Building Bridges: Viewing Active Learning from the Multi-Armed Bandit Lens

Ravi Ganti, Alexander Gray
Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence, PMLR R11:392-401, 2013.

Abstract

In this paper we propose a multi-armed ban- dit inspired, pool based active learning algo- rithm for the problem of binary classification. By carefully constructing an analogy between active learning and multi-armed bandits, we utilize ideas such as lower confidence bounds, and self-concordant regularization from the multi-armed bandit literature to design our proposed algorithm. Our algorithm is a se- quential algorithm, which in each round as- signs a sampling distribution on the pool, samples one point from this distribution, and queries the oracle for the label of this sam- pled point. The design of this sampling dis- tribution is also inspired by the analogy be- tween active learning and multi-armed ban- dits. We show how to derive lower confidence bounds required by our algorithm. Exper- imental comparisons to previously proposed active learning algorithms show superior per- formance on some standard UCI data-sets.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR11-ganti13a, title = {Building Bridges: Viewing Active Learning from the Multi-Armed Bandit Lens}, author = {Ganti, Ravi and Gray, Alexander}, booktitle = {Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence}, pages = {392--401}, year = {2013}, editor = {Nicholson, Ann and Smyth, Padhraic}, volume = {R11}, series = {Proceedings of Machine Learning Research}, month = {12--14 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/r11/main/assets/ganti13a/ganti13a.pdf}, url = {https://proceedings.mlr.press/r11/ganti13a.html}, abstract = {In this paper we propose a multi-armed ban- dit inspired, pool based active learning algo- rithm for the problem of binary classification. By carefully constructing an analogy between active learning and multi-armed bandits, we utilize ideas such as lower confidence bounds, and self-concordant regularization from the multi-armed bandit literature to design our proposed algorithm. Our algorithm is a se- quential algorithm, which in each round as- signs a sampling distribution on the pool, samples one point from this distribution, and queries the oracle for the label of this sam- pled point. The design of this sampling dis- tribution is also inspired by the analogy be- tween active learning and multi-armed ban- dits. We show how to derive lower confidence bounds required by our algorithm. Exper- imental comparisons to previously proposed active learning algorithms show superior per- formance on some standard UCI data-sets.}, note = {Reissued by PMLR on 04 October 2026.} }
Endnote
%0 Conference Paper %T Building Bridges: Viewing Active Learning from the Multi-Armed Bandit Lens %A Ravi Ganti %A Alexander Gray %B Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2013 %E Ann Nicholson %E Padhraic Smyth %F pmlr-vR11-ganti13a %I PMLR %P 392--401 %U https://proceedings.mlr.press/r11/ganti13a.html %V R11 %X In this paper we propose a multi-armed ban- dit inspired, pool based active learning algo- rithm for the problem of binary classification. By carefully constructing an analogy between active learning and multi-armed bandits, we utilize ideas such as lower confidence bounds, and self-concordant regularization from the multi-armed bandit literature to design our proposed algorithm. Our algorithm is a se- quential algorithm, which in each round as- signs a sampling distribution on the pool, samples one point from this distribution, and queries the oracle for the label of this sam- pled point. The design of this sampling dis- tribution is also inspired by the analogy be- tween active learning and multi-armed ban- dits. We show how to derive lower confidence bounds required by our algorithm. Exper- imental comparisons to previously proposed active learning algorithms show superior per- formance on some standard UCI data-sets. %Z Reissued by PMLR on 04 October 2026.
APA
Ganti, R. & Gray, A.. (2013). Building Bridges: Viewing Active Learning from the Multi-Armed Bandit Lens. Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R11:392-401 Available from https://proceedings.mlr.press/r11/ganti13a.html. Reissued by PMLR on 04 October 2026.

Related Material