Paper Info
Title
Relationship between diversity and correlation in multi-classifier systems
Abstract
Diversity plays an important role in the design of Multi-Classifier Systems, but its relationship to classification accuracy is still unclear from a theoretical perspective As a step towards the solution of this probelm, we take a different route and explore the relationship between diversity and correlation In this paper we provide a theoretical analysis and present a nonlinear function that relates diversity to correlation, which hence can be further related to accuracy This paper contributes to connecting existing research in diversity and correlation, and also providing a proxy to the relationship between diversity and accuracy Our experimental results reveal deeper insights into the role of diversity in Multi-Classifier Systems.
Year
DOI
Venue
2010
10.1007/978-3-642-13672-6_47
PAKDD (2)
Keywords
Field
DocType
theoretical perspective,important role,multi-classifier systems,deeper insight,multi-classifier system,nonlinear function,different route,classification accuracy,theoretical analysis,existing research
Data mining,Nonlinear system,Computer science,Correlation,Artificial intelligence,Classifier (linguistics),Machine learning
Conference
Volume
ISSN
ISBN
6119
0302-9743
3-642-13671-0
Citations 
PageRank 
References 
5
0.40
11
Authors
2
Name
Order
Citations
PageRank
Kuo-Wei Hsu1536.38
Jaideep Srivastava25845871.63