LLMs for Cold-Start Cutting Plane Separator Configuration

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Lawless, Connor, Li, Yingxi, Wikum, Anders, Udell, Madeleine, Vitercik, Ellen
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916965971394560
author Lawless, Connor
Li, Yingxi
Wikum, Anders
Udell, Madeleine
Vitercik, Ellen
author_facet Lawless, Connor
Li, Yingxi
Wikum, Anders
Udell, Madeleine
Vitercik, Ellen
contents Mixed integer linear programming (MILP) solvers expose hundreds of parameters that have an outsized impact on performance but are difficult to configure for all but expert users. Existing machine learning (ML) approaches require training on thousands of related instances, generalize poorly and can be difficult to integrate into existing solver workflows. We propose a large language model (LLM)-based framework that configures cutting plane separators using problem descriptions and solver-specific separator summaries. To reduce variance in LLM outputs, we introduce an ensembling strategy that clusters and aggregates candidate configurations into a small portfolio of high-performing configurations. Our method requires no custom solver interface, generates configurations in seconds via simple API calls, and requires solving only a small number of instances. Extensive experiments on standard synthetic and real-world MILPs show our approach matches or outperforms state-of-the-art configuration methods with a fraction of the data and computation.
format Preprint
id arxiv_https___arxiv_org_abs_2412_12038
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle LLMs for Cold-Start Cutting Plane Separator Configuration
Lawless, Connor
Li, Yingxi
Wikum, Anders
Udell, Madeleine
Vitercik, Ellen
Machine Learning
Mixed integer linear programming (MILP) solvers expose hundreds of parameters that have an outsized impact on performance but are difficult to configure for all but expert users. Existing machine learning (ML) approaches require training on thousands of related instances, generalize poorly and can be difficult to integrate into existing solver workflows. We propose a large language model (LLM)-based framework that configures cutting plane separators using problem descriptions and solver-specific separator summaries. To reduce variance in LLM outputs, we introduce an ensembling strategy that clusters and aggregates candidate configurations into a small portfolio of high-performing configurations. Our method requires no custom solver interface, generates configurations in seconds via simple API calls, and requires solving only a small number of instances. Extensive experiments on standard synthetic and real-world MILPs show our approach matches or outperforms state-of-the-art configuration methods with a fraction of the data and computation.
title LLMs for Cold-Start Cutting Plane Separator Configuration
topic Machine Learning
url https://arxiv.org/abs/2412.12038