Long Text Generation via Adversarial Training with Leaked Information

TLDR

Automatic generation of coherent, semantically meaningful text is crucial for applications such as machine translation, dialogue systems, and image captioning, yet existing GAN approaches guided by a scalar reward only after full generation lack intermediate structural signals, limiting success for long passages beyond about 20 words. This work introduces LeakGAN, a framework designed to overcome these limitations in long‑text generation. LeakGAN lets the discriminator leak high‑level extracted features to the generator, which a Manager module uses to produce latent vectors that guide a Worker module at every generation step. Experiments on synthetic and real‑world tasks, including Turing tests, show that LeakGAN excels at long‑text generation, improves short‑text performance, and can implicitly learn sentence structure without supervision.

Abstract

Automatically generating coherent and semantically meaningful text has many applications in machine translation, dialogue systems, image captioning, etc. Recently, by combining with policy gradient, Generative Adversarial Nets (GAN) that use a discriminative model to guide the training of the generative model as a reinforcement learning policy has shown promising results in text generation. However, the scalar guiding signal is only available after the entire text has been generated and lacks intermediate information about text structure during the generative process. As such, it limits its success when the length of the generated text samples is long (more than 20 words). In this paper, we propose a new framework, called LeakGAN, to address the problem for long text generation. We allow the discriminative net to leak its own high-level extracted features to the generative net to further help the guidance. The generator incorporates such informative signals into all generation steps through an additional Manager module, which takes the extracted features of current generated words and outputs a latent vector to guide the Worker module for next-word generation. Our extensive experiments on synthetic data and various real-world tasks with Turing test demonstrate that LeakGAN is highly effective in long text generation and also improves the performance in short text generation scenarios. More importantly, without any supervision, LeakGAN would be able to implicitly learn sentence structures only through the interaction between Manager and Worker.