LISTEN to Your Preferences: An LLM Framework for Multi-Objective Selection

Adam Jovine, Tinghan Ye, David Shmoys, Peter I. Frazier

Keywords: LLM, Preference Learning, Multi Objective Optimization

Abstract:

Multi-objective optimization often produces large sets of Pareto-optimal solutions, creating a bottleneck for human experts who must select the best option. This difficulty is compounded by the fact that expert preferences are often complex and hard to formalize. To address this, we introduce LISTEN, a framework that leverages a large language model (LLM) as a zero-shot preference oracle, guided only by an expert's high-level priorities in natural language.
To operate within LLM constraints like context windows and inference costs, we propose two iterative algorithms: LISTEN-U, which uses the LLM to refine a parametric utility function, and LISTEN-T, a non-parametric method that performs tournament-style selections over small batches of solutions.

Submission Number: 199