Achieving Robustness to Aleatoric Uncertainty with Heteroscedastic Bayesian Optimisation
arXiv:1910.07779 · doi:10.1088/2632-2153/ac298c
Abstract
Bayesian optimisation is a sample-efficient search methodology that holds great promise for accelerating drug and materials discovery programs. A frequently-overlooked modelling consideration in Bayesian optimisation strategies however, is the representation of heteroscedastic aleatoric uncertainty. In many practical applications it is desirable to identify inputs with low aleatoric noise, an example of which might be a material composition which consistently displays robust properties in response to a noisy fabrication process. In this paper, we propose a heteroscedastic Bayesian optimisation scheme capable of representing and minimising aleatoric noise across the input space. Our scheme employs a heteroscedastic Gaussian process (GP) surrogate model in conjunction with two straightforward adaptations of existing acquisition functions. First, we extend the augmented expected improvement (AEI) heuristic to the heteroscedastic setting and second, we introduce the aleatoric noise-penalised expected improvement (ANPEI) heuristic. Both methodologies are capable of penalising aleatoric noise in the suggestions and yield improved performance relative to homoscedastic Bayesian optimisation and random sampling on toy problems as well as on two real-world scientific datasets. Code is available at: \url{https://github.com/Ryan-Rhys/Heteroscedastic-BO}
arXiv commentPublished in Machine Learning: Science and Technology 2021 (https://iopscience.iop.org/article/10.1088/2632-2153/ac298c) Earlier version accepted to the 2019 NeurIPS Workshop on Safety and Robustness in Decision Making
References in corpus (22)
- Array Programming with NumPy
- Automatic chemical design using a data-driven continuous representation of molecules
- Accelerating high-throughput virtual screening through molecular pool-based active learning
- Practical heteroskedastic Gaussian process modeling for large simulation experiments
- Replication or exploration? Sequential design for stochastic simulation experiments
- Parallel and Distributed Thompson Sampling for Large-scale Accelerated Exploration of Chemical Space
- Bayesian Optimization of Combinatorial Structures
- Olympus: a benchmarking framework for noisy optimization and experiment planning
- Maximizing acquisition functions for Bayesian optimization
- A General Framework for Multi-fidelity Bayesian Optimization with Gaussian Processes
- ChemBO: Bayesian Optimization of Small Organic Molecules with Synthesizable Recommendations
- Constrained Bayesian Optimization for Automatic Chemical Design
- Latent Molecular Optimization for Targeted Therapeutic Design
- Gaussian Process Regression with Heteroscedastic or Non-Gaussian Residuals
- Heteroscedastic Gaussian processes for uncertainty modeling in large-scale crowdsourced traffic data
- Heteroscedastic Treed Bayesian Optimisation
- Gaussian Process Molecule Property Prediction with FlowMO
- Are we Forgetting about Compositional Optimisers in Bayesian Optimisation?
- Adaptive Sensor Placement for Continuous Spaces
- Kernels over Sets of Finite Sets using RKHS Embeddings, with Application to Bayesian (Combinatorial) Optimization
- A COLD Approach to Generating Optimal Samples
- Optimizing Photonic Nanostructures via Multi-fidelity Gaussian Processes