TY - GEN
T1 - Meta-classifiers and selective superiority
AU - Benton, Ryan
AU - Kubat, Miroslav
AU - Loganantharaj, Rasaiah
PY - 2000
Y1 - 2000
N2 - Given that no one classification method is the best in all tasks, a variety of approaches have evolved to prevent poor performance due to mismatch of capabilities. One approach to overcome this problem is to determine when a method may be appropriate for a given problem. A second, more popular approach is to combine the capabilities of two or more classification methods. This paper provides some evidence that the combining of classifiers can yield more robust solutions.
AB - Given that no one classification method is the best in all tasks, a variety of approaches have evolved to prevent poor performance due to mismatch of capabilities. One approach to overcome this problem is to determine when a method may be appropriate for a given problem. A second, more popular approach is to combine the capabilities of two or more classification methods. This paper provides some evidence that the combining of classifiers can yield more robust solutions.
UR - http://www.scopus.com/inward/record.url?scp=84957869946&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84957869946&partnerID=8YFLogxK
U2 - 10.1007/3-540-45049-1_53
DO - 10.1007/3-540-45049-1_53
M3 - Conference contribution
AN - SCOPUS:84957869946
SN - 3540676899
SN - 9783540450498
VL - 1821
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 434
EP - 442
BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
PB - Springer Verlag
T2 - 13th International Conference on Industrial and Engineering Applications of Artificial Intelligence and Expert Systems, IEA/AIE 2000
Y2 - 19 June 2000 through 22 June 2000
ER -