The Future of Software Development: The Art of Code GAN-eration

Abstract

Code generation: automatic creation or completion of code, is still one of the most challenging problems which have lately garnered a lot of attention. There are abounding tools/methods available in the market that have tried to tackle this problem differently. Some key approaches include a transformer model, recurrent neural network, and abstract syntax tree. With the advent of the generative adversarial network (GAN), text generation has become even more popular. Many researchers are leveraging it to develop models and tools for automatic coding. Code generation using GANs has several applications, including code review, code synthesis, and program repair. In this paper, we have analyzed how GAN is being leveraged for code generation while also looking at the limitations and future of the same vis-à-vis other standard deep learning techniques in use currently. We found that while code generation using GANs has shown promising results, several challenges remain unresolved, including the generation of code that is both syntactically and semantically correct, as well as the need for large amounts of training data. However, with continued research in this area, code generation using GANs has the potential to revolutionize software development and accelerate the creation of new software applications.