Decision-Focused Surrogate Modeling for Mixed-Integer Linear Optimization

TL;DR

This paper introduces a decision-focused surrogate modeling approach that constructs efficient linear program approximations of mixed-integer linear programs, enabling faster solutions in real-time decision-making while maintaining solution quality.

Contribution

The authors develop a data-driven, decision-focused method for creating surrogate LP models that incorporate original constraints, improving efficiency and accuracy over neural-network proxies.

Findings

Surrogate LPs achieve near-optimal solutions comparable to original MILPs.

The approach is highly data-efficient and outperforms neural-network-based proxies.

Case studies demonstrate significant improvements in solution speed and feasibility.

Abstract

Mixed-integer optimization is at the core of many online decision-making systems that demand frequent updates of decisions in real time. However, due to their combinatorial nature, mixed-integer linear programs (MILPs) can be difficult to solve, rendering them often unsuitable for time-critical online applications. To address this challenge, we develop a data-driven approach for constructing surrogate optimization models in the form of linear programs (LPs) that can be solved much more efficiently than the corresponding MILPs. We train these surrogate LPs in a decision-focused manner such that for different model inputs, they achieve the same or close to the same optimal solutions as the original MILPs. One key advantage of the proposed method is that it allows the incorporation of all the original MILP's linear constraints, which significantly increases the likelihood of obtaining feasible predicted solutions. Results from two computational case studies indicate that this decision-focused surrogate modeling approach is highly data-efficient and provides very accurate predictions of the optimal solutions. In these examples, it outperforms more commonly used neural-network-based optimization proxies.