Course scheduling problem is a nondeterministic polynomial time-complete problem which is hard to solve. One of the popular method for solving the problem is the so-called local search algorithm family, which is basically search a better solution in the neighborhood of previously known potential solution iteratively. Local search technique has a weakness that it gets stuck easily in a local optimum or on a plateau. Various efforts have been done in many literatures to circumvent this weakness. Current work investigates an already improved scattered local search by augmenting randomly generated initial population of schedules in the local search algorithm. This algorithm will be compared to an originally population based genetic algorithm which is enhanced by letting local search technique on each individual schedule in every generation. The result has shown that the population based local search performs better than the hybrid genetic algorithm.
展开▼
