Predictive Entropy Search for Efficient Global Optimization of Black-box Functions

Abstract

We propose a novel information-theoretic approach for Bayesian optimization called Predictive Entropy Search (PES). At each iteration, PES selects the next evaluation point that maximizes the expected information gained with respect to the global maximum. PES codifies this intractable acquisition function in terms of the expected reduction in the differential entropy of the predictive distribution. This reformulation allows PES to obtain approximations that are both more accurate and efficient than other alternatives such as Entropy Search (ES). Furthermore, PES can easily perform a fully Bayesian treatment of the model hyperparameters while ES cannot. We evaluate PES in both synthetic and real-world applications, including optimization problems in machine learning, finance, biotechnology, and robotics. We show that the increased accuracy of PES leads to significant gains in optimization performance.

References in corpus (3)

• 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
• Adaptive Hamiltonian and Riemann Manifold Monte Carlo Samplers

Cited by in corpus (150)

• Hyper-Parameter Optimization: A Review of Algorithms and Applications
• A Tutorial on Bayesian Optimization
• Dealing with Integer-valued Variables in Bayesian Optimization with Gaussian Processes
• Dealing with Categorical and Integer-valued Variables in Bayesian Optimization with Gaussian Processes
• CMA-ES for Hyperparameter Optimization of Deep Neural Networks
• Gryffin: An algorithm for Bayesian optimization of categorical variables informed by expert knowledge
• Predictive Entropy Search for Multi-objective Bayesian Optimization
• Fast Bayesian Optimization of Machine Learning Hyperparameters on Large Datasets
• An Efficient Batch Constrained Bayesian Optimization Approach for Analog Circuit Synthesis via Multi-objective Acquisition Ensemble
• Predictive Entropy Search for Bayesian Optimization with Unknown Constraints
• Batched Large-scale Bayesian Optimization in High-dimensional Spaces
• Multi-fidelity Bayesian Optimisation with Continuous Approximations
• Predictive Entropy Search for Multi-objective Bayesian Optimization with Constraints
• Parallel and Distributed Thompson Sampling for Large-scale Accelerated Exploration of Chemical Space
• Bayesian Optimization of Combinatorial Structures
• Multi-Fidelity Cost-Aware Bayesian Optimization
• ChemBO: Bayesian Optimization of Small Organic Molecules with Synthesizable Recommendations
• Parallel Predictive Entropy Search for Batch Global Optimization of Expensive Objective Functions
• Bayesian Optimization with Output-Weighted Optimal Sampling
• An Efficient Multi-fidelity Bayesian Optimization Approach for Analog Circuit Synthesis
• Tuning Hyperparameters without Grad Students: Scalable and Robust Bayesian Optimisation with Dragonfly
• Information-Guided Robotic Maximum Seek-and-Sample in Partially Observable Continuous Environments
• An Entropy Search Portfolio for Bayesian Optimization
• Parallel Bayesian Optimization of Multiple Noisy Objectives with Expected Hypervolume Improvement
• Bayesian Hyperparameter Optimization for Ensemble Learning
• Bayesian Optimization with Adaptive Kernels for Robot Control
• Efficient acquisition rules for model-based approximate Bayesian computation
• Self-Optimizing Grinding Machines using Gaussian Process Models and Constrained Bayesian Optimization
• Batched Gaussian Process Bandit Optimization via Determinantal Point Processes
• Cost-aware Bayesian Optimization
• Uncertainty as a Form of Transparency: Measuring, Communicating, and Using Uncertainty
• A Latent Variable Approach for Non-Hierarchical Multi-Fidelity Adaptive Sampling
• Multi-fidelity Bayesian Optimization with Max-value Entropy Search and its parallelization
• Incorporating Expert Prior in Bayesian Optimisation via Space Warping
• Multi-Fidelity Bayesian Optimization via Deep Neural Networks
• MEMe: An Accurate Maximum Entropy Method for Efficient Approximations in Large-Scale Machine Learning
• Inference of Regulatory Networks Through Temporally Sparse Data
• Noisy-Input Entropy Search for Efficient Robust Bayesian Optimization
• Scalable Constrained Bayesian Optimization
• Hyperparameter Learning via Distributional Transfer
• Critically Examining the Claimed Value of Convolutions over User-Item Embedding Maps for Recommender Systems
• Expected Coordinate Improvement for High-Dimensional Bayesian Optimization
• Why Non-myopic Bayesian Optimization is Promising and How Far Should We Look-ahead? A Study via Rollout
• On Simulation and Trajectory Prediction with Gaussian Process Dynamics
• Extending Expected Improvement for High-dimensional Stochastic Optimization of Expensive Black-Box Functions
• Unbounded Bayesian Optimization via Regularization
• Suggesting Cooking Recipes Through Simulation and Bayesian Optimization
• A Probabilistic Machine Learning Approach to Scheduling Parallel Loops with Bayesian Optimization
• Actively Learning what makes a Discrete Sequence Valid
• Practical Bayesian Optimization for Variable Cost Objectives
• Constant-Time Predictive Distributions for Gaussian Processes
• Neural-BO: A Black-box Optimization Algorithm using Deep Neural Networks
• Bayesian Optimization using Deep Gaussian Processes
• Multi-objective Bayesian Optimization using Pareto-frontier Entropy
• Query-limited Black-box Attacks to Classifiers
• Output Space Entropy Search Framework for Multi-Objective Bayesian Optimization
• A sequential Monte Carlo approach to Thompson sampling for Bayesian optimization
• Multi-level CNN for lung nodule classification with Gaussian Process assisted hyperparameter optimization
• Efficiently Sampling Functions from Gaussian Process Posteriors
• Fast Matrix Square Roots with Applications to Gaussian Processes and Bayesian Optimization
• Max-value Entropy Search for Multi-Objective Bayesian Optimization with Constraints
• Minimum Regret Search for Single- and Multi-Task Optimization
• Knowing The What But Not The Where in Bayesian Optimization
• Meta-Learning Acquisition Functions for Transfer Learning in Bayesian Optimization
• Epsilon-Greedy Thompson Sampling to Bayesian Optimization
• Parallel Predictive Entropy Search for Multi-objective Bayesian Optimization with Constraints
• Batch simulations and uncertainty quantification in Gaussian process surrogate approximate Bayesian computation
• Integrals over Gaussians under Linear Domain Constraints
• Pareto Front-Diverse Batch Multi-Objective Bayesian Optimization
• Bayesian Optimization with Binary Auxiliary Information
• Towards Information-Seeking Agents
• Corruption-Tolerant Gaussian Process Bandit Optimization
• PHOENICS: A universal deep Bayesian optimizer
• Bayesian Optimization for Policy Search via Online-Offline Experimentation
• Learned Hardware/Software Co-Design of Neural Accelerators
• Efficient Exploration in Binary and Preferential Bayesian Optimization
• Multi-objective Asynchronous Successive Halving
• GIBBON: General-purpose Information-Based Bayesian OptimisatioN
• Tight Regret Bounds for Bayesian Optimization in One Dimension
• Towards Assessing the Impact of Bayesian Optimization's Own Hyperparameters
• Kernels over Sets of Finite Sets using RKHS Embeddings, with Application to Bayesian (Combinatorial) Optimization
• Non-Myopic Multifidelity Bayesian Optimization
• Distributionally Robust Bayesian Quadrature Optimization
• Stochastic Variational Inference for Bayesian Sparse Gaussian Process Regression
• Advancing Bayesian Optimization: The Mixed-Global-Local (MGL) Kernel and Length-Scale Cool Down
• Budgeted Batch Bayesian Optimization With Unknown Batch Sizes
• Adaptive Expansion Bayesian Optimization for Unbounded Global Optimization
• Information-Theoretic Multi-Objective Bayesian Optimization with Continuous Approximations
• A New Knowledge Gradient-based Method for Constrained Bayesian Optimization
• Multipoint-BAX: A New Approach for Efficiently Tuning Particle Accelerator Emittance via Virtual Objectives
• Cautious Bayesian Optimization for Efficient and Scalable Policy Search
• Incorporating Expert Prior Knowledge into Experimental Design via Posterior Sampling
• Pareto-efficient Acquisition Functions for Cost-Aware Bayesian Optimization
• An Asymptotically Optimal Multi-Armed Bandit Algorithm and Hyperparameter Optimization
• Nonmyopic Gaussian Process Optimization with Macro-Actions
• Bayesian Algorithm Execution: Estimating Computable Properties of Black-box Functions Using Mutual Information
• Adaptive Rate of Convergence of Thompson Sampling for Gaussian Process Optimization
• BOSH: Bayesian Optimization by Sampling Hierarchically
• Regret bounds for meta Bayesian optimization with an unknown Gaussian process prior
• Discretization-free Knowledge Gradient Methods for Bayesian Optimization
• Scalable Thompson Sampling using Sparse Gaussian Process Models
• Quantity vs. Quality: On Hyperparameter Optimization for Deep Reinforcement Learning
• Active Contextual Entropy Search
• BoA-PTA, A Bayesian Optimization Accelerated Error-Free SPICE Solver
• Practical Batch Bayesian Optimization for Less Expensive Functions
• No-Regret Bayesian Optimization with Unknown Hyperparameters
• Bayesian Optimisation for Constrained Problems
• Bayesian Optimization for Probabilistic Programs
• Bayesian Optimization for Iterative Learning
• Optimizing Photonic Nanostructures via Multi-fidelity Gaussian Processes
• Lenient Regret and Good-Action Identification in Gaussian Process Bandits
• Sequential- and Parallel- Constrained Max-value Entropy Search via Information Lower Bound
• Using Distance Correlation for Efficient Bayesian Optimization
• Active multi-fidelity Bayesian online changepoint detection
• Time-varying Gaussian Process Bandit Optimization with Non-constant Evaluation Time
• Uncertainty Quantification for Bayesian Optimization
• Batch Multi-Fidelity Bayesian Optimization with Deep Auto-Regressive Networks
• Reinforced Few-Shot Acquisition Function Learning for Bayesian Optimization
• A Nonmyopic Approach to Cost-Constrained Bayesian Optimization
• Efficient Transfer Bayesian Optimization with Auxiliary Information
• Sparse Spectrum Gaussian Process for Bayesian Optimization
• Transfer Learning Based Co-surrogate Assisted Evolutionary Bi-objective Optimization for Objectives with Non-uniform Evaluation Times
• Efficient Batch Black-box Optimization with Deterministic Regret Bounds
• Gaussian Process Bandit Optimization of the Thermodynamic Variational Objective
• BORE: Bayesian Optimization by Density-Ratio Estimation
• Reinforcement Learning To Adapt Speech Enhancement to Instantaneous Input Signal Quality
• Bayesian Optimization for Categorical and Category-Specific Continuous Inputs
• A General Framework for Constrained Bayesian Optimization using Information-based Search
• Quantization Algorithms for Random Fourier Features
• Multi-Fidelity Multi-Objective Bayesian Optimization: An Output Space Entropy Search Approach
• Optimising Placement of Pollution Sensors in Windy Environments
• Efficient Bayesian Optimization using Multiscale Graph Correlation
• Funneled Bayesian Optimization for Design, Tuning and Control of Autonomous Systems
• Efficient Rollout Strategies for Bayesian Optimization
• Accelerating Psychometric Screening Tests With Bayesian Active Differential Selection
• Lookahead Acquisition Functions for Finite-Horizon Time-Dependent Bayesian Optimization and Application to Quantum Optimal Control
• Gaussian Process Sampling and Optimization with Approximate Upper and Lower Bounds
• Neural fidelity warping for efficient robot morphology design
• Recursive Two-Step Lookahead Expected Payoff for Time-Dependent Bayesian Optimization
• Contextual Bayesian optimization with binary outputs
• Classified Regression for Bayesian Optimization: Robot Learning with Unknown Penalties
• Trusted-Maximizers Entropy Search for Efficient Bayesian Optimization
• Bayesian Optimization using Pseudo-Points
• Beyond Marginal Uncertainty: How Accurately can Bayesian Regression Models Estimate Posterior Predictive Correlations?
• One-parameter family of acquisition functions for efficient global optimization
• Targeted Active Learning for Bayesian Decision-Making
• OpEvo: An Evolutionary Method for Tensor Operator Optimization
• Doubly Bayesian Optimization
• Covariance Function Pre-Training with m-Kernels for Accelerated Bayesian Optimisation
• Scalable Bayesian Optimization with Sparse Gaussian Process Models