TREGO: a trust-region framework for efficient global optimization
From MaRDI portal
Abstract: Efficient Global Optimization (EGO) is the canonical form of Bayesian optimization that has been successfully applied to solve global optimization of expensive-to-evaluate black-box problems. However, EGO struggles to scale with dimension, and offers limited theoretical guarantees. In this work, a trust-region framework for EGO (TREGO) is proposed and analyzed. TREGO alternates between regular EGO steps and local steps within a trust region. By following a classical scheme for the trust region (based on a sufficient decrease condition), the proposed algorithm enjoys global convergence properties, while departing from EGO only for a subset of optimization steps. Using extensive numerical experiments based on the well-known COCO {bound constrained problems}, we first analyze the sensitivity of TREGO to its own parameters, then show that the resulting algorithm is consistently outperforming EGO and getting competitive with other state-of-the-art black-box optimization methods.
Recommendations
- An analysis of covariance parameters in Gaussian process-based optimization
- Batch sequential adaptive designs for global optimization
- Tracking global optima in dynamic environments with efficient global optimization
- Convergence rates of efficient global optimization algorithms
- No-regret Bayesian optimization with unknown hyperparameters
Cites work
- A merit function approach for direct search
- A merit function approach for evolution strategies
- A particle swarm pattern search method for bound constrained global optimization
- A progressive barrier for derivative-free nonlinear programming
- A stochastic Levenberg-Marquardt method using random models with complexity results
- Algorithm 909: NOMAD: nonlinear optimization with the MADS algorithm
- Algorithm 1027: NOMAD Version 4: Nonlinear Optimization with the MADS Algorithm
- Analysis of direct searches for discontinuous functions
- Analysis of Generalized Pattern Searches
- Bayesian optimization in a billion dimensions via random embeddings
- COCO: a platform for comparing continuous optimizers in a black-box setting
- Convergence properties of the expected improvement algorithm with fixed mean and covariance functions
- Convergence rates of efficient global optimization algorithms
- Derivative-free and blackbox optimization
- Derivative-free optimization: a review of algorithms and comparison of software implementations
- Efficient global optimization for high-dimensional constrained problems by using the Kriging models combined with the partial least squares method
- Efficient global optimization of expensive black-box functions
- Gaussian processes for machine learning.
- Global optimization by multilevel coordinate search
- Globally convergent evolution strategies
- Globally convergent evolution strategies for constrained optimization
- scientific article; zbMATH DE number 2002582 (Why is no real title available?)
- scientific article; zbMATH DE number 944037 (Why is no real title available?)
- scientific article; zbMATH DE number 5060482 (Why is no real title available?)
- scientific article; zbMATH DE number 2231192 (Why is no real title available?)
- Information-Theoretic Regret Bounds for Gaussian Process Optimization in the Bandit Setting
- Introduction to Derivative-Free Optimization
- Mesh Adaptive Direct Search Algorithms for Constrained Optimization
- Multi-fidelity optimization via surrogate modelling
- Noisy kriging-based optimization methods: a unified implementation within the DiceOptim package
- Optimization and nonsmooth analysis
- Stochastic mesh adaptive direct search for blackbox optimization using probabilistic estimates
- Stochastic optimization using a trust-region method and random models
Cited in
(6)- Tracking global optima in dynamic environments with efficient global optimization
- An analysis of covariance parameters in Gaussian process-based optimization
- Feasible set estimation under functional uncertainty by Gaussian process modelling
- A cooperative approach to efficient global optimization
- A principle for global optimization with gradients
- Explainable parameter calibration via importance-driven sequential design with an application to building energy systems
This page was built for publication: TREGO: a trust-region framework for efficient global optimization
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6102171)
