M. Pawan Kumar 
HOME

EFFICIENT OPTIMIZATION FOR AVERAGE PRECISION SVM P. Mohapatra, C. V. Jawahar and M. Pawan Kumar In Proceedings of Advances in Neural Information Processing Systems (NIPS), 2014 The accuracy of information retrieval systems is often measured using average precision (AP). Given a set of positive (relevant) and negative (nonrelevant) samples, the parameters of a retrieval system can be estimated using the APSVM framework, which minimizes a regularized convex upper bound on the empirical AP loss. However, the high computational complexity of lossaugmented inference, which is required for learning an APSVM, prohibits its use on large training datasets. To alleviate this deficiency, we propose three complementary approaches. The first approach guarantees an asymptotic decrease in the computational complexity of lossaugmented inference by exploiting the problem structure. The second approach takes advantage of the fact that we do not require a full ranking during lossaugmented inference. This helps us to avoid the expensive step of sorting the negative samples according to their individual scores. The third approach approximates the AP loss over all samples by the AP loss over difficult samples (for example, those that are incorrectly classified by a binary SVM), while ensuring the correct classification of the remaining samples. Using the PASCAL VOC action classification dataset, we show that our approaches provide significant speedups during training without degrading the test accuracy of APSVM. 