How Bayesian Should Bayesian Optimisation Be?
arXiv:2105.00894 · doi:10.1145/3449726.3463164
Abstract
Bayesian optimisation (BO) uses probabilistic surrogate models - usually Gaussian processes (GPs) - for the optimisation of expensive black-box functions. At each BO iteration, the GP hyperparameters are fit to previously-evaluated data by maximising the marginal likelihood. However, this fails to account for uncertainty in the hyperparameters themselves, leading to overconfident model predictions. This uncertainty can be accounted for by taking the Bayesian approach of marginalising out the model hyperparameters. We investigate whether a fully-Bayesian treatment of the Gaussian process hyperparameters in BO (FBBO) leads to improved optimisation performance. Since an analytic approach is intractable, we compare FBBO using three approximate inference schemes to the maximum likelihood approach, using the Expected Improvement (EI) and Upper Confidence Bound (UCB) acquisition functions paired with ARD and isotropic Matern kernels, across 15 well-known benchmark problems for 4 observational noise settings. FBBO using EI with an ARD kernel leads to the best performance in the noise-free setting, with much less difference between combinations of BO components when the noise is increased. FBBO leads to over-exploration with UCB, but is not detrimental with EI. Therefore, we recommend that FBBO using EI with an ARD kernel as the default choice for BO.
arXiv commentTo appear in the Proceedings of Genetic and Evolutionary Computation Conference Companion (GECCO 2021), ACM. 10 pages (main paper) + 26 pages (supplement)
References in corpus (17)
- Auto-Encoding Variational Bayes
- Practical Bayesian Optimization of Machine Learning Algorithms
- Variational Inference: A Review for Statisticians
- A Tutorial on Bayesian Optimization of Expensive Cost Functions, with Application to Active User Modeling and Hierarchical Reinforcement Learning
- The No-U-Turn Sampler: Adaptively Setting Path Lengths in Hamiltonian Monte Carlo
- Variational Inference with Normalizing Flows
- GPyTorch: Blackbox Matrix-Matrix Gaussian Process Inference with GPU Acceleration
- Monte Carlo Implementation of Gaussian Process Models for Bayesian Regression and Classification
- Deep Gaussian Processes
- Automatic Differentiation Variational Inference
- Max-value Entropy Search for Efficient Bayesian Optimization
- Greed is Good: Exploration and Exploitation Trade-offs in Bayesian Optimisation
- Slice sampling covariance hyperparameters of latent Gaussian models
- Input Warping for Bayesian Optimization of Non-stationary Functions
- Approximate Inference for Fully Bayesian Gaussian Process Regression
- What do you Mean? The Role of the Mean Function in Bayesian Optimisation
- A hierarchical expected improvement method for Bayesian optimization