This repository is dedicated to advanced explorations in neural network models, focusing on Variational Autoencoders (VAEs) and Generative Adversarial Networks (GANs), utilizing a uniquely modified MNIST dataset where digits are colored. It encompasses detailed exercises on three VAE models: continuous, discrete, and combined. Additionally, the repository features a Cycle GAN exercise designed to generate digits with colors associated with other digits, demonstrating the transformative capabilities of GANs in a research and academic context. The repository has two folders: one for VAE exercises and the other for the Cycle GAN exercise. In addition, the repository contains a comprehensive report and instructions files in PDF format.
yonatanko / coloredmnist-vae-gan Goto Github PK
View Code? Open in Web Editor NEWAdvanced Study of VAEs and GANs using a Colored MNIST Dataset.