eduzhai > Applied Sciences > Engineering >

A Survey on Generative Adversarial Networks Variants Applications and Training

  • Save

... pages left unread,continue reading

Document pages: 38 pages

Abstract: The Generative Models have gained considerable attention in the field ofunsupervised learning via a new and practical framework called GenerativeAdversarial Networks (GAN) due to its outstanding data generation capability.Many models of GAN have proposed, and several practical applications emerged invarious domains of computer vision and machine learning. Despite GAN sexcellent success, there are still obstacles to stable training. The problemsare due to Nash-equilibrium, internal covariate shift, mode collapse, vanishinggradient, and lack of proper evaluation metrics. Therefore, stable training isa crucial issue in different applications for the success of GAN. Herein, wesurvey several training solutions proposed by different researchers tostabilize GAN training. We survey, (I) the original GAN model and its modifiedclassical versions, (II) detail analysis of various GAN applications indifferent domains, (III) detail study about the various GAN training obstaclesas well as training solutions. Finally, we discuss several new issues as wellas research outlines to the topic.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×