Online Program Home
  My Program

Abstract Details

Activity Number: 478 - Online Machine Learning for Prediction and Sequential Decision Making
Type: Invited
Date/Time: Wednesday, August 2, 2017 : 10:30 AM to 12:20 PM
Sponsor: IMS
Abstract #321949
Title: Improved Strongly Adaptive Online Learning Using Coin Betting
Author(s): Rebecca Willett* and Kwang-Sung Jun and Francesco Orabona and Stephen Wright
Companies: University of Wisconsin-Madison and University of Wisconsin-Madison and Stony Brook University and University of Wisconsin-Madison
Keywords: online learning ; learning with expert advice ; regret ; parameter free ; bandits
Abstract:

In this talk, I will describe a new parameter-free online learning algorithm for changing environments. Our algorithm, Coin Betting for Changing Environments (CBCE), leverages elements of Sleeping Bandits methods and coin betting algorithms for parameter-free online learning. CBCE admits a strongly adaptive regret bound that is a factor of at least \sqrt{log(T)} better than competing algorithms with the same time complexity as ours, where T is the time horizon. In addition, CBCE is "parameter free" -- that is, it is unnecessary to know a priori how often the environment changes to set tuning parameters. Empirical results show that our algorithm outperforms state-of-the-art methods in learning with expert advice and metric learning scenarios.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2017 program

 
 
Copyright © American Statistical Association