GLASSES: Relieving The Myopia Of Bayesian Optimisation

Abstract

We present GLASSES: Global optimisation with Look-Ahead through Stochastic Simulation and Expected-loss Search. The majority of global optimisation approaches in use are myopic, in only considering the impact of the next function value; the non-myopic approaches that do exist are able to consider only a handful of future evaluations. Our novel algorithm, GLASSES, permits the consideration of dozens of evaluations into the future. This is done by approximating the ideal look-ahead loss function, which is expensive to evaluate, by a cheaper alternative in which the future steps of the algorithm are simulated beforehand. An Expectation Propagation algorithm is used to compute the expected value of the loss.We show that the far-horizon planning thus enabled leads to substantive performance gains in empirical tests.

12 pages, 9 figures

References in corpus (4)

• Practical Bayesian Optimization of Machine Learning Algorithms
• A Tutorial on Bayesian Optimization of Expensive Cost Functions, with Application to Active User Modeling and Hierarchical Reinforcement Learning
• Expectation Propagation for approximate Bayesian inference
• Batch Bayesian Optimization via Local Penalization

Cited by in corpus (19)

• A Tutorial on Bayesian Optimization
• Replication or exploration? Sequential design for stochastic simulation experiments
• Efficient acquisition rules for model-based approximate Bayesian computation
• Batched Gaussian Process Bandit Optimization via Determinantal Point Processes
• A Stratified Analysis of Bayesian Optimization Methods
• Strong convergence rates of probabilistic integrators for ordinary differential equations
• Why Non-myopic Bayesian Optimization is Promising and How Far Should We Look-ahead? A Study via Rollout
• Deep Adaptive Design: Amortizing Sequential Bayesian Experimental Design
• GIBBON: General-purpose Information-Based Bayesian OptimisatioN
• Bayesian Optimization for Cascade-type Multi-stage Processes
• Multi-Step Budgeted Bayesian Optimization with Unknown Evaluation Costs
• Efficient Nonmyopic Bayesian Optimization via One-Shot Multi-Step Trees
• Distributionally Ambiguous Optimization Techniques for Batch Bayesian Optimization
• Correcting boundary over-exploration deficiencies in Bayesian optimization with virtual derivative sign observations
• BINOCULARS for Efficient, Nonmyopic Sequential Experimental Design
• Reinforced Few-Shot Acquisition Function Learning for Bayesian Optimization
• Recursive Two-Step Lookahead Expected Payoff for Time-Dependent Bayesian Optimization
• Efficient Rollout Strategies for Bayesian Optimization
• Lookahead Acquisition Functions for Finite-Horizon Time-Dependent Bayesian Optimization and Application to Quantum Optimal Control