TY - GEN
T1 - RV-SVM
T2 - 13th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2009
AU - Hwanjo, Yu
AU - Youngdae, Kim
AU - Seungwon, Hwang
PY - 2009/7/23
Y1 - 2009/7/23
N2 - Learning ranking (or preference) functions has become an important data mining task in recent years, as various applications have been found in information retrieval. Among rank learning methods, ranking SVMhas been favorably applied to various applications, e.g., optimizing search engines, improving data retrieval quality. In this paper, we first develop a 1-norm ranking SVM that is faster in testing than the standard ranking SVM, and propose Ranking Vector SVM (RV-SVM) that revises the 1-norm ranking SVM for faster training. The number of variables in the RV-SVM is significantly smaller, thus the RV-SVM trains much faster than the other ranking SVMs.We experimentally compared the RV-SVM with the state-of-the-art rank learning method provided in SVM-light. The RV-SVMuses much less support vectors and trains much faster for nonlinear kernels than the SVM-light. The accuracies of RV-SVM and SVM-light are comparable on relatively large data sets. Our implementation of RV-SVM is posted at http://iis.postech.ac.kr/rv-svm.
AB - Learning ranking (or preference) functions has become an important data mining task in recent years, as various applications have been found in information retrieval. Among rank learning methods, ranking SVMhas been favorably applied to various applications, e.g., optimizing search engines, improving data retrieval quality. In this paper, we first develop a 1-norm ranking SVM that is faster in testing than the standard ranking SVM, and propose Ranking Vector SVM (RV-SVM) that revises the 1-norm ranking SVM for faster training. The number of variables in the RV-SVM is significantly smaller, thus the RV-SVM trains much faster than the other ranking SVMs.We experimentally compared the RV-SVM with the state-of-the-art rank learning method provided in SVM-light. The RV-SVMuses much less support vectors and trains much faster for nonlinear kernels than the SVM-light. The accuracies of RV-SVM and SVM-light are comparable on relatively large data sets. Our implementation of RV-SVM is posted at http://iis.postech.ac.kr/rv-svm.
UR - http://www.scopus.com/inward/record.url?scp=67650686496&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=67650686496&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-01307-2_39
DO - 10.1007/978-3-642-01307-2_39
M3 - Conference contribution
AN - SCOPUS:67650686496
SN - 3642013066
SN - 9783642013065
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 426
EP - 438
BT - 13th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2009
Y2 - 27 April 2009 through 30 April 2009
ER -