Probabilistic-Programming & Bayesian Data Analysis

Probabilistic Programming and Bayesian Data Analysis (COGS516 - METU - Graduate Informatics Course)

This course will examine the basic principles of probabilistic programming and Bayesian modelling, for analysis of data which may come from observational or experimental cognitive science studies. A variety of Bayesian data analysis will be discussed and implemented in an expressive probabilistic programming language. Approaches for model building, model checking and model validation will be discussed following a Bayesian workflow.

Introduction to Probabilistic Programming; generative modelling, Bayesian inference, executing probabilistic programs; exact inference; rejection sampling; importance sampling; Markov Chain Monte Carlo (MCMC); efficient MCMC Techniques; deep Probabilistic Programming; Probabilistic Programming applications.

1. Probability and Inference
2. Regression
3. Categories and Curves
4. Causality: Confounders and Colliders
5. Overfitting and model comparison
6. Inference, Markov Chain Monte Carlo, Assessing Convergence
7. Generalized Linear Models
8. Multi-level Models
9. Ordered categorical outcomes
10. Gaussian Processes
11. Review, Further Topics in PPL