What's on in Computing Science?

Date: Thursday, 26 February, 2009
Time: 11:00
Location: Sir Alwyn Williams Building, 422 Seminar Room
[ Inference Seminar ] Gaussian Processes for Multi-task Learning
Edwin Bonilla (University of Edinburgh)
Multi-task learning is an interesting scenario in machine learning where the learner aims to improve generalization by exploiting the shared information across different but related tasks. In this talk I will describe a class of multi-task learning models within the context of Gaussian processes. The main idea is that of modelling task dependencies directly so that predictions on one task are affected by the observations on the others. This is achieved by considering a shared covariance function on input features and a free-form covariance matrix over tasks, which allows for good flexibility when modelling inter-task dependencies while avoiding the need for large amounts of data for training. I will present applications of these models including compiler performance prediction, exam score prediction and learning of robot inverse dynamics.

Contact: Dr Rónán Daly (rdaly@dcs.gla.ac.uk)

Add to my calendar