Lagrange Neural Network for Solving Constraint Satisfaction Problem

概要

論文の詳細を見る
The constraint satisfaction problem (CSP) is a combinatorial problem to find a solution which satisfies all given constraints. The CSP can represent various problems which appear in the fields of Artificial Intelligence. In this paper, we propose a neural network called LPPH-CSP to solve the CSP. Every equilibrium point of this neural network is a solution of the CSP, and vice versa. Thus the LPPH-CSP is not trapped by any point which is not a solution of the CSP. We compare the LPPH-CSP with the GENET which is a famous CSP-solver. Experimental results show that our method is as efficient as the GENET. However our neural network can update all neuron simultaneously for solving the CSP. In contrast, the GENET must update variables sequentially. We consider that this is an advantage for the VLSI implementation.
社団法人電子情報通信学会の論文
2005-03-22