Quality-Diversity Generative Sampling for Learning with Synthetic Data
This work addresses bias in synthetic training data for machine learning, offering a model-agnostic solution to enhance fairness in applications like facial recognition, though it is incremental as it builds on existing generative sampling methods.
The paper tackles the problem of generative models transferring biases to downstream tasks by proposing quality-diversity generative sampling (QDGS), a framework that samples synthetic data uniformly across user-defined measures to protect quality and diversity. Results show that using QDGS-generated balanced datasets debiases classifiers on color-biased shape datasets and improves fairness while maintaining accuracy on facial recognition benchmarks.
Generative models can serve as surrogates for some real data sources by creating synthetic training datasets, but in doing so they may transfer biases to downstream tasks. We focus on protecting quality and diversity when generating synthetic training datasets. We propose quality-diversity generative sampling (QDGS), a framework for sampling data uniformly across a user-defined measure space, despite the data coming from a biased generator. QDGS is a model-agnostic framework that uses prompt guidance to optimize a quality objective across measures of diversity for synthetically generated data, without fine-tuning the generative model. Using balanced synthetic datasets generated by QDGS, we first debias classifiers trained on color-biased shape datasets as a proof-of-concept. By applying QDGS to facial data synthesis, we prompt for desired semantic concepts, such as skin tone and age, to create an intersectional dataset with a combined blend of visual features. Leveraging this balanced data for training classifiers improves fairness while maintaining accuracy on facial recognition benchmarks. Code available at: https://github.com/Cylumn/qd-generative-sampling.