Canada Research Chairs
Classical machine learning often requires large amounts of data, which can be very expensive. But interactive learning helps address this problem: while classical machine learning models assume that data have either been randomly selected or are extremely inadequate, interactive learning models assume carefully selected data.
Dr. Sandra Zilles, Canada Research Chair in Computational Learning Theory, is designing and analyzing formal models of interactive learning and developing algorithms that can solve complex learning problems efficiently using economical amounts of data. The models and algorithmic techniques that she and her research team provide may change the way machine learning is used. Although Zilles is aiming for theoretical guarantees, her findings could also lead to improvements in applied machine learning as well.