https://github.com/timudk/gmin

Jupyter notebooks for generative models using NumPy
https://github.com/timudk/gmin

Last synced: 5 days ago
JSON representation

Jupyter notebooks for generative models using NumPy

• Host: GitHub
• URL: https://github.com/timudk/gmin
• Owner: timudk
• License: mit
• Created: 2019-08-20T14:55:27.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2019-09-04T13:05:08.000Z (over 5 years ago)
• Last Synced: 2024-12-22T14:26:49.855Z (about 2 months ago)
• Language: Jupyter Notebook
• Homepage:
• Size: 173 KB
• Stars: 1
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Generative models in NumPy

This repository contains implementations of generative models using NumPy.

## Available implementations

1. Naive Bayes (classifier)

2. Boltzmann machine

3. Restricted Boltzmann machine

## In progress

1. GAN

## Motivation

In my opinion, the best way to grasp a new generative model is by implementing it in NumPy. Only by doing so, one can understand the full picture of (probabilisitc) model assumptions, optimization and sampling. 

## Sources of inspiration

1. [Danilo Rezende's slides on deep generative models](https://docs.google.com/presentation/d/e/2PACX-1vSwRVxRHDarUx2mwXrsrlrtdTVTyEiFkWjJ9TvJ5ad6gbB3PDZSgD9yHAUiB6DcO1zP7LXBpxzc0SzC/pub?start=true&loop=true&delayms=10000&slide=id.gd9c453428_0_16)

2. Papers on generative models:

  * Introducing GANs: [http://papers.nips.cc/paper/5423-generative-adversarial-nets.pdf](http://papers.nips.cc/paper/5423-generative-adversarial-nets.pdf)

  * Introducing Variational autoencoders: [Auto-Encoding Variational Bayes](https://arxiv.org/pdf/1312.6114.pdf)

  * [A Practical Guide to Training Restricted Boltzmann Machines](https://www.cs.toronto.edu/~hinton/absps/guideTR.pdf)

  * [Stochastic Backpropagation and Approximate Inferencein Deep Generative Models](https://arxiv.org/pdf/1401.4082.pdf)

  * Introducing normalizing flows: [Variational Inference with Normalizing Flows](https://arxiv.org/pdf/1505.05770.pdf)