A Quality-Diversity Controllable GAN for Text Generation

Xingyu Lou, Kaihe Xu, Zhongliang Li, Tian Xia, Shaojun Wang, Jing Xiao

Keywords: text generation, GAN, quality-diversity, generalized Jensen-Shannon divergence

TL;DR: A GAN that can control quality-diversity trade-off through a single hyper-parameter and is more competitive with MLE model than other GANs variants.

Abstract: Text generation is a critical and difficult natural language processing task. Maximum likelihood estimate (MLE) based models have been arguably suffered from exposure bias in the inference stage and thus varieties of language generative adversarial networks (GANs) bypassing this problem have emerged. However, recent study has demonstrated that MLE models can constantly outperform GANs models over quality-diversity space under several metrics. In this paper, we propose a quality-diversity controllable language GAN.