SeqGAN: Sequence Generative Adversarial Nets with Policy Gradient

TL;DR

SeqGAN introduces a reinforcement learning-based framework for sequence generation that effectively trains generative models with discrete outputs, overcoming the gradient passing issues faced by traditional GANs.

Contribution

It proposes a novel sequence generation method, SeqGAN, which uses policy gradients and Monte Carlo search to improve training of discrete sequence models.

Findings

Significant improvements over baseline models on synthetic data.

Effective training of sequence models with discrete tokens.

Demonstrated success on real-world sequence generation tasks.

Abstract

As a new way of training generative models, Generative Adversarial Nets (GAN) that uses a discriminative model to guide the training of the generative model has enjoyed considerable success in generating real-valued data. However, it has limitations when the goal is for generating sequences of discrete tokens. A major reason lies in that the discrete outputs from the generative model make it difficult to pass the gradient update from the discriminative model to the generative model. Also, the discriminative model can only assess a complete sequence, while for a partially generated sequence, it is non-trivial to balance its current score and the future one once the entire sequence has been generated. In this paper, we propose a sequence generation framework, called SeqGAN, to solve the problems. Modeling the data generator as a stochastic policy in reinforcement learning (RL), SeqGAN bypasses the generator differentiation problem by directly performing gradient policy update. The RL reward signal comes from the GAN discriminator judged on a complete sequence, and is passed back to the intermediate state-action steps using Monte Carlo search. Extensive experiments on synthetic data and real-world tasks demonstrate significant improvements over strong baselines.