GAN-Implementation

Overview

Implementation of a Deep Convolutional Generative Adversarial Network using TensorFlow and Jupyter Notebook as final project for CSC 59929 Intro to Machine Learning at the City College of New York. This was able to effectively learn the distribution of the MNIST dataset and produce images that contain convincing handwritten digits.

Screenshots from training

Before training:

after 180,000 batches:

after 530,000 batches:

after 790,000 batches:

after 1,140,000 batches:

Usage (warning: can take a LLOOOONNNNNGGGGG time)

$ python DCGAN_MNIST.py

Dependencies

• TensorFlow
• Numpy
• MatplotLib
• Python3