Constrained Preferential Bayesian Optimization and Its Application in Banner Ad Design

2025-05-16

AI Summary

Existing preference-based Bayesian optimization (PBO) struggles to handle inequality constraints, limiting its applicability in real-world constrained human-AI collaborative optimization tasks. This paper proposes Constrained Preference-Based Bayesian Optimization (CPBO), the first framework to explicitly incorporate inequality constraints into the PBO paradigm, enabling optimization of human subjective preferences while satisfying hard constraints (e.g., minimum ad click-through rate). Methodologically, CPBO introduces a constraint-aware Gaussian process preference model and a novel constraint-weighted acquisition function to guide efficient exploration within the feasible region. In a user study with professional advertising designers, CPBO significantly accelerates convergence to preference-optimal solutions under constraints and demonstrates superior practical utility. Results validate CPBO’s effectiveness and generalizability for human-AI co-creative optimization in realistic constrained settings.

Technology Category

• Constraint Satisfaction and Optimization: Constraint Optimization
• Humans and AI: Learning Human Values and Preferences
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Fairness, privacy, and diversity in economic environments
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Preferential Bayesian optimization (PBO) is a variant of Bayesian optimization that observes relative preferences (e.g., pairwise comparisons) instead of direct objective values, making it especially suitable for human-in-the-loop scenarios. However, real-world optimization tasks often involve inequality constraints, which existing PBO methods have not yet addressed. To fill this gap, we propose constrained preferential Bayesian optimization (CPBO), an extension of PBO that incorporates inequality constraints for the first time. Specifically, we present a novel acquisition function for this purpose. Our technical evaluation shows that our CPBO method successfully identifies optimal solutions by focusing on exploring feasible regions. As a practical application, we also present a designer-in-the-loop system for banner ad design using CPBO, where the objective is the designer's subjective preference, and the constraint ensures a target predicted click-through rate. We conducted a user study with professional ad designers, demonstrating the potential benefits of our approach in guiding creative design under real-world constraints.

Problem

• Extends PBO to handle inequality constraints
• Proposes novel acquisition function for CPBO
• Applies CPBO in designer-in-the-loop ad design

Innovation

• Extends PBO with inequality constraints handling
• Introduces novel acquisition function for CPBO
• Applies CPBO in designer-in-the-loop ad design