Active Learning for Probabilistic Structured Prediction of Cuts and Matchings

Sima Behpour, Anqi Liu, Brian Ziebart
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:563-572, 2019.

Abstract

Active learning methods, like uncertainty sampling, combined with probabilistic prediction techniques have achieved success in various problems like image classification and text classification. For more complex multivariate prediction tasks, the relationships between labels play an important role in designing structured classifiers with better performance. However, computational time complexity limits prevalent probabilistic methods from effectively supporting active learning. Specifically, while non-probabilistic methods based on structured support vector ma-chines can be tractably applied to predicting cuts and bipartite matchings, conditional random fields are intractable for these structures. We propose an adversarial approach for active learning with structured prediction domains that is tractable for cuts and matching. We evaluate this approach algorithmically in two important structured prediction problems: multi-label classification and object tracking in videos. We demonstrate better accuracy and computational efficiency for our proposed method.

Cite this Paper

BibTeX
@InProceedings{pmlr-v97-behpour19a, title = {Active Learning for Probabilistic Structured Prediction of Cuts and Matchings}, author = {Behpour, Sima and Liu, Anqi and Ziebart, Brian}, booktitle = {Proceedings of the 36th International Conference on Machine Learning}, pages = {563--572}, year = {2019}, editor = {Chaudhuri, Kamalika and Salakhutdinov, Ruslan}, volume = {97}, series = {Proceedings of Machine Learning Research}, month = {09--15 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v97/behpour19a/behpour19a.pdf}, url = {https://proceedings.mlr.press/v97/behpour19a.html}, abstract = {Active learning methods, like uncertainty sampling, combined with probabilistic prediction techniques have achieved success in various problems like image classification and text classification. For more complex multivariate prediction tasks, the relationships between labels play an important role in designing structured classifiers with better performance. However, computational time complexity limits prevalent probabilistic methods from effectively supporting active learning. Specifically, while non-probabilistic methods based on structured support vector ma-chines can be tractably applied to predicting cuts and bipartite matchings, conditional random fields are intractable for these structures. We propose an adversarial approach for active learning with structured prediction domains that is tractable for cuts and matching. We evaluate this approach algorithmically in two important structured prediction problems: multi-label classification and object tracking in videos. We demonstrate better accuracy and computational efficiency for our proposed method.} }
Endnote
%0 Conference Paper %T Active Learning for Probabilistic Structured Prediction of Cuts and Matchings %A Sima Behpour %A Anqi Liu %A Brian Ziebart %B Proceedings of the 36th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2019 %E Kamalika Chaudhuri %E Ruslan Salakhutdinov %F pmlr-v97-behpour19a %I PMLR %P 563--572 %U https://proceedings.mlr.press/v97/behpour19a.html %V 97 %X Active learning methods, like uncertainty sampling, combined with probabilistic prediction techniques have achieved success in various problems like image classification and text classification. For more complex multivariate prediction tasks, the relationships between labels play an important role in designing structured classifiers with better performance. However, computational time complexity limits prevalent probabilistic methods from effectively supporting active learning. Specifically, while non-probabilistic methods based on structured support vector ma-chines can be tractably applied to predicting cuts and bipartite matchings, conditional random fields are intractable for these structures. We propose an adversarial approach for active learning with structured prediction domains that is tractable for cuts and matching. We evaluate this approach algorithmically in two important structured prediction problems: multi-label classification and object tracking in videos. We demonstrate better accuracy and computational efficiency for our proposed method.
APA
Behpour, S., Liu, A. & Ziebart, B.. (2019). Active Learning for Probabilistic Structured Prediction of Cuts and Matchings. Proceedings of the 36th International Conference on Machine Learning, in Proceedings of Machine Learning Research 97:563-572 Available from https://proceedings.mlr.press/v97/behpour19a.html.

Related Material