-------------------------------
Title: Online learning and prediction: just play along!
Speaker: Purushottam Kar
Day: October 29, 2015, Thursday
Time: 5:15 PM
Venue: KD101 (CSE main)

Abstract:
Online learning and optimization techniques have formed the basis of some
of the most successful learning algorithms for large scale optimization,
deep learning etc. However, the roots of these algorithms can be traced to
a most beautiful interpretation of the process of machine learning as an
iterated game, one that is played between the teacher and the student.

This interpretation not only leads to a deep understanding of the learning
process, but also reveals the existence of several learning paradigms such
as passive learning, active learning and online learning. In this talk we
shall explore the world of online learning and prediction as an iterated
game. We will look at simple but powerful algorithms for online
prediction, as well as optimization.
-------------------------------