Bayesian Statistics for the Physical Sciences

Course Description

This course provides an introduction to Bayesian statistics with a focus on applications in physical sciences and engineering. Emphasis will be placed on casting Bayesian statistics as a framework for quantitative scientific reasoning and discovery in the context of physical models. Students will learn to apply Bayesian logic and methodology to set up and solve inference and decision problems in scientific contexts. The course covers fundamental concepts such as conditional and marginal probability, Bayesian inference, latent variables, locally and globally informative data summaries, missing data problems, model comparison, and experimental design. Computational techniques covered will include Markov Chain Monte Carlo methods and variational approaches. We will discuss ML approaches to solving traditionally intractable models that are specified implicitly through physical model simulations including variational techniques with neural density estimators.

Code

EN.553.635

Credits

3

Subject

Applied Mathematics & Statistics

Programs That Include This Course

• Master's Applied Mathematics and Statistics, Master of Science in Engineering
• Master's Data Science, Master of Science in Engineering