Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/pierresegonne/VINF

Repository for DTU Special Course, focusing on Variational Inference using Normalizing Flows (VINF). Supervised by Michael Riis Andersen
https://github.com/pierresegonne/VINF

distributions python3 tensorflow variational-inference

Last synced: 2 months ago
JSON representation

Repository for DTU Special Course, focusing on Variational Inference using Normalizing Flows (VINF). Supervised by Michael Riis Andersen

Host: GitHub
URL: https://github.com/pierresegonne/VINF
Owner: pierresegonne
Created: 2020-03-02T12:34:31.000Z (almost 5 years ago)
Default Branch: master
Last Pushed: 2020-06-11T20:40:25.000Z (over 4 years ago)
Last Synced: 2024-11-16T03:22:14.679Z (2 months ago)
Topics: distributions, python3, tensorflow, variational-inference
Language: Python
Homepage: https://pierresegonne.github.io/VINF/
Size: 86.4 MB
Stars: 25
Watchers: 4
Forks: 3
Open Issues: 1
Metadata Files:
- Readme: README.md

Awesome Lists containing this project

awesome-normalizing-flows - Variational Inference using Normalizing Flows (VINF)

README

        ![ex_space_warp](https://raw.githubusercontent.com/pierresegonne/VINF/master/space_warp/Figure_2%20copy.png)

# Variational Inference using Normalizing Flows (VINF)

This repository provides a hands-on `tensorflow` implementation of Normalizing Flows as presented in the [paper](https://arxiv.org/pdf/1505.05770.pdf)

introducing the concept (D. Rezende & S. Mohamed). This code was developed as part of a Special Course at DTU (Denmarks Tekniske Universitet), supervised

by Michael Riis Andersen. The final report of the course, that details all experiments run with this repository can directly be accessed at [https://pierresegonne.github.io/VINF/](https://pierresegonne.github.io/VINF/)

## Implementation

This repository provides an implementation of 

- ADVI (Automatic Differential Variational Inference, with Diagonal Gaussian, baseline) 

- Planar Flow

- Radial Flow

## Demonstrative distributions 

True posterior

![true_energies](https://raw.githubusercontent.com/pierresegonne/VINF/master/assets/energy_densities.png)

Samples generated from the trained variational approximation

![energy_1](https://raw.githubusercontent.com/pierresegonne/VINF/master/assets/energy_1_pf_hexbin.png)

![energy_2](https://raw.githubusercontent.com/pierresegonne/VINF/master/assets/energy_2_pf_hexbin.png)

![energy_3](https://raw.githubusercontent.com/pierresegonne/VINF/master/assets/energy_3_pf_hexbin.png)

![energy_4](https://raw.githubusercontent.com/pierresegonne/VINF/master/assets/energy_4_pf_hexbin.png)

## TODO

- [ ] Run additional experiments on radial flows

- [ ] Add requirements.txt

- [ ] Improve models with the use of bijectors. See [this thread](https://stackoverflow.com/questions/61717694/embed-trainable-bijector-into-keras-model/62284510#62284510) for a starting point

- [ ] Include new flow models.

  - [ ] Glow [paper](https://papers.nips.cc/paper/8224-glow-generative-flow-with-invertible-1x1-convolutions.pdf)

  - [ ] MAF [paper](https://papers.nips.cc/paper/6581-improved-variational-inference-with-inverse-autoregressive-flow.pdf)

  - [ ] IAF [paper](https://papers.nips.cc/paper/6828-masked-autoregressive-flow-for-density-estimation.pdf)