Chapter 2 - GAN models in natural language processing and image translation

Abstract

Recently, generative adversarial networks (GANs) have actively been used in various research fields such as natural language processing, image generation, translation, cyber security, medicine, geo-informatics, etc. But in the field of Natural language processing and Image generation and translation, GANs have shown their remarkable performance when compared to other fields. This chapter provides insights on various GAN models classified based on their learning mechanism. The comparison of GAN models based on activation function, optimization function, and loss functions has been outlined. Further, how these GAN models are adapted for various applications of natural language processing, image generation, and translation are also discussed. This chapter also discusses the comparison of GAN models among NLP, image generation, and translation. This chapter outlines the various NLP and image datasets available for research. The evaluation metrics to evaluate the GAN performance, tools, and the languages used for GAN research are also discussed in this chapter. Finally, it discusses the open challenges for further research.