TY - JOUR
T1 - MILES
T2 - Multiple-instance learning via embedded instance selection
AU - Chen, Yixin
AU - Bi, Jinbo
AU - Wang, James Z.
N1 - Funding Information:
Y. Chen is supported by a Louisiana Board of Regents RCS program under Grant No. LBOR0077NR00C, the US National Science Foundation EPSCoR program under Grant No. NSF/ LEQSF(2005)-PFUND-39, the Research Institute for Children, and the University of New Orleans. This research work was conducted when Y. Chen was with the Department of Computer Science at University of New Orleans, New Orleans, Lousiana and when J. Bi was with the Department of Mathematical Sciences at Rensselaer Polytechnic Institute, 110 8th Street, Troy, New York. J.Z. Wang is supported by the US National Science Foundation under Grant Nos. IIS-0219272, IIS-0347148, and ANI-0202007, The Pennsylvania State University, and the PNC Foundation. The authors would like to thank the anonymous reviewers and the associate editor for their comments which have led to improvements of this paper. They would also like to thank Timor Kadir for providing the salient region detector and Rob Fergus for sharing the details on the object class recognition experiments.
PY - 2006
Y1 - 2006
N2 - Multiple-instance problems arise from the situations where training class labels are attached to sets of samples (named bags), instead of individual samples within each bag (called instances). Most previous multiple-instance learning (MIL) algorithms are developed based on the assumption that a bag is positive if and only if at least one of its instances is positive. Although the assumption works well in a drug activity prediction problem, it is rather restrictive for other applications, especially those in the computer vision area. We propose a learning method, MILES (Multiple-Instance Learning via Embedded instance Selection), which converts the multiple-instance learning problem to a standard supervised learning problem that does not impose the assumption relating instance labels to bag labels. MILES maps each bag into a feature space defined by the instances in the training bags via an instance similarity measure. This feature mapping often provides a large number of redundant or irrelevant features. Hence, 1-norm SVM is applied to select important features as well as construct classifiers simultaneously. We have performed extensive experiments. In comparison with other methods, MILES demonstrates competitive classification accuracy, high computation efficiency, and robustness to labeling uncertainty.
AB - Multiple-instance problems arise from the situations where training class labels are attached to sets of samples (named bags), instead of individual samples within each bag (called instances). Most previous multiple-instance learning (MIL) algorithms are developed based on the assumption that a bag is positive if and only if at least one of its instances is positive. Although the assumption works well in a drug activity prediction problem, it is rather restrictive for other applications, especially those in the computer vision area. We propose a learning method, MILES (Multiple-Instance Learning via Embedded instance Selection), which converts the multiple-instance learning problem to a standard supervised learning problem that does not impose the assumption relating instance labels to bag labels. MILES maps each bag into a feature space defined by the instances in the training bags via an instance similarity measure. This feature mapping often provides a large number of redundant or irrelevant features. Hence, 1-norm SVM is applied to select important features as well as construct classifiers simultaneously. We have performed extensive experiments. In comparison with other methods, MILES demonstrates competitive classification accuracy, high computation efficiency, and robustness to labeling uncertainty.
UR - http://www.scopus.com/inward/record.url?scp=33947180489&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=33947180489&partnerID=8YFLogxK
U2 - 10.1109/TPAMI.2006.248
DO - 10.1109/TPAMI.2006.248
M3 - Article
C2 - 17108368
AN - SCOPUS:33947180489
SN - 0162-8828
VL - 28
SP - 1931
EP - 1947
JO - IEEE Transactions on Pattern Analysis and Machine Intelligence
JF - IEEE Transactions on Pattern Analysis and Machine Intelligence
IS - 12
ER -