Fair Generative Modeling via Weak Supervision
This addresses the problem of dataset bias in unsupervised machine learning for researchers and practitioners, offering a method to improve fairness in generative models, though it is incremental as it builds on existing techniques like density ratio and GANs.
The paper tackles bias in real-world datasets for deep generative models by proposing a weakly supervised algorithm that uses a small unlabeled reference dataset to detect and mitigate bias without explicit labels. It reduces bias with respect to latent factors by an average of up to 34.6% over baselines in image generation tasks.
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.