Adversarial Fisher Vectors for Unsupervised Representation Learning
AuthorsShuangfei Zhai, Walter Talbott, Carlos Guestrin, Joshua M. Susskind
We examine Generative Adversarial Networks (GANs) through the lens of deep Energy Based Models (EBMs), with the goal of exploiting the density model that follows from this formulation. In contrast to a traditional view where the discriminator learns a constant function when reaching convergence, here we show that it can provide useful information for downstream tasks, e.g., feature extraction for classification. To be concrete, in the EBM formulation, the discriminator learns an unnormalized density function (i.e., the negative energy term) that characterizes the data manifold. We propose to evaluate both the generator and the discriminator by deriving corresponding Fisher Score and Fisher Information from the EBM. We show that by assuming that the generated examples form an estimate of the learned density, both the Fisher Information and the normalized Fisher Vectors are easy to compute. We also show that we are able to derive a distance metric between examples and between sets of examples. We conduct experiments showing that the GAN-induced Fisher Vectors demonstrate competitive performance as unsupervised feature extractors for classification and perceptual similarity tasks.
Apple attended the 33rd Conference and Workshop on Neural Information Processing Systems (NeurIPS) in December. The conference took place in Vancouver, Canada from December 8th to 14th.
Apple product teams are engaged in state of the art research in machine hearing, speech recognition, natural language processing, machine translation, text-to-speech, and artificial intelligence, improving the lives of millions of customers every day.