Abstract
We present a novel model called OCGAN for the classical problem of one-classnovelty detection, where, given a set of examples from a particular class, thegoal is to determine if a query example is from the same class. Our solution isbased on learning latent representations of in-class examples using a denoisingauto-encoder network. The key contribution of our work is our proposal toexplicitly constrain the latent space to exclusively represent the given class.In order to accomplish this goal, firstly, we force the latent space to havebounded support by introducing a tanh activation in the encoder's output layer.Secondly, using a discriminator in the latent space that is trainedadversarially, we ensure that encoded representations of in-class examplesresemble uniform random samples drawn from the same bounded space. Thirdly,using a second adversarial discriminator in the input space, we ensure allrandomly drawn latent samples generate examples that look real. Finally, weintroduce a gradient-descent based sampling technique that explores points inthe latent space that generate potential out-of-class examples, which are fedback to the network to further train it to generate in-class examples fromthose points. The effectiveness of the proposed method is measured across fourpublicly available datasets using two one-class novelty detection protocolswhere we achieve state-of-the-art results.