Causality and Learning
Course Description: In many contexts, we need to be able to learn and reason about causal structures in our environment. Over the past thirty years, researchers from multiple fields—including philosophy, computer science, statistics, and epidemiology—have developed a computational framework and algorithms for causal learning and modeling. This course will first introduce students to the framework of probabilistic graphical models (including causal Bayesian networks, causal structural equation models, and others). We will then turn to a range of different structure learning and reasoning algorithms, focusing not just on the algorithms themselves, but also on their assumptions, power, and limitations. Given this foundation in causal modeling and learning, we will explore a range of recent research developments and open questions, with a particular eye towards understanding the suitability of causal modeling frameworks in different contexts, including learning and reasoning about psychological, neuroscientific, economic, and other causal structures.
Schedule of all courses