CSAIL Event Calendar

Multi-task Learning and Structured Sparsity

A fundamental limitation of supervised learning is the cost incurred by the preparation of the large training samples required for good generalization. A potential remedy is offered by multi-task learning: in many cases, while individual sample sizes are rather small, there are samples to represent a large number of learning tasks, which share some constraining or generative property. If this property is sufficiently simple it should allow for better estimation of the individual tasks despite their small individual sample sizes. In this talk we review different classes of regularizers which implement task relatedness assumptions, building upon ideas from kernel methods and sparse estimation. We address the predictive properties of the methods and describe optimisation techniques to solve the underlying regularization problem. Finally, we comment on some potential applications of the methods in computer vision.

See other events happening in February 2013