Selecting the Best in GANs Family: a Post Selection Inference Framework
"Which Generative Adversarial Networks (GANs) generates the most plausible images?" has been a frequently asked question among researchers. To address this problem, we first propose an incomplete U-statistics estimate of maximum mean discrepancy MMD_inc to measure the distribution discrepancy between generated and real images. MMD_inc enjoys the advantages of asymptotic normality, computation efficiency, and model agnosticity. We then propose a GANs analysis framework to select and test the "best" member in GANs family using the Post Selection Inference (PSI) with MMD_inc. In the experiments, we adopt the proposed framework on 7 GANs variants and compare their MMD_inc scores.
READ FULL TEXT