VEEGAN - Reducing Mode Collapse in GANs Using Implicit Variational Learning
University of Central Florida via YouTube
Overview
This course teaches learners how to reduce mode collapse in Generative Adversarial Networks (GANs) using Implicit Variational Learning. The learning outcomes include understanding GANs, mode collapse, and how VEEGAN addresses this issue. Students will also learn about the reconstructor network, its objective function, and how to approximately invert the generator G. The course uses a theoretical approach to explain the concepts and includes practical results for synthetic datasets, stacked MNIST, and CIFAR. The intended audience for this course is individuals interested in deep learning, specifically in improving GAN performance.
Syllabus
What are Generative Adversarial Networks
What is mode collapse
How does VEEGAN address mode collapse?
How the reconstructor works
Approximately invert the generator Gyl
Reconstructor Network Objective Function
Results for the Synthetic datasets
Results for stacked MNIST
Stacked MNIST and CIFAR
Taught by
UCF CRCV