Online Passive-Aggressive Algorithms

This week we will cover the paper –
Online Passive-Aggressive Algorithms
by Crammer et al. This group will be especially interesting thanks to Koby who has kindly agreed to share some insights into this work.

Abstract
We present a family of margin based online learning algorithms for various prediction tasks. In particular we derive and analyze algorithms for binary and multiclass categorization, regression, uniclass prediction and sequence prediction. The update steps of our different algorithms are all based on analytical solutions to simple constrained optimization problems. This unified view allows us to prove worst-case loss bounds for the different algorithms and for the various decision problems based on a single lemma. Our bounds on the cumulative loss of the algorithms are relative to the smallest loss that can be attained by any fixed hypothesis, and as such are applicable to both realizable and unrealizable settings. We demonstrate some of the merits of the proposed algorithms in a series of experiments with synthetic and real data sets.

We have moved the time back to 9:30.

See you on Wednesday (2/1/2012)!

Leave a Reply

Your email address will not be published. Required fields are marked *