1: \begin{abstract}

2: Recently, sampling methods have been successfully applied to enhance the sample quality of Generative Adversarial Networks (GANs). However, in practice, they typically have poor sample efficiency because of the independent proposal sampling from the generator. 

3: % In this work, we propose REP-GAN, a novel sampling method with dependent proposal, which is further reparametrized through a pairing Markov Chain (CMC) in the latent space of the generator. 

4: In this work, we propose REP-GAN, a novel sampling method that allows general dependent proposals by REParameterizing the Markov chains

5: % with tractable dependent proposals reparameterized 

6: into the latent space of the generator.

7: % Theoretically, we show that given a perfect discriminator, and certain regularity conditions, our proposed method guarantees convergence to the true data distribution. 

8: Theoretically, we show that our reparameterized proposal admits a closed-form Metropolis-Hastings acceptance ratio.

9: Empirically, extensive experiments on synthetic and real datasets demonstrate that our REP-GAN largely improves the sample efficiency and obtains better sample quality simultaneously.

10: \end{abstract}

11: