Active Sensing

Shipeng Yu, Balaji Krishnapuram, Romer Rosales, R. Bharat Rao
Proceedings of the Twelth International Conference on Artificial Intelligence and Statistics, PMLR 5:639-646, 2009.

Abstract

Labels are often expensive to get, and this motivates \emphactive learning which chooses the most informative samples for label acquisition. In this paper we study \emphactive sensing in a multi-view setting, motivated from many problems where grouped features are also expensive to obtain and need to be acquired (or \emphsensed) actively (e.g., in cancer diagnosis each patient might go through many tests such as CT, Ultrasound and MRI to get valuable features). The strength of this model is that one actively sensed (sample, view) pair would improve the \emphjoint multi-view classification on all the samples. For this purpose we extend the Bayesian co-training framework such that it can handle missing views in a principled way, and introduce two criteria for view acquisition. Experiments on one toy data and two real-world medical problems show the effectiveness of this model.

Cite this Paper


BibTeX
@InProceedings{pmlr-v5-yu09a, title = {Active Sensing}, author = {Yu, Shipeng and Krishnapuram, Balaji and Rosales, Romer and Rao, R. Bharat}, booktitle = {Proceedings of the Twelth International Conference on Artificial Intelligence and Statistics}, pages = {639--646}, year = {2009}, editor = {van Dyk, David and Welling, Max}, volume = {5}, series = {Proceedings of Machine Learning Research}, address = {Hilton Clearwater Beach Resort, Clearwater Beach, Florida USA}, month = {16--18 Apr}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v5/yu09a/yu09a.pdf}, url = {https://proceedings.mlr.press/v5/yu09a.html}, abstract = {Labels are often expensive to get, and this motivates \emphactive learning which chooses the most informative samples for label acquisition. In this paper we study \emphactive sensing in a multi-view setting, motivated from many problems where grouped features are also expensive to obtain and need to be acquired (or \emphsensed) actively (e.g., in cancer diagnosis each patient might go through many tests such as CT, Ultrasound and MRI to get valuable features). The strength of this model is that one actively sensed (sample, view) pair would improve the \emphjoint multi-view classification on all the samples. For this purpose we extend the Bayesian co-training framework such that it can handle missing views in a principled way, and introduce two criteria for view acquisition. Experiments on one toy data and two real-world medical problems show the effectiveness of this model.} }
Endnote
%0 Conference Paper %T Active Sensing %A Shipeng Yu %A Balaji Krishnapuram %A Romer Rosales %A R. Bharat Rao %B Proceedings of the Twelth International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2009 %E David van Dyk %E Max Welling %F pmlr-v5-yu09a %I PMLR %P 639--646 %U https://proceedings.mlr.press/v5/yu09a.html %V 5 %X Labels are often expensive to get, and this motivates \emphactive learning which chooses the most informative samples for label acquisition. In this paper we study \emphactive sensing in a multi-view setting, motivated from many problems where grouped features are also expensive to obtain and need to be acquired (or \emphsensed) actively (e.g., in cancer diagnosis each patient might go through many tests such as CT, Ultrasound and MRI to get valuable features). The strength of this model is that one actively sensed (sample, view) pair would improve the \emphjoint multi-view classification on all the samples. For this purpose we extend the Bayesian co-training framework such that it can handle missing views in a principled way, and introduce two criteria for view acquisition. Experiments on one toy data and two real-world medical problems show the effectiveness of this model.
RIS
TY - CPAPER TI - Active Sensing AU - Shipeng Yu AU - Balaji Krishnapuram AU - Romer Rosales AU - R. Bharat Rao BT - Proceedings of the Twelth International Conference on Artificial Intelligence and Statistics DA - 2009/04/15 ED - David van Dyk ED - Max Welling ID - pmlr-v5-yu09a PB - PMLR DP - Proceedings of Machine Learning Research VL - 5 SP - 639 EP - 646 L1 - http://proceedings.mlr.press/v5/yu09a/yu09a.pdf UR - https://proceedings.mlr.press/v5/yu09a.html AB - Labels are often expensive to get, and this motivates \emphactive learning which chooses the most informative samples for label acquisition. In this paper we study \emphactive sensing in a multi-view setting, motivated from many problems where grouped features are also expensive to obtain and need to be acquired (or \emphsensed) actively (e.g., in cancer diagnosis each patient might go through many tests such as CT, Ultrasound and MRI to get valuable features). The strength of this model is that one actively sensed (sample, view) pair would improve the \emphjoint multi-view classification on all the samples. For this purpose we extend the Bayesian co-training framework such that it can handle missing views in a principled way, and introduce two criteria for view acquisition. Experiments on one toy data and two real-world medical problems show the effectiveness of this model. ER -
APA
Yu, S., Krishnapuram, B., Rosales, R. & Rao, R.B.. (2009). Active Sensing. Proceedings of the Twelth International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 5:639-646 Available from https://proceedings.mlr.press/v5/yu09a.html.

Related Material