Nonmyopic ε-Bayes-Optimal Active Learning of Gaussian Processes

Trong Nghia Hoang, Bryan Kian Hsiang Low, Patrick Jaillet, Mohan Kankanhalli
Proceedings of the 31st International Conference on Machine Learning, PMLR 32(2):739-747, 2014.

Abstract

A fundamental issue in active learning of Gaussian processes is that of the exploration-exploitation trade-off. This paper presents a novel nonmyopic ε-Bayes-optimal active learning (ε-BAL) approach that jointly and naturally optimizes the trade-off. In contrast, existing works have primarily developed myopic/greedy algorithms or performed exploration and exploitation separately. To perform active learning in real time, we then propose an anytime algorithm based on ε-BAL with performance guarantee and empirically demonstrate using synthetic and real-world datasets that, with limited budget, it outperforms the state-of-the-art algorithms.

Cite this Paper

BibTeX
@InProceedings{pmlr-v32-hoang14, title = {Nonmyopic $\epsilon$-Bayes-Optimal Active Learning of Gaussian Processes}, author = {Hoang, Trong Nghia and Low, Bryan Kian Hsiang and Jaillet, Patrick and Kankanhalli, Mohan}, booktitle = {Proceedings of the 31st International Conference on Machine Learning}, pages = {739--747}, year = {2014}, editor = {Xing, Eric P. and Jebara, Tony}, volume = {32}, number = {2}, series = {Proceedings of Machine Learning Research}, address = {Bejing, China}, month = {22--24 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v32/hoang14.pdf}, url = {https://proceedings.mlr.press/v32/hoang14.html}, abstract = {A fundamental issue in active learning of Gaussian processes is that of the exploration-exploitation trade-off. This paper presents a novel nonmyopic ε-Bayes-optimal active learning (ε-BAL) approach that jointly and naturally optimizes the trade-off. In contrast, existing works have primarily developed myopic/greedy algorithms or performed exploration and exploitation separately. To perform active learning in real time, we then propose an anytime algorithm based on ε-BAL with performance guarantee and empirically demonstrate using synthetic and real-world datasets that, with limited budget, it outperforms the state-of-the-art algorithms.} }
Endnote
%0 Conference Paper %T Nonmyopic ε-Bayes-Optimal Active Learning of Gaussian Processes %A Trong Nghia Hoang %A Bryan Kian Hsiang Low %A Patrick Jaillet %A Mohan Kankanhalli %B Proceedings of the 31st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2014 %E Eric P. Xing %E Tony Jebara %F pmlr-v32-hoang14 %I PMLR %P 739--747 %U https://proceedings.mlr.press/v32/hoang14.html %V 32 %N 2 %X A fundamental issue in active learning of Gaussian processes is that of the exploration-exploitation trade-off. This paper presents a novel nonmyopic ε-Bayes-optimal active learning (ε-BAL) approach that jointly and naturally optimizes the trade-off. In contrast, existing works have primarily developed myopic/greedy algorithms or performed exploration and exploitation separately. To perform active learning in real time, we then propose an anytime algorithm based on ε-BAL with performance guarantee and empirically demonstrate using synthetic and real-world datasets that, with limited budget, it outperforms the state-of-the-art algorithms.
RIS
TY - CPAPER TI - Nonmyopic ε-Bayes-Optimal Active Learning of Gaussian Processes AU - Trong Nghia Hoang AU - Bryan Kian Hsiang Low AU - Patrick Jaillet AU - Mohan Kankanhalli BT - Proceedings of the 31st International Conference on Machine Learning DA - 2014/06/18 ED - Eric P. Xing ED - Tony Jebara ID - pmlr-v32-hoang14 PB - PMLR DP - Proceedings of Machine Learning Research VL - 32 IS - 2 SP - 739 EP - 747 L1 - http://proceedings.mlr.press/v32/hoang14.pdf UR - https://proceedings.mlr.press/v32/hoang14.html AB - A fundamental issue in active learning of Gaussian processes is that of the exploration-exploitation trade-off. This paper presents a novel nonmyopic ε-Bayes-optimal active learning (ε-BAL) approach that jointly and naturally optimizes the trade-off. In contrast, existing works have primarily developed myopic/greedy algorithms or performed exploration and exploitation separately. To perform active learning in real time, we then propose an anytime algorithm based on ε-BAL with performance guarantee and empirically demonstrate using synthetic and real-world datasets that, with limited budget, it outperforms the state-of-the-art algorithms. ER -
APA
Hoang, T.N., Low, B.K.H., Jaillet, P. & Kankanhalli, M.. (2014). Nonmyopic ε-Bayes-Optimal Active Learning of Gaussian Processes. Proceedings of the 31st International Conference on Machine Learning, in Proceedings of Machine Learning Research 32(2):739-747 Available from https://proceedings.mlr.press/v32/hoang14.html.

Related Material