Search

Artificial neural network models have been successfully applied to solve a job-shop scheduling problem (JSSP) known as a Nonpolynomial (NP-complete) constraint satisfaction problem. Our main contribution is an improvement of the algorithm proposed in the literature. It consists in using a procedure optimizing the initial value of the starting time. The aim is to speed a Hopfield Neural Network (HNN) and therefore reduce the number of searching cycles. This new heuristic provides several advantages; mainly to improve the searching speed of an optimal or near optimal solution of a deterministic JSSP using HNN and reduce the makespan. Simulation results of the proposed method have been performed on various benchmarks and compared with current algorithms such as genetic algorithm, constraint satisfaction adaptive neural networks, simulated annealing, threshold accepting, flood method, and priority rules such as shortest processing time (SPT) to mention a few. As the simulation results show, and Brandts algorithm, combined with the proposed heuristic method, is efficient with respect to the resolution speed, quality of the solution, and the reduction of the computation time. [ABSTRACT FROM AUTHOR]

In this paper, we present a new pruning algorithm formed by combining the Statistical Stepwise Method (SSM) [1] with the Iterative Pruning (IP) [4] algorithms. This proposed algorithm (SSIP) is used to simultaneously remove unnecessary neurons or weight connections from a given feed-forward neural network (NN) in order to "optimize" its structure. Some modifications to the previous pruning algorithms published in [1] and [4] are also reported. Two versions of the combined SSIP are considered: In the first version, SSIP1, the modified IP is first applied to the given neural network in order to prune insignificant units, and then the modified SSM is applied to the pruned network to remove unnecessary links. In the second version, SSIP2, the above procedure is applied to each layer in turn, working from the input layer to the output layer. The performances of the algorithms are compared using two real world applications, brain disease detection and texture classification, and the superiority of the SSIP pruning algorithm is demonstrated. This new algorithm can eliminate approximately 59% of the links in the initial oversized network in order to improve performance by approximately +39% and +26% for the sensitivity of learning, and generalization, respectively. [ABSTRACT FROM AUTHOR]