Understanding Deep Generative Models with Generalized Empirical Likelihoods


Understanding how well a deep generative model captures a distribution of high-dimensional data remains an important open challenge. It is especially difficult for certain model classes, such as Generative Adversarial Networks and Diffusion Models, whose models do not admit exact likelihoods. In this work, we demonstrate that generalized empirical likelihood (GEL) methods offer a family of diagnostic tools that can identify many issues of deep generative models (DGMs). We show, with appropriate specification of moment conditions, that the proposed method can identify which modes have been dropped, the degree to which DGMs are mode imbalanced, and whether DGMs sufficiently capture intraclass diversity. We show how to combine techniques from Maximum Mean Discrepancy and Generalized Empirical Likelihood to create not only distribution tests that retain per sample interpretability, but also metrics that include label information. We find that such tests predict the amount of mode dropping and mode imbalance up to 60% better than metrics such as improved precision/recall.

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)