Advanced Search
DSpace@MIT

Statistical Learning: Stability is Sufficient for Generalization and Necessary and Sufficient for Consistency of Empirical Risk Minimization

Research and Teaching Output of the MIT Community

Statistical Learning: Stability is Sufficient for Generalization and Necessary and Sufficient for Consistency of Empirical Risk Minimization

Download
Author: Mukherjee, Sayan; Niyogi, Partha; Poggio, Tomaso; Rifkin, Ryan
Date Issued: 2002-12-01
Abstract:
Solutions of learning problems by Empirical Risk Minimization (ERM) need to be consistent, so that they may be predictive. They also need to be well-posed, so that they can be used robustly. We show that a statistical form of well-posedness, defined in terms of the key property of L-stability, is necessary and sufficient for consistency of ERM.
Description:
revised July 2003
URI: http://hdl.handle.net/1721.3/5507
Other Identifiers: AIM-2002-024
CBCL-223
Series/Report no.: AIM-2002-024, CBCL-223
Keywords: AI, Theory of Learning, Great Discoveries, Consistency, ERM, Stability

Show full item metadata

Files in this item

Name Size Format Description
AIM-2002-024.ps 1.768Mb Postscript
AIM-2002-024.pdf 391.1Kb PDF

This item appears in the following Collection(s)

MIT-Mirage