This repository contains implementations for inference and learning in probabilistic graphical models from scratch. Detailed code and tutorials are provided for both Directed and Undirected Graphical Models, in MATLAB and Python. Additional algorithms may be added in the future.

Directed Graphical Models (Bayesian Networks)

• Belief Propagation Algorithm
• Variable Elimination
• Probabilistic Approximation Algorithms:
  • Likelihood Weighted Sampling
  • Monte Carlo Markov Chain (MCMC)
• Deterministic Approximation Algorithms:
  • Variational Mean-field Posterior Probability Inference
• (Learning) Maximum Likelihood Estimation (MLE)

Undirected Graphical Models (Markov and Conditional Random Fields)

• Belief Propagation:
  • Sum-Product Algorithm
  • Max-Sum Algorithm
• (Learning) Optimization Technique:
  • Stochastic Gradient Descent