Elicitation-Augmented Bayesian Optimization

2026-05-12

AI Summary

This work addresses a key limitation in traditional human-in-the-loop Bayesian optimization, where expert prior knowledge must be explicitly quantified despite often being implicit and difficult to articulate precisely. The authors propose a novel framework that treats pairwise comparisons provided by experts as noisy observations of the underlying objective function, integrating them with direct function evaluations within a unified Gaussian process model. A cost-aware acquisition function is introduced to dynamically balance the trade-off between expensive direct evaluations and cheaper pairwise queries. This approach uniquely formalizes pairwise comparisons as computable inputs representing implicit expert knowledge, substantially improving sample efficiency when query costs are low. When pairwise queries become costly or highly noisy, the method gracefully degrades to standard Bayesian optimization, robustly approximating the convex hull of trajectories informed by multiple sources of information.

Abstract

Human-in-the-loop Bayesian optimization (HITL BO) methods utilize human expertise to improve the sample-efficiency of BO. Most HITL BO methods assume that a domain expert can quantify their knowledge, for instance by pinpointing query locations or specifying their prior beliefs about the location of the maximum as a probability distribution. However, since human expertise is often tacit and cannot be explicitly quantified, we consider a setting where domain knowledge of an expert is elicited via pairwise comparisons of designs. We interpret the expert's pairwise judgements as noisy evidence about the values of the observable objective function and develop a principled method for combining the information obtained via direct observations and pairwise queries. Specifically, we derive a cost-aware value-of-information acquisition function that balances direct observations against pairwise queries. The proposed method approaches the convex hull of the trajectories of the individual information sources: when pairwise queries are cheap it substantially improves sample-efficiency over observation-only BO, and when pairwise queries are costly or noisy, it recovers the performance of standard BO by relying on direct observations alone.

Problem

• Bayesian optimization
• human-in-the-loop
• pairwise comparisons
• expert knowledge elicitation
• sample efficiency

Innovation

• Human-in-the-loop Bayesian Optimization
• Pairwise Comparisons
• Value-of-Information Acquisition Function
• Elicitation-Augmented Optimization
• Cost-aware Sampling