1: \begin{abstract}

2: %  Generative adversarial networks(GAN) are drawing attention not only for good performance on generating data, but also for various applications such as semi-supervised learning, reinforcement learning etc. Behind its usefulness and potential, GANs are still less understood. Convergence and Stability issue of non minmax case are especially so. Various approaches have been proposed to solve such problem, and one promising approach is probabilistic modeling with sampling update. Training GAN by sampling may be accompanied by training of multiple models. So, it will be necessary to learn them sparse to save parameters for real world application.  We propose a new GAN training algorithm that theoretically guarantees convergence and can be applied for sparse model.

3:     

4:     The Generative Adversarial Net (GAN) is a highly effective generative model when the process of generating data, such as images, videos and text, is ambiguous. However, it is very difficult to train due to an instability issue known as model collapse, i.e., there is a lack of diversity among generated data. In this paper, we propose a new method to train GAN using an adaptive momentum stochastic gradient Langevin dynamics algorithm, which provides a theoretical guarantee for the convergence of the generator to the underlying data distribution. The numerical results favor to the proposed algorithm.  

5: \end{abstract}

6: