Paper Info
Title
The Problem with Noise and Small Disjuncts
Abstract
Many systems that learn from examples expressthe learned concept as a disjunction. Thosedisjuncts that cover only a few examples arereferred to as small disjuncts. The problem withsmall disjuncts is that they have a much highererror rate than large disjuncts but are necessary toachieve a high level of predictive accuracy. Thispaper investigates the effect of noise on smalldisjuncts. In particular, we show that when noiseis added to two real-world domains, a significant,and...
Year
Venue
Keywords
1998
ICML
small disjuncts
Field
DocType
ISBN
Pattern recognition,Computer science,Artificial intelligence,Machine learning
Conference
1-55860-556-8
Citations 
PageRank 
References 
20
2.93
5
Authors
2
Name
Order
Citations
PageRank
Gary M. Weiss11808105.84
Haym Hirsh21839277.74