Implicit Online Learning

Implicit Online Learning,Brian Kulis,Peter L. Bartlett

Implicit Online Learning   (Citations: 3)
BibTex | RIS | RefWorks Download
Online learning algorithms have recently risen to prominence due to their strong theoretical guar- antees and an increasing number of practical ap- plications for large-scale data analysis problems. In this paper, we analyze a class of online learn- ing algorithms based on fixed potentials and non- linearized losses, which yields algorithms with implicit update rules. We show how to effi- ciently compute these updates, and we prove re- gret bounds for the algorithms. We apply our for- mulation to several special cases where our ap- proach has benefits over existing online learning methods. In particular, we provide improved al- gorithms and bounds for the online metric learn- ing problem, and show improved robustness for online linear prediction problems. Results over a variety of data sets demonstrate the advantages of our framework.
Cumulative Annual
View Publication
The following links allow you to view full publications. These links are maintained by other sources not affiliated with Microsoft Academic Search.
Sort by: