Improved Training of Mixture-of-Experts Language GANs
This work addresses the problem of adversarial text generation for natural language processing, representing an incremental improvement over existing methods.
The paper tackles the challenge of training Generative Adversarial Networks (GANs) for generating human language by enhancing generator representation capacity with a mixture-of-experts approach and using Feature Statistics Alignment to provide fine-grained learning signals, resulting in superior performance on synthetic and real benchmarks.
Despite the dramatic success in image generation, Generative Adversarial Networks (GANs) still face great challenges in synthesizing sequences of discrete elements, in particular human language. The difficulty in generator training arises from the limited representation capacity and uninformative learning signals obtained from the discriminator. In this work, we (1) first empirically show that the mixture-of-experts approach is able to enhance the representation capacity of the generator for language GANs and (2) harness the Feature Statistics Alignment (FSA) paradigm to render fine-grained learning signals to advance the generator training. Specifically, FSA forces the mean statistics of the distribution of fake data to approach that of real samples as close as possible in the finite-dimensional feature space. Empirical study on synthetic and real benchmarks shows the superior performance in quantitative evaluation and demonstrates the effectiveness of our approach to adversarial text generation.