ReGAN: RE[LAX|BAR|INFORCE] based Sequence Generation using GANs
This work addresses a specific challenge in natural language processing for researchers, but it is incremental as it builds on existing GAN methods for sequence generation.
The paper tackles the problem of generating longer text sequences using Generative Adversarial Networks (GANs), which have seen limited success in this domain, by applying efficient policy gradient estimators like REBAR, RELAX, and REINFORCE, resulting in improved training stability and quality of generated sequences on synthetic datasets with varying lengths.
Generative Adversarial Networks (GANs) have seen steep ascension to the peak of ML research zeitgeist in recent years. Mostly catalyzed by its success in the domain of image generation, the technique has seen wide range of adoption in a variety of other problem domains. Although GANs have had a lot of success in producing more realistic images than other approaches, they have only seen limited use for text sequences. Generation of longer sequences compounds this problem. Most recently, SeqGAN (Yu et al., 2017) has shown improvements in adversarial evaluation and results with human evaluation compared to a MLE based trained baseline. The main contributions of this paper are three-fold: 1. We show results for sequence generation using a GAN architecture with efficient policy gradient estimators, 2. We attain improved training stability, and 3. We perform a comparative study of recent unbiased low variance gradient estimation techniques such as REBAR (Tucker et al., 2017), RELAX (Grathwohl et al., 2018) and REINFORCE (Williams, 1992). Using a simple grammar on synthetic datasets with varying length, we indicate the quality of sequences generated by the model.