Active learning and A-Optimal Experimental Design - Eldad Haber, Professor, UBC

Abstract:

In this work we discuss the problem of active learning. We present an approach that is based on A-optimal experimental design of ill-posed problems and show how one can optimally label a data set by partially probing it, and use it to train a model. We present two methods. The first is based on a Bayesian interpretation of the semi-supervised learning problem with the graph Laplacian and the second is based on a frequentist approach. The frequentist approach allows to slowly probe the data and adaptively label the data to reduce the error in the recovery of the labels.

Registrants will be sent the Zoom link closer to the event date. Please watch your contact email for that information.