Teaching to Achieve Meta-Learning

Having taught courses in machine learning, one of the fastest-growing subfields of computer science where paradigm shifts in modeling and applications occur every couple of months (!), I attempt to summarize my philosophy on teaching through the term, meta-learning. In his seminal book on Machine Learning, Tom Mitchell briefly defines learning: A computer program is said to learn from ex perience E with respect to some class of tasks T and performance measure P, if its performance on tasks in T, as measured by P, improves with experience E. In general, students generally gain knowledge through lectures, recitations, and home- work (experience E) on topics predefined in a syllabus (tasks T) with the goal that they improve on some measurable score (performance P) with increased experience E. However, this may break down for courses meant to tackle the most recent research topics. Whenever new shifts in knowledge happen (new tasks in T), instructors may have to redesign the course and students have to learn these new concepts (experience E).

Meta-learning, as its name suggests, aims to teach students the process of learning to learn, rather than what to learn. In a similar vein as above, meta-learning can be defined as: A computer program is said to meta-learn from experience E with respect to some class of taskns T and performance measure P, if its performance on new tasks from a similar distribution but not seen in T, as measured by P, improves with experience E. How can we structure teaching such that we achieve student meta-learning? In my teaching, I have emphasized teaching fundamental concepts that permeate past, current, and future methods, and encouraged curiosity to seek out open questions and accomplish open-ended research. Finally, achieving the above at scale through enabling inclusive access to course materials and long-term mentoring of junior students.

I'm honored to have had the opportunity to collaborate and learn from the best faculty: Louis-Philippe Morency, Ruslan Salakhutdinov, Manuel Blum, Lenore Blum, Yonatan Bisk, Daniel Fried, and many fellow TAs and students. To my students who have now embarked on exciting journeys in academia and industry: I'm proud of you and I thank you for giving me the opportunity to teach!