Adversarial Fisher Vectors for Unsupervised Representation Learning
Joshua Susskind · Shuangfei Zhai · Walter Talbott · Carlos Guestrin

Tue Dec 10th 05:15 -- 05:20 PM @ West Exhibition Hall C + B3

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. Code is available at \url{}.

Author Information

Joshua Susskind (Apple Inc.)

Did my PhD work at the University of Toronto in machine learning and cognitive neuroscience with Dr. Geoff Hinton and Adam Anderson, then moved to San Diego for a post-doctoral position at UC San Diego. Before coming to Apple I co-founded Emotient in 2012 and led the deep learning effort for facial expression and demographics recognition. Since joining Apple, I have led the Face ID neural network team responsible for face recognition, and started a machine learning research group within the hardware organization.

Shuangfei Zhai (Apple)
Walter Talbott (Apple)
Carlos Guestrin (Apple & University of Washington)

Carlos Guestrin is the Director of Machine Learning at Apple and the Amazon Professor of Machine Learning in Computer Science and Engineering at the University of Washington. Carlos was the cofounder and CEO of Turi (formerly Dato and GraphLab), a startup that developed large-scale machine learning tools. A world-recognized leader in the field of machine learning, Carlos was named one of the 2008 Brilliant 10 by Popular Science. He received the 2009 IJCAI Computers and Thought Award for his contributions to artificial intelligence, and a Presidential Early Career Award for Scientists and Engineers (PECASE).

Related Events (a corresponding poster, oral, or spotlight)