The Community for Technology Leaders
RSS Icon
Subscribe
Issue No.12 - December (2006 vol.28)
pp: 1931-1947
Jinbo Bi , IEEE
Yixin Chen , IEEE
ABSTRACT
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.
INDEX TERMS
Multiple-instance learning, feature subset selection, 1-norm support vector machine, image categorization, object recognition, drug activity prediction.
CITATION
Jinbo Bi, Yixin Chen, "MILES: Multiple-Instance Learning via Embedded Instance Selection", IEEE Transactions on Pattern Analysis & Machine Intelligence, vol.28, no. 12, pp. 1931-1947, December 2006, doi:10.1109/TPAMI.2006.248
23 ms
(Ver 2.0)

Marketing Automation Platform Marketing Automation Tool