The Community for Technology Leaders
Green Image
Issue No. 04 - April (2013 vol. 35)
ISSN: 0162-8828
pp: 911-924
M. Ranjbar , Sch. of Comput. Sci., Simon Fraser Univ., Burnaby, BC, Canada
Tian Lan , Sch. of Comput. Sci., Simon Fraser Univ., Burnaby, BC, Canada
Yang Wang , Dept. of Comput. Sci., Univ. of Manitoba, Winnipeg, MB, Canada
S. N. Robinovitch , Sch. of Eng. Sci., Simon Fraser Univ., Burnaby, BC, Canada
Ze-Nian Li , Sch. of Comput. Sci., Simon Fraser Univ., Burnaby, BC, Canada
G. Mori , Sch. of Comput. Sci., Simon Fraser Univ., Burnaby, BC, Canada
ABSTRACT
We develop an algorithm for structured prediction with nondecomposable performance measures. The algorithm learns parameters of Markov Random Fields (MRFs) and can be applied to multivariate performance measures. Examples include performance measures such as $(F_{\beta })$ score (natural language processing), intersection over union (object category segmentation), Precision/Recall at k (search engines), and ROC area (binary classifiers). We attack this optimization problem by approximating the loss function with a piecewise linear function. The loss augmented inference forms a Quadratic Program (QP), which we solve using LP relaxation. We apply this approach to two tasks: object class-specific segmentation and human action retrieval from videos. We show significant improvement over baseline approaches that either use simple loss functions or simple scoring functions on the PASCAL VOC and H3D Segmentation datasets, and a nursing home action recognition dataset.
INDEX TERMS
Loss measurement, Piecewise linear approximation, Labeling, Training, Vectors, Prediction algorithms, Optimization,structural SVM, Optimization, large-margin
CITATION
M. Ranjbar, Tian Lan, Yang Wang, S. N. Robinovitch, Ze-Nian Li, G. Mori, "Optimizing Nondecomposable Loss Functions in Structured Prediction", IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. 35, no. , pp. 911-924, April 2013, doi:10.1109/TPAMI.2012.168
3447 ms
(Ver 3.3 (11022016))