On the Value of Prior in Online Learning to Rank

Branislav Kveton, Ofer Meshi, Masrour Zoghi, Zhen Qin
Proceedings of The 25th International Conference on Artificial Intelligence and Statistics, PMLR 151:6880-6892, 2022.

Abstract

This paper addresses the cold-start problem in online learning to rank (OLTR). We show both theoretically and empirically that priors improve the quality of ranked lists presented to users interactively based on user feedback. These priors can come in the form of unbiased estimates of the relevance of the ranked items, or more practically, can be obtained from offline-learned models. Our experiments show the effectiveness of priors in improving the short-term regret of tabular OLTR algorithms, based on Thompson sampling and BayesUCB.

Cite this Paper

BibTeX
@InProceedings{pmlr-v151-kveton22a, title = { On the Value of Prior in Online Learning to Rank }, author = {Kveton, Branislav and Meshi, Ofer and Zoghi, Masrour and Qin, Zhen}, booktitle = {Proceedings of The 25th International Conference on Artificial Intelligence and Statistics}, pages = {6880--6892}, year = {2022}, editor = {Camps-Valls, Gustau and Ruiz, Francisco J. R. and Valera, Isabel}, volume = {151}, series = {Proceedings of Machine Learning Research}, month = {28--30 Mar}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v151/kveton22a/kveton22a.pdf}, url = {https://proceedings.mlr.press/v151/kveton22a.html}, abstract = { This paper addresses the cold-start problem in online learning to rank (OLTR). We show both theoretically and empirically that priors improve the quality of ranked lists presented to users interactively based on user feedback. These priors can come in the form of unbiased estimates of the relevance of the ranked items, or more practically, can be obtained from offline-learned models. Our experiments show the effectiveness of priors in improving the short-term regret of tabular OLTR algorithms, based on Thompson sampling and BayesUCB. } }
Endnote
%0 Conference Paper %T On the Value of Prior in Online Learning to Rank %A Branislav Kveton %A Ofer Meshi %A Masrour Zoghi %A Zhen Qin %B Proceedings of The 25th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2022 %E Gustau Camps-Valls %E Francisco J. R. Ruiz %E Isabel Valera %F pmlr-v151-kveton22a %I PMLR %P 6880--6892 %U https://proceedings.mlr.press/v151/kveton22a.html %V 151 %X This paper addresses the cold-start problem in online learning to rank (OLTR). We show both theoretically and empirically that priors improve the quality of ranked lists presented to users interactively based on user feedback. These priors can come in the form of unbiased estimates of the relevance of the ranked items, or more practically, can be obtained from offline-learned models. Our experiments show the effectiveness of priors in improving the short-term regret of tabular OLTR algorithms, based on Thompson sampling and BayesUCB.
APA
Kveton, B., Meshi, O., Zoghi, M. & Qin, Z.. (2022). On the Value of Prior in Online Learning to Rank . Proceedings of The 25th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 151:6880-6892 Available from https://proceedings.mlr.press/v151/kveton22a.html.

Related Material