Batch Bayesian Optimization via Local Penalization

Abstract

The popularity of Bayesian optimization methods for efficient exploration of parameter spaces has lead to a series of papers applying Gaussian processes as surrogates in the optimization of functions. However, most proposed approaches only allow the exploration of the parameter space to occur sequentially. Often, it is desirable to simultaneously propose batches of parameter values to explore. This is particularly the case when large parallel processing facilities are available. These facilities could be computational or physical facets of the process being optimized. E.g. in biological experiments many experimental set ups allow several samples to be simultaneously processed. Batch methods, however, require modeling of the interaction between the evaluations in the batch, which can be expensive in complex scenarios. We investigate a simple heuristic based on an estimate of the Lipschitz constant that captures the most important aspect of this interaction (i.e. local repulsion) at negligible computational overhead. The resulting algorithm compares well, in running time, with much more elaborate alternatives. The approach assumes that the function of interest, $f$, is a Lipschitz continuous function. A wrap-loop around the acquisition function is used to collect batches of points of certain size minimizing the non-parallelizable computational effort. The speed-up of our method with respect to previous approaches is significant in a set of computationally expensive experiments.

11 pages, 10 figures

Cited by in corpus (72)

• BatchBALD: Efficient and Diverse Batch Acquisition for Deep Bayesian Active Learning
• Gryffin: An algorithm for Bayesian optimization of categorical variables informed by expert knowledge
• Bayesian Optimization is Superior to Random Search for Machine Learning Hyperparameter Tuning: Analysis of the Black-Box Optimization Challenge 2020
• An Efficient Batch Constrained Bayesian Optimization Approach for Analog Circuit Synthesis via Multi-objective Acquisition Ensemble
• Batched Large-scale Bayesian Optimization in High-dimensional Spaces
• Parallel and Distributed Thompson Sampling for Large-scale Accelerated Exploration of Chemical Space
• Bayesian Batch Active Learning as Sparse Subset Approximation
• A System for Massively Parallel Hyperparameter Tuning
• GPflowOpt: A Bayesian Optimization Library using TensorFlow
• GLASSES: Relieving The Myopia Of Bayesian Optimisation
• Tuning Hyperparameters without Grad Students: Scalable and Robust Bayesian Optimisation with Dragonfly
• An Efficient Asynchronous Batch Bayesian Optimization Approach for Analog Circuit Synthesis
• Preferential Bayesian Optimization
• Batched Gaussian Process Bandit Optimization via Determinantal Point Processes
• Robust Maximization of Non-Submodular Objectives
• GPdoemd: a Python package for design of experiments for model discrimination
• Gaussian Process Molecule Property Prediction with FlowMO
• Multi-fidelity Bayesian Optimization with Max-value Entropy Search and its parallelization
• Is novelty predictable?
• BOSS: Bayesian Optimization over String Spaces
• Neural Architecture Generator Optimization
• Uniform Error Bounds for Gaussian Process Regression with Application to Safe Control
• Provably Efficient Online Hyperparameter Optimization with Population-Based Bandits
• Bayesian Optimisation over Multiple Continuous and Categorical Inputs
• Inverse Reinforcement Learning from Summary Data
• Optimal Transport Kernels for Sequential and Parallel Neural Architecture Search
• Co-Learning Bayesian Optimization
• Parallel Predictive Entropy Search for Multi-objective Bayesian Optimization with Constraints
• Rapidly adapting robot swarms with Swarm Map-based Bayesian Optimisation
• DHA: End-to-End Joint Optimization of Data Augmentation Policy, Hyper-parameter and Architecture
• Where to Drill Next? A Dual-Weighted Approach to Adaptive Optimal Design of Groundwater Surveys
• GIBBON: General-purpose Information-Based Bayesian OptimisatioN
• Using Bayesian Optimization to Accelerate Virtual Screening for the Discovery of Therapeutics Appropriate for Repurposing for COVID-19
• Budgeted Batch Bayesian Optimization With Unknown Batch Sizes
• Bayesian Optimization for Cascade-type Multi-stage Processes
• Optimal approximation for unconstrained non-submodular minimization
• Sampling Acquisition Functions for Batch Bayesian Optimization
• An Asymptotically Optimal Multi-Armed Bandit Algorithm and Hyperparameter Optimization
• Scalable Thompson Sampling using Sparse Gaussian Process Models
• Discretization-free Knowledge Gradient Methods for Bayesian Optimization
• Nonmyopic Gaussian Process Optimization with Macro-Actions
• BOSH: Bayesian Optimization by Sampling Hierarchically
• Faster Improvement Rate Population Based Training
• Derivative free optimization via repeated classification
• Distributionally Ambiguous Optimization Techniques for Batch Bayesian Optimization
• A portfolio approach to massively parallel Bayesian optimization
• Practical Batch Bayesian Optimization for Less Expensive Functions
• BINOCULARS for Efficient, Nonmyopic Sequential Experimental Design
• On Batch Bayesian Optimization
• Open Loop Hyperparameter Optimization and Determinantal Point Processes
• Efficient Batch Black-box Optimization with Deterministic Regret Bounds
• Bayesian Optimization for Categorical and Category-Specific Continuous Inputs
• Efficient Transfer Bayesian Optimization with Auxiliary Information
• Batch Multi-Fidelity Bayesian Optimization with Deep Auto-Regressive Networks
• Attentive Neural Processes and Batch Bayesian Optimization for Scalable Calibration of Physics-Informed Digital Twins
• Preferential Batch Bayesian Optimization
• Trusted-Maximizers Entropy Search for Efficient Bayesian Optimization
• Batch Selection for Parallelisation of Bayesian Quadrature
• A Hierarchical Approach to Scaling Batch Active Search Over Structured Data
• Efficient Phase Diagram Sampling by Active Learning
• Bayesian Optimization using Pseudo-Points
• Practical Bayesian Optimization of Objectives with Conditioning Variables
• DEEP-BO for Hyperparameter Optimization of Deep Networks
• Surrogate Assisted Evolutionary Algorithm for Medium Scale Expensive Multi-Objective Optimisation Problems
• Parallel Contextual Bandits in Wireless Handover Optimization
• Bayesian optimization for modular black-box systems with switching costs
• Robust Policy Search for Robot Navigation
• LinEasyBO: Scalable Bayesian Optimization Approach for Analog Circuit Synthesis via One-Dimensional Subspaces
• A tree-based radial basis function method for noisy parallel surrogate optimization
• Collaborative and Federated Black-box Optimization: A Bayesian Optimization Perspective
• Ply-drop design of non-conventional composites using Bayesian optimization
• Designing over uncertain outcomes with stochastic sampling Bayesian optimization