Rigorous Systems Research Group (RSRG) Seminar

Thursday October 18, 2018 12:00 PM

Avoiding the Pitfalls of Active Learning with Robust Predictors for Covariate Shift

Speaker: Angie Liu, Computing & Mathematical Science, Caltech
Location: Annenberg 213

Pool-based active learning promises to significantly reduce the labeling burden of black-box supervised machine learning methods, but often doesn't deliver in practice.  In fact, standard active learning techniques frequently provide worse performance than passive learners (i.e., datapoints labeled at random).  This talk will illuminate the fundamental issue of covariate shift hindering pool-based active learning methods, present a new approach using adversarial estimation for addressing it, demonstrate the benefits of the approach on classification tasks, and discuss extensions of this idea for other prediction tasks.

Series Rigorous Systems Research Group (RSRG) Seminar Series

For more information visit: http://cms.caltech.edu/seminars