Broad Area Colloquium For AI-Geometry-Graphics-Robotics-Vision
(CS 528)

A Unified Algorithmic Framework for Online Learning

Yoram Singer, Hebrew University
October 11, 2004, 4:15PM
TCSeq 200


A unified algorithmic framework for numerous problems in online learning is presented. The basic algorithm works by projecting an instantaneous hypothesis onto a single hyperplane which forms the basis for the next instantaneous hypothesis. In particular we discuss classification, regression, and uniclass problems. The analysis is based on simple convexity properties combined with mistake bound techniques. After describing the basic algorithmic setup we discuss a few extensions to more complex problems. Specifically, we describe online learning algorithms for hierarchical and multiclass categorization, rank-ordering learning, sequence prediction, and pseudo-metric learning. I will conclude with illustrations of a few systems developed at the Hebrew university that employ the online projection method.

* Based on joint works with Koby Crammer (UPenn), Ofer Dekel (HUJI), Shai Shwartz (HUJI), Joseph Keshet (HUJI), and Andrew Ng (Stanford).

About the Speaker

Professor Yoram Singer is a faculty member at the School of Computer Science and Engineering of The Hebrew University. His work focuses on the design, analysis, and implementation of machine learning algorithms.


Back to the Colloquium Page