New face generation with GANs
Published:
Paper
UNSUPERVISED REPRESENTATION LEARNING WITH DEEP CONVOLUTIONAL GENERATIVE ADVERSARIAL NETWORKS
Important: the project does not implement the paper, but uses the paper to answer some questions. Paper is just as a reference for better understanding.
About
Generative adversial network is a big trend today and showed as a powerful tool for unsupervised learning. In the project we will implement a simple GAN to generate unexisting portrets.
From Wikipedia:
A generative adversarial network (GAN) is a class of machine learning frameworks designed by Ian Goodfellow and his colleagues in 2014. Two neural networks contest with each other in a game (in the form of a zero-sum game, where one agent’s gain is another agent’s loss).
We will use CelebA dataset for training.
Implementation
For the implementation we will use Python3 and PyTorch.
Environment
We use Jupyter notebooks to run the code and observe the results.
Code
All code is available on the Github