A Tutorial on Bayesian Optimization

Abstract

Bayesian optimization is an approach to optimizing objective functions that take a long time (minutes or hours) to evaluate. It is best-suited for optimization over continuous domains of less than 20 dimensions, and tolerates stochastic noise in function evaluations. It builds a surrogate for the objective and quantifies the uncertainty in that surrogate using a Bayesian machine learning technique, Gaussian process regression, and then uses an acquisition function defined from this surrogate to decide where to sample. In this tutorial, we describe how Bayesian optimization works, including Gaussian process regression and three common acquisition functions: expected improvement, entropy search, and knowledge gradient. We then discuss more advanced techniques, including running multiple function evaluations in parallel, multi-fidelity and multi-information source optimization, expensive-to-evaluate constraints, random environmental conditions, multi-task Bayesian optimization, and the inclusion of derivative information. We conclude with a discussion of Bayesian optimization software and future research directions in the field. Within our tutorial material we provide a generalization of expected improvement to noisy evaluations, beyond the noise-free setting where it is more commonly applied. This generalization is justified by a formal decision-theoretic argument, standing in contrast to previous ad hoc modifications.

References in corpus (14)

• 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
• Predictive Entropy Search for Efficient Global Optimization of Black-box Functions
• Bayesian optimization for materials design
• Fast Bayesian Optimization of Machine Learning Hyperparameters on Large Datasets
• High Dimensional Bayesian Optimisation and Bandits via Additive Models
• Bayesian Optimization with Gradients
• The Parallel Knowledge Gradient Method for Batch Bayesian Optimization
• Freeze-Thaw Bayesian Optimization
• Predictive Entropy Search for Bayesian Optimization with Unknown Constraints
• Parallel Bayesian Global Optimization of Expensive Functions
• GLASSES: Relieving The Myopia Of Bayesian Optimisation
• Practical Bayesian Optimization for Variable Cost Objectives
• Multi-Step Bayesian Optimization for One-Dimensional Feasibility Determination

Cited by in corpus (153)

• A Review of Uncertainty Quantification in Deep Learning: Techniques, Applications and Challenges
• Hyper-Parameter Optimization: A Review of Algorithms and Applications
• Accelerating high-throughput virtual screening through molecular pool-based active learning
• Training of Quantum Circuits on a Hybrid Quantum Computer
• Bayesian Optimization is Superior to Random Search for Machine Learning Hyperparameter Tuning: Analysis of the Black-Box Optimization Challenge 2020
• Testing Scenario Library Generation for Connected and Automated Vehicles: An Adaptive Framework
• Reward Design for Driver Repositioning Using Multi-Agent Reinforcement Learning
• Fast differentiable DNA and protein sequence optimization for molecular design
• Bayesian Optimization with Output-Weighted Optimal Sampling
• AutoTune: Automatically Tuning Convolutional Neural Networks for Improved Transfer Learning
• Upper Trust Bound Feasibility Criterion for Mixed Constrained Bayesian Optimization with Application to Aircraft Design
• ENTMOOT: A Framework for Optimization over Ensemble Tree Models
• Real-world Video Adaptation with Reinforcement Learning
• High-Dimensional Bayesian Optimization with Sparse Axis-Aligned Subspaces
• Parallel Bayesian Optimization of Multiple Noisy Objectives with Expected Hypervolume Improvement
• AdaLead: A simple and robust adaptive greedy search algorithm for sequence design
• Revisiting Bayesian Optimization in the light of the COCO benchmark
• Learning Search Space Partition for Black-box Optimization using Monte Carlo Tree Search
• Bilevel methods for image reconstruction
• Cost-aware Bayesian Optimization
• Better Security Bug Report Classification via Hyperparameter Optimization
• Multi-Objective Constrained Optimization for Energy Applications via Tree Ensembles
• Finding Critical Scenarios for Automated Driving Systems: A Systematic Literature Review
• MPC Controller Tuning using Bayesian Optimization Techniques
• Black-box Adversarial Attacks with Bayesian Optimization
• The Cost of Improving the Precision of the Variational Quantum Eigensolver for Quantum Chemistry
• Automated Configuration of Genetic Algorithms by Tuning for Anytime Performance
• Investigation of REFINED CNN ensemble learning for anti-cancer drug sensitivity prediction
• Machine Learning based optimization for interval uncertainty propagation
• Structure Prediction of Epitaxial Inorganic Interfaces by Lattice and Surface Matching with Ogre
• Saddle Point Optimization with Approximate Minimization Oracle and its Application to Robust Berthing Control
• Scalable Constrained Bayesian Optimization
• Bayesian Optimisation for Sequential Experimental Design with Applications in Additive Manufacturing
• Bayesian Hyperparameter Optimization with BoTorch, GPyTorch and Ax
• Accelerating Bayesian microseismic event location with deep learning
• Bayesian Optimization using Deep Gaussian Processes
• Are we Forgetting about Compositional Optimisers in Bayesian Optimisation?
• Think Global and Act Local: Bayesian Optimisation over High-Dimensional Categorical and Mixed Search Spaces
• A Tutorial on Distance Metric Learning: Mathematical Foundations, Algorithms, Experimental Analysis, Prospects and Challenges (with Appendices on Mathematical Background and Detailed Algorithms Explanation)
• On Local Optimizers of Acquisition Functions in Bayesian Optimization
• Multi-Objective Bayesian Optimization over High-Dimensional Search Spaces
• Planar Robot Casting with Real2Sim2Real Self-Supervised Learning
• Reliable training and estimation of variance networks
• Bayesian Optimisation over Multiple Continuous and Categorical Inputs
• Verification of deep probabilistic models
• High-Dimensional Bayesian Optimisation with Variational Autoencoders and Deep Metric Learning
• Knowing The What But Not The Where in Bayesian Optimization
• Benchmarking the Performance of Bayesian Optimization across Multiple Experimental Materials Science Domains
• Adaptive Structural Hyper-Parameter Configuration by Q-Learning
• Optimal Transport Kernels for Sequential and Parallel Neural Architecture Search
• Global Stochastic Optimization of Stellarator Coil Configurations
• Additive Tree-Structured Conditional Parameter Spaces in Bayesian Optimization: A Novel Covariance Function and a Fast Implementation
• Scalable Combinatorial Bayesian Optimization with Tractable Statistical models
• Applying Bayesian Optimization with Gaussian Process Regression to Computational Fluid Dynamics Problems
• Attacking Vision-based Perception in End-to-End Autonomous Driving Models
• On the Numerical Performance of Derivative-Free Optimization Methods Based on Finite-Difference Approximations
• Hyper-parameter estimation method with particle swarm optimization
• Kernels over Sets of Finite Sets using RKHS Embeddings, with Application to Bayesian (Combinatorial) Optimization
• Towards Assessing the Impact of Bayesian Optimization's Own Hyperparameters
• Combinatorial 3D Shape Generation via Sequential Assembly
• Run2Survive: A Decision-theoretic Approach to Algorithm Selection based on Survival Analysis
• Bayesian Optimization of 2D Echocardiography Segmentation
• Fair Sequential Selection Using Supervised Learning Models
• Reproducible Performance Optimization of Complex Applications on the Edge-to-Cloud Continuum
• High-Dimensional Experimental Design and Kernel Bandits
• Semi-supervised Embedding Learning for High-dimensional Bayesian Optimization
• Latency-Aware Neural Architecture Search with Multi-Objective Bayesian Optimization
• Neuroadaptive electroencephalography: a proof-of-principle study in infants
• Achieving Real-Time LiDAR 3D Object Detection on a Mobile Device
• NOMU: Neural Optimization-based Model Uncertainty
• Interpretable Disease Prediction based on Reinforcement Path Reasoning over Knowledge Graphs
• Neural Process for Black-Box Model Optimization Under Bayesian Framework
• Multi-Step Budgeted Bayesian Optimization with Unknown Evaluation Costs
• A Principled Approach to Design Using High Fidelity Fluid-Structure Interaction Simulations
• Cautious Bayesian Optimization for Efficient and Scalable Policy Search
• Addressing Variance Shrinkage in Variational Autoencoders using Quantile Regression
• Combining Latent Space and Structured Kernels for Bayesian Optimization over Combinatorial Spaces
• Quantity vs. Quality: On Hyperparameter Optimization for Deep Reinforcement Learning
• Kernel Interpolation for Scalable Online Gaussian Processes
• A Learning-Based Tune-Free Control Framework for Large Scale Autonomous Driving System Deployment
• Computationally Efficient High-Dimensional Bayesian Optimization via Variable Selection
• Bayesian Optimization over Hybrid Spaces
• Bayesian Algorithm Execution: Estimating Computable Properties of Black-box Functions Using Mutual Information
• Trainless Model Performance Estimation for Neural Architecture Search
• Pathwise Conditioning of Gaussian Processes
• Practical Transfer Learning for Bayesian Optimization
• You Only Query Once: Effective Black Box Adversarial Attacks with Minimal Repeated Queries
• Efficient Nonmyopic Bayesian Optimization via One-Shot Multi-Step Trees
• A gray-box approach for curriculum learning
• Kriging prediction with isotropic Matérn correlations: Robustness and experimental design
• BISTRO: Berkeley Integrated System for Transportation Optimization
• On the Correspondence between Gaussian Processes and Geometric Harmonics
• Sequential Model Optimization for Software Process Control
• An In-depth Summary of Recent Artificial Intelligence Applications in Drug Design
• Bayesian Optimisation for Constrained Problems
• Multi-Task Gaussian Processes and Dilated Convolutional Networks for Reconstruction of Reproductive Hormonal Dynamics
• Bayesian Optimization for Iterative Learning
• Additive Tree-Structured Covariance Function for Conditional Parameter Spaces in Bayesian Optimization
• Flexible mean field variational inference using mixtures of non-overlapping exponential families
• How Low Can We Go: Trading Memory for Error in Low-Precision Training
• Bayesian Screening: Multi-test Bayesian Optimization Applied to in silico Material Screening
• Adaptive Compression-based Lifelong Learning
• Data Troubles in Sentence Level Confidence Estimation for Machine Translation
• Orchestrate: Infrastructure for Enabling Parallelism during Hyperparameter Optimization
• Programming and Deployment of Autonomous Swarms using Multi-Agent Reinforcement Learning
• Uncertainty Quantification for Bayesian Optimization
• Emergent Hand Morphology and Control from Optimizing Robust Grasps of Diverse Objects
• PointBA: Towards Backdoor Attacks in 3D Point Cloud
• BORE: Bayesian Optimization by Density-Ratio Estimation
• Hyperparameter Transfer Learning with Adaptive Complexity
• Problem-fluent models for complex decision-making in autonomous materials research
• Reinforced Few-Shot Acquisition Function Learning for Bayesian Optimization
• Using Distance Correlation for Efficient Bayesian Optimization
• Bayesian Optimization with Approximate Set Kernels
• A Nonmyopic Approach to Cost-Constrained Bayesian Optimization
• Approximate Bayesian inference from noisy likelihoods with Gaussian process emulated MCMC
• Quantum evolution kernel : Machine learning on graphs with programmable arrays of qubits
• Geometric Graph Representations and Geometric Graph Convolutions for Deep Learning on Three-Dimensional (3D) Graphs
• A Simple and Fast Baseline for Tuning Large XGBoost Models
• Designing MacPherson Suspension Architectures using Bayesian Optimization
• Exploration and Exploitation in Symbolic Regression using Quality-Diversity and Evolutionary Strategies Algorithms
• Combined Global and Local Search for Optimization with Gaussian Process Models
• Function Optimization with Posterior Gaussian Derivative Process
• Contact Map based Crystal Structure Prediction using Global Optimization
• Bayesian Optimization of Area-based Road Pricing
• Sequential Bayesian Risk Set Inference for Robust Discrete Optimization via Simulation
• Quasi-Newton Quasi-Monte Carlo for variational Bayes
• Practical Bayesian Optimization of Objectives with Conditioning Variables
• Noncooperative dynamics in election interference
• Mixed Strategies for Robust Optimization of Unknown Objectives
• Rate-optimal Bayesian Simple Regret in Best Arm Identification
• An Algorithmic Theory of Metacognition in Minds and Machines
• Scaling Bayesian Optimization With Game Theory
• BOAssembler: a Bayesian Optimization Framework to Improve RNA-Seq Assembly Performance
• Formal Verification of Safety Critical Autonomous Systems via Bayesian Optimization
• Sonic: A Sampling-based Online Controller for Streaming Applications
• Uncertainty Estimation For Community Standards Violation In Online Social Networks
• Collaborative Multidisciplinary Design Optimization with Neural Networks
• Bayesian Optimisation with Formal Guarantees
• Fusing heterogeneous data sets
• Explaining Inference Queries with Bayesian Optimization
• JUMBO: Scalable Multi-task Bayesian Optimization using Offline Data
• Information Directed Reward Learning for Reinforcement Learning
• Scalable Hyperparameter Optimization with Lazy Gaussian Processes
• Comparison of Syntactic and Semantic Representations of Programs in Neural Embeddings
• Bias-Robust Bayesian Optimization via Dueling Bandits
• Quality and Computation Time in Optimization Problems
• Practical Bayesian Optimization with Threshold-Guided Marginal Likelihood Maximization
• Approximate Bayesian Computation for Physical Inverse Modeling
• Posterior Temperature Optimization in Variational Inference for Inverse Problems
• Simple and Scalable Parallelized Bayesian Optimization
• Learning How to Optimize Black-Box Functions With Extreme Limits on the Number of Function Evaluations
• Bayesian optimization for modular black-box systems with switching costs