Neural networks for facility location problems

Abstract

This paper presents a new self-organizing neural network approach for solving graph-based facility location problems. It is designed to have small amount of parameters and to not need much tuning. We test our approach on several groups of problems and show that it consistently finds good feasible solutions.