Preference-Driven Multi-Objective Combinatorial Optimization with Conditional Computation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Fan, Mingfeng, Zhou, Jianan, Zhang, Yifeng, Wu, Yaoxin, Chen, Jinbiao, Sartoretti, Guillaume Adrien
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910060366528512
author Fan, Mingfeng
Zhou, Jianan
Zhang, Yifeng
Wu, Yaoxin
Chen, Jinbiao
Sartoretti, Guillaume Adrien
author_facet Fan, Mingfeng
Zhou, Jianan
Zhang, Yifeng
Wu, Yaoxin
Chen, Jinbiao
Sartoretti, Guillaume Adrien
contents Recent deep reinforcement learning methods have achieved remarkable success in solving multi-objective combinatorial optimization problems (MOCOPs) by decomposing them into multiple subproblems, each associated with a specific weight vector. However, these methods typically treat all subproblems equally and solve them using a single model, hindering the effective exploration of the solution space and thus leading to suboptimal performance. To overcome the limitation, we propose POCCO, a novel plug-and-play framework that enables adaptive selection of model structures for subproblems, which are subsequently optimized based on preference signals rather than explicit reward values. Specifically, we design a conditional computation block that routes subproblems to specialized neural architectures. Moreover, we propose a preference-driven optimization algorithm that learns pairwise preferences between winning and losing solutions. We evaluate the efficacy and versatility of POCCO by applying it to two state-of-the-art neural methods for MOCOPs. Experimental results across four classic MOCOP benchmarks demonstrate its significant superiority and strong generalization.
format Preprint
id arxiv_https___arxiv_org_abs_2506_08898
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Preference-Driven Multi-Objective Combinatorial Optimization with Conditional Computation
Fan, Mingfeng
Zhou, Jianan
Zhang, Yifeng
Wu, Yaoxin
Chen, Jinbiao
Sartoretti, Guillaume Adrien
Artificial Intelligence
Recent deep reinforcement learning methods have achieved remarkable success in solving multi-objective combinatorial optimization problems (MOCOPs) by decomposing them into multiple subproblems, each associated with a specific weight vector. However, these methods typically treat all subproblems equally and solve them using a single model, hindering the effective exploration of the solution space and thus leading to suboptimal performance. To overcome the limitation, we propose POCCO, a novel plug-and-play framework that enables adaptive selection of model structures for subproblems, which are subsequently optimized based on preference signals rather than explicit reward values. Specifically, we design a conditional computation block that routes subproblems to specialized neural architectures. Moreover, we propose a preference-driven optimization algorithm that learns pairwise preferences between winning and losing solutions. We evaluate the efficacy and versatility of POCCO by applying it to two state-of-the-art neural methods for MOCOPs. Experimental results across four classic MOCOP benchmarks demonstrate its significant superiority and strong generalization.
title Preference-Driven Multi-Objective Combinatorial Optimization with Conditional Computation
topic Artificial Intelligence
url https://arxiv.org/abs/2506.08898