Active Learning Is Planning: Nonmyopic ε-Bayes-Optimal Active Learning of Gaussian Processes
Author(s)Hoang, Trong Nghia; Low, Kian Hsiang; Jaillet, Patrick; Kankanhalli, Mohan
MetadataShow full item record
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 optimizes the trade-off. In contrast, existing works have primarily developed 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 a real-world dataset that, with limited budget, it outperforms the state-of-the-art algorithms.
DepartmentMassachusetts Institute of Technology. Department of Electrical Engineering and Computer Science
Machine Learning and Knowledge Discovery in Databases
Hoang, Trong Nghia, Kian Hsiang Low, Patrick Jaillet, and Mohan Kankanhalli. “Active Learning Is Planning: Nonmyopic ε-Bayes-Optimal Active Learning of Gaussian Processes.” Lecture Notes in Computer Science (2014): 494–498.
Author's final manuscript