An introduction to Bayesian methods in ML

A workshop introducing basic Bayesian modelling, using practical examples and probabilistic programming.

Learning outcomes

  • Understanding the Bayesian methodology, and the foundations of approximate bayesian inference.
  • First steps in probabilistic programming in pyro.
  • Getting to know the model based machine learning approach:
    • Learn to make the right assumptions.
    • Criticise your own model.
  • Getting to know many model types and comparing their performance.
  • Applying these models to a wide range of areas.
  • Combining probabilistic modelling with deep learning.

Structure of the workshop

Part 1

  • Introduction to Bayesian methods and probabilistic programming languages.
  • Inference with discrete variables: Who has the right skill set for your job offering?
  • Inference with Markov chain Monte Carlo methods: Marriages, divorces and causality.
  • The Lotka-Volterra predator-prey model. Advantages and necessity of hybrid approaches combining deterministic and statistical models.

Part 2

  • Stochastic variational inference and the mean field assumption.
  • Clustering images with mixture models and latent Dirichlet allocation like models.
  • Understanding the development of asthma through Markov chains.
  • Become a wine connoisseur with Bayesian neural networks.
  • Modeling faces through the ages with variational autoencoders.

In this series