A survey on text generation using generative adversarial networks

TL;DR

This survey reviews recent advancements in text generation using Generative Adversarial Networks, highlighting challenges and solutions like Gumbel-Softmax, Reinforcement Learning, and modified training methods.

Contribution

It provides a comprehensive overview of recent approaches and critically analyzes their objectives, methodologies, and experimental results in adversarial text generation.

Findings

Adversarial learning offers promising alternatives for natural language generation.

Challenges include handling discrete text data with GANs originally designed for continuous data.

Recent methods include Gumbel-Softmax, Reinforcement Learning, and modified training objectives.

Abstract

This work presents a thorough review concerning recent studies and text generation advancements using Generative Adversarial Networks. The usage of adversarial learning for text generation is promising as it provides alternatives to generate the so-called "natural" language. Nevertheless, adversarial text generation is not a simple task as its foremost architecture, the Generative Adversarial Networks, were designed to cope with continuous information (image) instead of discrete data (text). Thus, most works are based on three possible options, i.e., Gumbel-Softmax differentiation, Reinforcement Learning, and modified training objectives. All alternatives are reviewed in this survey as they present the most recent approaches for generating text using adversarial-based techniques. The selected works were taken from renowned databases, such as Science Direct, IEEEXplore, Springer, Association for Computing Machinery, and arXiv, whereas each selected work has been critically analyzed and assessed to present its objective, methodology, and experimental results.