jgsimard / pgm-pytorch

Probabilstic graphical models in Pytorch

The goal of this repository is to provide a pytorch implementation of the main probabilistic graphical models and accompanying notebooks to explain the theory behind the models, show a peek of the implementation and a use case.

TODO

Implementations

• KMeans
• Gaussiam Mixture Model
• Hidden Marvov Model
  • Multinomial Emission
  • Gaussian Emission
• Factor Analysis
• Principal Component Analysis
• Kalman Filters
  • Basic
  • Extended Kalman Filter
  • Uscented Kalman Filter
• CRF
  • Linear
  • Fully Connected
• MCMC - Gibbs sampling on ising model
• Mean Field on ising model
• LDA
• Particle filters

Notebooks

• KMeans
• Gaussiam Mixture Model
• Hidden Marvov Model
• Factor Analysis
• Kalman Filters
• CRFs
• MCMC
• Variational inference