LGSYMar 11, 2025

An Analysis of Safety Guarantees in Multi-Task Bayesian Optimization

arXiv:2503.08555v32 citationsh-index: 3
Originality Incremental advance
AI Analysis

This work addresses safety-critical optimization problems where multiple data sources are available, but it is incremental as it builds on existing Bayesian optimization frameworks.

The paper tackles the problem of integrating multiple information sources into Bayesian optimization while maintaining safety constraints, by modeling interdependencies with an unknown correlation matrix and adjusting error bounds. The results show significant improvement in sample efficiency, demonstrated on benchmark functions and a controller optimization problem.

This paper addresses the integration of additional information sources into a Bayesian optimization framework while ensuring that safety constraints are satisfied. The interdependencies between these information sources are modeled using an unknown correlation matrix. We explore how uniform error bounds must be adjusted to maintain constraint satisfaction throughout the optimization process, considering both Bayesian and frequentist statistical perspectives. This is achieved by appropriately scaling the error bounds based on a confidence interval that can be estimated from the data. Furthermore, the efficacy of the proposed approach is demonstrated through experiments on two benchmark functions and a controller parameter optimization problem. Our results highlight a significant improvement in sample efficiency, demonstrating the methods suitability for optimizing expensive-to-evaluate functions.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes