This study is to present a particle swarm optimization (PSO) to solve a job shop scheduling problem with the objective of minimizing a makespan. The solution representation of a particle that is encoded with a permutation of all operations on jobs. The operations are ordered by precedence constraints. For the performance evaluation, each particle is evaluated by a quality of the fitness function. In the particle swarm optimization model, the effect of parameters, like the swarm size, on performance are investigated in this paper. In different swarm sizes, the performance of proposed methodology is tested on forty well-known benchmark problems from the OR-library. The computational results demonstrate that the proposed methodology can effectively solve a job shop scheduling problem and also take less processing time although run on a personal computer.

This paper proposes a novel swarm intelligence technique, which is an adaptation of Abbass’s marriage in honey-bee optimization (MBO), with the aim to achieve better overall performance than the original version of the MBO while also lowering the computation time for finding the optimal solution. The original MBO has been proven to be one of the best swarm intelligence algorithms for solving optimization problems. However, many parameters need to be properly set in order for the MBO to perform at its best. Therefore, long computation time caused by a large number of trial and error iterations involved in trying to find the right combination of parameters is unavoidable. The framework of the proposed algorithm is similar to the original MBO, which is based on the marriage behavior of honey-bees. In order to improve the efficiency of the MBO algorithm, several aspects of the original MBO have been adapted, such as (1) the proposed algorithm is adapted to obtain the ability to automatically search for the proper number of queens, (2) the proposed algorithm divides the problem space into several colonies, each of which has its own queen. In order to keep the number of colonies to a minimum, the proposed algorithm, therefore, encourages the queens to compete with each other for a larger colony and also urges the newly-born brood which is fitter than the queen of the colony to overthrow the queen. (3) the fuzzy c-means algorithm is employed to assign the drones to the proper colonies. The proposed algorithm has been evaluated and compared to the original MBO algorithm. The experimental results on six benchmark problems demonstrate the potential of the proposed algorithm in offering an efficient and effective solution to the problem.

This paper proposes a modified marriage in honey-bee optimization for solving multiobjective optimization problems. Unlike the original marriage in honey-bee optimization, the proposed algorithm divides the objective space into several colonies, each of which has its own queen. The fitness of each solution is based on 3 parameters: the size of the colony, the number of dominating solutions, and the number of dominated solutions. The nondominated solutions with highest fitness values are preferentially assigned to be the queens while the rest are assigned to be the drones. Next, all drones are assigned to the colony according to their distances from the queens of the colonies. In order to maximize a genetic variance in the population, the multiple mating is used. The multiple mating requires the queen to mate with drones from the other colonies. The proposed algorithm has been evaluated and compared to two state-of-the-art metaheuristic algorithms: the Pareto archived evolution strategy and the nondominated sorting genetic algorithm. The experimental results on 5 different ZDT benchmark functions illustrate that the proposed algorithm is able to converge to the true Pareto fronts and has better spread of solutions, as compared with the published results of the two state-of-the-art algorithms.

This paper proposes a new neural network approach to the off-line handwritten Thai character recognition. This new neural network is a hierarchical neural network; it employs the concept of a weighting matrix in measuring the similarity between the incoming input pattern and the reference patterns. The experiments have been conducted to recognize both slipshod and proper handwritten characters. The results demonstrate a very promising performance of the proposed approach.