/approximate-inference

Expectation Maximisation, Variational Bayes, ARD, Loopy Belief Propagation, Gaussian Process Regression

Primary LanguagePython

Approximate Inference and Learning in Probabilistic Models

Assignment for Approximate Inference and Learning in Probabilistic Models (COMP0085) at UCL 2022

  • Variational Bayes & Automatic Relevance Determination (ARD)
  • Mean Field Learning
  • Loopy Belief Propagation
  • Gaussian Process Regression

To set up your python environment:

  1. Install poetry
pip install poetry
  1. Install dependencies
poetry install

Variational Bayes Automatic Relevance Determination:

Automatic Relevance Determination (8 Actual Latent Factors)

Mean Field Learning:

Learned Latent Factors

EM Free Energy

Loopy Belief Propagation:

Learned Latent Factors

EM Free Energy (doesn't converge)

Gaussian Process Regression:

CO2 Extrapolation

Kernel Hyper-Parameter Visualisation