Coin Betting and Parameter-Free Online Learning

Abstract

In the recent years, a number of parameter-free algorithms have been developed for online linear optimization over Hilbert spaces and for learning with expert advice. These algorithms achieve optimal regret bounds that depend on the unknown competitors, without having to tune the learning rates with oracle choices. We present a new intuitive framework to design parameter-free algorithms for \emph{both} online linear optimization over Hilbert spaces and for learning with expert advice, based on reductions to betting on outcomes of adversarial coins. We instantiate it using a betting algorithm based on the Krichevsky-Trofimov estimator. The resulting algorithms are simple, with no parameters to be tuned, and they improve or match previous results in terms of regret guarantee and per-round complexity.

Fixed an compilation error in the latex

Cited by in corpus (8)

• The Many Faces of Exponential Weights in Online Learning
• MetaGrad: Multiple Learning Rates in Online Learning
• Online Learning Without Prior Information
• Training Deep Networks without Learning Rates Through Coin Betting
• Improved Strongly Adaptive Online Learning using Coin Betting
• Online Parameter-Free Learning of Multiple Low Variance Tasks
• Approximately Exact Line Search
• Parameter-free Gradient Temporal Difference Learning