The real-world deployment of machine learning models faces a series of lateral challenges affecting model trustworthiness, such as domain generalization, dataset shifts, causal validity, explainability, fairness, representativeness, and transparency. These challenges become increasingly important in techno-social systems affecting human high-stake decision making, which is often regulated by law. In this course, students will learn techniques for robust model evaluation, model selection, causal discovery, explainable and fair artificial intelligence, and interpretable models. In addition, students will reason about representativeness, transparency, and legal aspects of techno-social systems. The course will review both cutting-edge research and relevant portions of recent open-access textbooks. Coursework includes reading recent research papers, programming assignments, and a final group project. After completing the course, students should be able to develop, investigate, evaluate, and deploy artificial intelligence systems more responsibly.
Open to Graduate Computer Science students only. AS THE COURSE HAS INTERDISCIPLINARY PARTS, TOWARD STATISTICS AND BEHAVIORAL SCIENCES, GRADUATE STUDENTS FROM OTHER DEPARTMENTS MAY ENROLL, IF SEATS ARE AVAILABLE. FOLLOWING BACKGROUND IS ASSUMED: INTRODUCTORY MACHINE LEARNING OR DATA SCIENCE (COMPSCI 348 OR 383 OR 389 OR 589 OR EQUIVALENT), BASIC STATISTICS (COMPSCI 240 OR STAT 240 OR PSYCH 240 OR OIM 240 OR STAT 515 OR EQUIVALENT), GOOD PROGRAMMING SKILLS IN PYTHON (LIBRARIES: NUMPY, PANDAS, MATPLOTLIB, SKLEARN). STUDENTS OUTSIDE OF CS MUST REQUEST OVERRIDES VIA THE ON-LINE FORM AND DESCRIBE HOW THEY MEET PREREQUISITES UNDER "ADDITIONAL INFORMATION": https://www.cics.umass.edu/overrides. STUDENTS WHO ARE UNCERTAIN WHETHER THEIR BACKGROUND IS SUFFICIENT SHOULD EMAIL THE INSTRUCTOR ATTACHING THEIR ACADEMIC TRANSCRIPTS TO DETERMINE WHETHER THEY HAVE THE APPROPRIATE BACKGROUND TO BE SUCCESSFUL IN THE COURSE.
Permission is required for interchange registration during the add/drop period only.
