A Quadrature Approach for General-Purpose Batch Bayesian Optimization via Probabilistic Lifting
This work addresses efficiency and flexibility issues in Bayesian optimization for researchers and practitioners, though it appears incremental as it builds on existing methods with modular improvements.
The paper tackles challenges in parallel Bayesian optimization, such as flexibility with acquisition functions and kernel choices, handling discrete and continuous variables, model misspecification, and fast parallelization, by introducing SOBER, a framework using probabilistic lifting with kernel quadrature, which achieves versatility, gradient-free sampling, and adaptive batch sizes as a Python library.
Parallelisation in Bayesian optimisation is a common strategy but faces several challenges: the need for flexibility in acquisition functions and kernel choices, flexibility dealing with discrete and continuous variables simultaneously, model misspecification, and lastly fast massive parallelisation. To address these challenges, we introduce a versatile and modular framework for batch Bayesian optimisation via probabilistic lifting with kernel quadrature, called SOBER, which we present as a Python library based on GPyTorch/BoTorch. Our framework offers the following unique benefits: (1) Versatility in downstream tasks under a unified approach. (2) A gradient-free sampler, which does not require the gradient of acquisition functions, offering domain-agnostic sampling (e.g., discrete and mixed variables, non-Euclidean space). (3) Flexibility in domain prior distribution. (4) Adaptive batch size (autonomous determination of the optimal batch size). (5) Robustness against a misspecified reproducing kernel Hilbert space. (6) Natural stopping criterion.