macio-matheus / gan-network-experiment

Gan Network Experiments

Gan Network Experiments

Generative Adversary Networks (GANs) are architectures of deep neural networks composed of two networks pitted against each other. This is one of the newest and most fascinating architectures in Deep Learning.

This project aims to show the application of the network in some conventional datasets.

Basic architecture of the GAN network:

Result of Mnist

Result after 1000 seasons in the mnist dataset

Result of Fashion Mnist

Result after 2000 seasons in the Fashion Mnist dataset

Usage

Run with docker compose

 cd gan-network-experiment
 docker-compose up -d --build

Go to #local-ip#:8111/lab and run the "gan-network-experiments.ipynb" notebook to observe the exploration process

TODO:

- Include more experiments with different datasets