A New Metric-Based Approach to Model Selection

A New Metric-Based Approach to Model Selection (1997)

Cached

Download Links

[www.cis.upenn.edu]

Other Repositories/Bibliography

DBLP

by Dale Schuurmans

Venue:	In Proceedings of the Fourteenth National Conference on Artificial Intelligence (AAAI-97
Citations:	37 - 5 self

Versions

Version History

Metadata Version 3

User correction supplied by SystemCorrections

Datum	Value	Source
TITLE	A New Metric-Based Approach to Model Selection	user correction - Legacy Corrections
AUTHOR NAME	Dale Schuurmans	SVM HeaderParse 0.1
AUTHOR AFFIL	; Institute for Research in Cognitive Science; University of Pennsylvania; NEC Research Institute	SVM HeaderParse 0.2
AUTHOR ADDR	; Philadelphia, PA 19104-6228; 4 Independence Way; Princeton, NJ 08540	SVM HeaderParse 0.1
ABSTRACT	We introduce a new approach to model selection that performs better than the standard complexitypenalization and hold-out error estimation techniques in many cases. The basic idea is to exploit the intrinsic metric structure of a hypothesis space, as determined by the natural distribution of unlabeled training patterns, and use this metric as a reference to detect whether the empirical error estimates derived from a small (labeled) training sample can be trusted in the region around an empirically optimal hypothesis. Using simple metric intuitions we develop new geometric strategies for detecting overfitting and performing robust yet responsive model selection in spaces of candidate functions. These new metric-based strategies dramatically outperform previous approaches in experimental studies of classical polynomial curve fitting. Moreover, the technique is simple, efficient, and can be applied to most function learning tasks. The only requirement is access to an auxiliary collection ...	user correction - Legacy Corrections
YEAR	1997	INFERENCE
VENUE	In Proceedings of the Fourteenth National Conference on Artificial Intelligence (AAAI-97	INFERENCE
VENUE TYPE	CONFERENCE	INFERENCE
PAGES	552--558	INFERENCE
CITATIONS	20 found	ParsCit 1.0