Brigham Young University Homepage
Route Y Secure Sign In
College of Physical and Mathematical Sciences College of Physical and Mathematical Sciences

Spring Research Presentation

First Name:Jeremy
Last Name:West
Title:A Theoretical Foundation for Inductive Transfer
Department:Computer Science
Additional Authors:Dan Ventura, Sean Warnick
Faculty Advisor:Dan Ventura
Type:Talk
Abstract:
Inductive transfer refers to any algorithmic process by which structure or knowledge derived from a learning problem is used to enhance learning on a related problem. In this paper, we develop a formal description of inductive transfer using least squares regression for both linear and non-linear models. We then prove necessary and sufficient conditions under which inductive transfer is beneficial to learning accuracy. Finally, we empirically validate our result by demonstrating transfer in both artifical and real-world problems using several learning algorithms.

Maintained by the College of Physical and Mathematical Sciences Webmaster
Copyright © 2009. Brigham Young University. All Rights Reserved.