Uri Shalit - Causal Inference and Machine Learning: a Two-Way Street (IS Colloquium)
Where does causal thinking meet machine learning? We will discuss several such cases. We first show how we use learning theory to guide us in building algorithms for inferring individual-level causal effects, and how we apply these ideas to create deep-learning causal-effect inference methods. We then show how ideas from causal inference can help us in two important machine learning tasks: learning robust classifiers and interpreting deep image recognition system. If time permits, we’ll discuss a recent application of machine learning for learning individualized treatments for patients in an acute hospital setting.