Open AccessIntroduction to Generative Adversarial Networks Challenges and Solutions
Author(s) -
Harmeet Kaur Khanuja,
AARTI AMOD AGARKAR
Publication year - 2021
Publication title -
international journal of next-generation computing
Language(s) - English
Resource type - Journals
eISSN - 2229-4678
pISSN - 0976-5034
DOI - 10.47164/ijngc.v12i5.468
Subject(s) - discriminator , computer science , artificial intelligence , generator (circuit theory) , generative grammar , deep learning , adversarial system , machine learning , artificial neural network , field (mathematics) , unsupervised learning , variation (astronomy) , mathematics , telecommunications , power (physics) , physics , quantum mechanics , detector , pure mathematics , astrophysics
Deep learning has received spectacular adoption in the field of artificial intelligence. With this many deep learning models have been developed. Generative Adversarial Networks (GAN) is one of the deep learning model, which is based on the two-player game from the Game theory. Two neural networks namely; generator and discriminator compete with each other. The proposition of the model variation is to achieve the data distribution through unsupervised learning to generate more realistic data. At present, GANs have been widely studied due to the extensive application prospects which include computer vision like generating lot of data, image to image translation etc. In this paper, the background of the GAN with theoretic model is introduced. Finally the existing research challenges of GANs are discussed with probable solutions giving a wider area of research.