We reconsider the training objective of Generative Adversarial Networks (GANs) from the mixed Nash Equilibria (NE) perspective. Inspired by the classical prox methods, we develop a novel algorithmic framework for GANs via an infinite-dimensional two-player ...
In this work, we propose a dynamical system based strategy for establishing a stable contact with convex shaped surfaces during non-contact/contact scenarios. A contact is called stable if the impact occurs only once and the robot remains in contact with t ...
In Proceedings of the Workshop on Towards Robots that Exhibit Manipulation Intelligence.2018
