Normalized Online Learning

John Langford, Paul Mineiro, Stephane Ross
Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence, PMLR R11:491-499, 2013.

Abstract

We introduce online learning algorithms which are independent of feature scales, proving regret bounds dependent on the ratio of scales existent in the data rather than the absolute scale. This has several useful effects: there is no need to pre- normalize data, the test-time and test-space com- plexity are reduced, and the algorithms are more robust.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR11-langford13a, title = {Normalized Online Learning}, author = {Langford, John and Mineiro, Paul and Ross, Stephane}, booktitle = {Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence}, pages = {491--499}, 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/langford13a/langford13a.pdf}, url = {https://proceedings.mlr.press/r11/langford13a.html}, abstract = {We introduce online learning algorithms which are independent of feature scales, proving regret bounds dependent on the ratio of scales existent in the data rather than the absolute scale. This has several useful effects: there is no need to pre- normalize data, the test-time and test-space com- plexity are reduced, and the algorithms are more robust.}, note = {Reissued by PMLR on 04 October 2026.} }
Endnote
%0 Conference Paper %T Normalized Online Learning %A John Langford %A Paul Mineiro %A Stephane Ross %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-langford13a %I PMLR %P 491--499 %U https://proceedings.mlr.press/r11/langford13a.html %V R11 %X We introduce online learning algorithms which are independent of feature scales, proving regret bounds dependent on the ratio of scales existent in the data rather than the absolute scale. This has several useful effects: there is no need to pre- normalize data, the test-time and test-space com- plexity are reduced, and the algorithms are more robust. %Z Reissued by PMLR on 04 October 2026.
APA
Langford, J., Mineiro, P. & Ross, S.. (2013). Normalized Online Learning. Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R11:491-499 Available from https://proceedings.mlr.press/r11/langford13a.html. Reissued by PMLR on 04 October 2026.

Related Material