Skip to content

sacktock/SGHMC

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

290 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Reproducing Stochastic Gradient Hamiltonian Monte Carlo

This is a group project for the Advanced Topics in Machine Learning course at Oxford. We are reproducing the experiments in the paper "Stochastic Gradient Hamiltonian Monte Carlo" [1] by Tianqi Chen, Emily B. Fox and Carlos Guestrin.

We implemented the following algorithms to be used directly with Pyro - a universal probabilistic programming language (PPL) written in Python [2]: Hamiltonian Monte Carlo (HMC), Stochastic Gradient Hamiltonian Monte Carlo (SGHMC), Stochastic Gradient with Langevin Dynamics (SGLD), Stochastic Gradient Descent (SGD), and SGD with Nesterov momentum, Stochastic Gradient No U-Turn Sampler (SGNUTS).

  • bnn/ reproducing MNIST classification with Bayesian neural networks (BNN) from [1], with some additional experiments and demos.
  • demo.ipynb a simple demo that goes through the caveats of our implementation and to how get started using it.
  • examples/ a series of examples that demonstrate some of the algorithms and other options.
  • experiments/ reproducing Figure 1, Figure 2 and Figure 3 from [1].
  • kernel/ contains our implementations of HMC, SGHMC, SGLD, SGD, SGNUTS.

About

Implementation of Stochastic Gradient Hamiltonian Monte Carlo using Pyro (Probabilistic Programming in Python)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors