Regret Bounds for Deterministic Gaussian Process Bandits

Details Regret Bounds for Deterministic Gaussian Process Bandits Regret Bounds for Deterministic Gaussian Process Bandits

2012-03-09

arxiv.org/abs/1203.2177v1

Computer Science

Learning

17 pages, 5 figures

Scientific paper

This paper analyses the problem of Gaussian process (GP) bandits with deterministic observations. The analysis uses a branch and bound algorithm that is related to the UCB algorithm of (Srinivas et al., 2010). For GPs with Gaussian observation noise, with variance strictly greater than zero, (Srinivas et al., 2010) proved that the regret vanishes at the approximate rate of $O(\frac{1}{\sqrt{t}})$, where t is the number of observations. To complement their result, we attack the deterministic case and attain a much faster exponential convergence rate. Under some regularity assumptions, we show that the regret decreases asymptotically according to $O(e^{-\frac{\tau t}{(\ln t)^{d/4}}})$ with high probability. Here, d is the dimension of the search space and $\tau$ is a constant that depends on the behaviour of the objective function near its global maximum.

Affiliated with

Also associated with

No associations

Rating

Regret Bounds for Deterministic Gaussian Process Bandits does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.

If you have personal experience with Regret Bounds for Deterministic Gaussian Process Bandits, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Regret Bounds for Deterministic Gaussian Process Bandits will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFWR-SCP-O-487949