Nonmyopic ϵ-Bayes-Optimal Active Learning of Gaussian Processes
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 develo...
| Main Authors: | , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | en_US |
| Published: |
Association for Computing Machinery (ACM)
2015
|
| Online Access: | http://hdl.handle.net/1721.1/100444 https://orcid.org/0000-0002-8585-6566 |
| _version_ | 1826201981164716032 |
|---|---|
| author | Hoang, Trong Nghia Low, Bryan Kian Hsiang Jaillet, Patrick Kankanhalli, Mohan |
| author2 | Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science |
| author_facet | Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science Hoang, Trong Nghia Low, Bryan Kian Hsiang Jaillet, Patrick Kankanhalli, Mohan |
| author_sort | Hoang, Trong Nghia |
| collection | MIT |
| description | 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. |
| first_indexed | 2024-09-23T12:00:02Z |
| format | Article |
| id | mit-1721.1/100444 |
| institution | Massachusetts Institute of Technology |
| language | en_US |
| last_indexed | 2024-09-23T12:00:02Z |
| publishDate | 2015 |
| publisher | Association for Computing Machinery (ACM) |
| record_format | dspace |
| spelling | mit-1721.1/1004442022-09-27T23:25:51Z Nonmyopic ϵ-Bayes-Optimal Active Learning of Gaussian Processes Hoang, Trong Nghia Low, Bryan Kian Hsiang Jaillet, Patrick Kankanhalli, Mohan Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science Jaillet, Patrick 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. Singapore. National Research Foundation (Singapore-MIT Alliance for Research and Technology Center) 2015-12-19T02:35:30Z 2015-12-19T02:35:30Z 2014 Article http://purl.org/eprint/type/ConferencePaper 1938-7228 http://hdl.handle.net/1721.1/100444 Hoang, Trong Nghia, Bryan Kian Hsiang Low, Patrick Jaillet, and Mohan Kankanhalli. "Nonmyopic ϵ-Bayes-Optimal Active Learning of Gaussian Processes." 31st International Conference on Machine Learning (2014). https://orcid.org/0000-0002-8585-6566 en_US http://jmlr.org/proceedings/papers/v32/ Proceedings of the 31st International Conference on Machine Learning Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use. application/pdf Association for Computing Machinery (ACM) MIT web domain |
| spellingShingle | Hoang, Trong Nghia Low, Bryan Kian Hsiang Jaillet, Patrick Kankanhalli, Mohan Nonmyopic ϵ-Bayes-Optimal Active Learning of Gaussian Processes |
| title | Nonmyopic ϵ-Bayes-Optimal Active Learning of Gaussian Processes |
| title_full | Nonmyopic ϵ-Bayes-Optimal Active Learning of Gaussian Processes |
| title_fullStr | Nonmyopic ϵ-Bayes-Optimal Active Learning of Gaussian Processes |
| title_full_unstemmed | Nonmyopic ϵ-Bayes-Optimal Active Learning of Gaussian Processes |
| title_short | Nonmyopic ϵ-Bayes-Optimal Active Learning of Gaussian Processes |
| title_sort | nonmyopic ϵ bayes optimal active learning of gaussian processes |
| url | http://hdl.handle.net/1721.1/100444 https://orcid.org/0000-0002-8585-6566 |
| work_keys_str_mv | AT hoangtrongnghia nonmyopicebayesoptimalactivelearningofgaussianprocesses AT lowbryankianhsiang nonmyopicebayesoptimalactivelearningofgaussianprocesses AT jailletpatrick nonmyopicebayesoptimalactivelearningofgaussianprocesses AT kankanhallimohan nonmyopicebayesoptimalactivelearningofgaussianprocesses |