Wednesday, 29.3.2017, 11:30
We study the problem of transferring a sample in one domain to an analog sample in another domain. Given two related domains, S and T, we would like to learn a generative function G that maps an input sample from S to the domain T, such that the output of a given function f, which accepts inputs in either domains, would remain unchanged. Other than the function f, the training data is unsupervised and consist of a set of samples from each domain. The Domain Transfer Network (DTN) we present employs a compound loss function that includes a multiclass GAN loss, an f-constancy component, and a regularizing component that encourages G to map samples from T to themselves. We apply our method to visual domains including digits and face images and demonstrate its ability to generate convincing novel images of previously unseen entities, while preserving their identity.
Adam Polyak is a research engineer at Facebook AI Research(FAIR) in Facebook Tel-Aviv, where he works on developing and understanding systems with human level intelligence. Before joining FAIR, Adam completed his M.Sc in computer science at Tel-Aviv University, under the supervision of Prof. Lior Wolf. His thesis focused on methods to accelerate and compress neural networks to allow their deployment on devices with limited computational power.