Randomised Gaussian Process Upper Confidence Bound for Bayesian Optimisation
This work addresses optimization efficiency for machine learning and engineering applications, but it is incremental as it builds on the established GP-UCB method.
The authors tackled the problem of improving Bayesian optimization performance by developing a modified GP-UCB acquisition function that samples the exploration-exploitation trade-off parameter from a distribution, proving it allows better adaptation without compromising regret bounds and showing it outperforms GP-UCB in real-world and synthetic problems.
In order to improve the performance of Bayesian optimisation, we develop a modified Gaussian process upper confidence bound (GP-UCB) acquisition function. This is done by sampling the exploration-exploitation trade-off parameter from a distribution. We prove that this allows the expected trade-off parameter to be altered to better suit the problem without compromising a bound on the function's Bayesian regret. We also provide results showing that our method achieves better performance than GP-UCB in a range of real-world and synthetic problems.