversarial networks.
arXiv commentFirst two authors contributed equally
References in corpus (25)
- GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
- Large Scale GAN Training for High Fidelity Natural Image Synthesis
- Stochastic Backpropagation and Approximate Inference in Deep Generative Models
- Conditional Image Synthesis With Auxiliary Classifier GANs
- Equality of Opportunity in Supervised Learning
- Progressive Growing of GANs for Improved Quality, Stability, and Variation
- NICE: Non-linear Independent Components Estimation
- Snorkel: Rapid Training Data Creation with Weak Supervision
- MMD GAN: Towards Deeper Understanding of Moment Matching Network
- Data Decisions and Theoretical Implications when Adversarially Learning Fair Representations
- Variational Approaches for Auto-Encoding Generative Adversarial Networks
- Fairness Without Demographics in Repeated Loss Minimization
- Learning in Implicit Generative Models
- Censoring Representations with an Adversary
- Counterpoint by Convolution
- InclusiveFaceNet: Improving Face Attribute Detection with Race and Gender Diversity
- Flow-GAN: Combining Maximum Likelihood and Adversarial Learning in Generative Models
- Bias Correction of Learned Generative Models using Likelihood-Free Importance Weighting
- Learning Controllable Fair Representations
- Comparison of Maximum Likelihood and GAN-based training of Real NVPs
- Fair Coresets and Streaming Algorithms for Fair k-Means Clustering
- Towards GAN Benchmarks Which Require Generalization
- Metropolis-Hastings Generative Adversarial Networks
- Quantitatively Evaluating GANs With Divergences Proposed for Training
- Boosted Generative Models