/Improving-MCMC-by-Proposal-Learning

Implementation the article "GENERALIZING HAMILTONIAN MONTE CARLO WITH NEURAL NETWORKS"

Primary LanguageJupyter Notebook

Improving MCMC by Proposal Learning

Link to the paper: Generalizing Hamiltonian Monte Carlo with Neural Networks

Project goal

In the paper, the authors propose the generalization of Hamiltonian Monte Carlo method parametrized by deep neural networks. Following their idea, we seek to

  • build a model described in the paper;
  • reproduce the experiments of the authors;
  • apply the proposed sampling method to the task of text generation.

Team members: Anastasiia Baiandina, Kristina Rakova.