Constraints Satisfaction through Recursive Neural Networks withMixed Penalties: a Case Study

Authors:
C. Privault;L. Hérault
Affiliations:
CEA-LETI, DSYS, CEA-Grenoble, 17 rue des Martyrs, 38054 Grenoble Cedex 9, France E-mail: laurent/herault@cea.fr;CEA-LETI, DSYS, CEA-Grenoble, 17 rue des Martyrs, 38054 Grenoble Cedex 9, France E-mail: laurent/herault@cea.fr
Venue:
Neural Processing Letters
Year:
1998

Citing 4
Cited 1

Theory of linear and integer programming

Theory of linear and integer programming
Competitive neural architecture for hardware solution to the assignment problem

Neural Networks
Path-following methods for linear programming

SIAM Review
Solving a Real World Assignment Problem with a Metaheuristic

Journal of Heuristics

Solving a Real World Assignment Problem with a Metaheuristic

Journal of Heuristics

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper investigates an industrial assignment problem. It ismodelized as a constraint satisfaction problem of large size withlinear inequalities and binary variables. A new analog neuron-likenetwork is proposed to find out feasible solutions to problems havingseveral thousands of 0/1 variables.The approach developed in this paper is based on mixed-penaltyfunctions: exterior penalty functions together with interior penaltyfunctions. Starting from a near-binary solution satisfying eachlinear inequality, the network generates trial solutions locatedoutside or inside the feasible set, in order to minimize an energyfunction which measures the total binary infeasibility of the system.The performances of the network are demonstrated on real data sets.