Search

• The constrained clustering problem is studied. • An efficient local search algorithm is proposed. • A node filtering strategy is introduced for improving efficiency. • The proposed algorithm is more effective than state-of-the-art heuristics. Constrained clustering extends clustering by integrating user constraints, and aims to determine an optimal assignment under the constraints. In this paper, we propose a local search algorithm called FastCCP to solve the constrained clustering problem. In the algorithm, instances connected by must-link constraints are first merged into nodes, and then, a local search method is performed to handle the cannot-link constraints while minimizing the Within-Cluster Sum of Squares (WCSS). Several strategies are proposed to enhance the solution diversity and achieve a trade-off between constraint satisfaction and WCSS minimization during the search. Furthermore, a node-filtering strategy is proposed to improve the efficiency of the algorithm. Experiments are performed on benchmark datasets to evaluate our algorithm. The comparative results indicate that our algorithm outperforms state-of-the-art algorithms in terms of both the solution quality and CPU runtime. [ABSTRACT FROM AUTHOR]

Abstract: The service network design problem (SNDP) is a core problem in freight transportation. It involves the determination of the most cost-effective transportation network and the characteristics of the corresponding services, subject to various constraints. The scale of the problem in real-world applications is usually very large, especially when the network contains both the geographical information and the temporal constraints which are necessary for modelling multiple service-classes and dynamic events. The development of time-efficient algorithms for this problem is, therefore, crucial for successful real-world applications. Earlier research indicated that guided local search (GLS) was a promising solution method for this problem. One of the advantages of GLS is that it makes use of both the information collected during the search as well as any special structures which are present in solutions. Building upon earlier research, this paper carries out in-depth investigations into several mechanisms that could potentially speed up the GLS algorithm for the SNDP. Specifically, the mechanisms that we have looked at in this paper include a tabu list (as used by tabu search), short-term memory, and an aspiration criterion. An efficient hybrid algorithm for the SNDP is then proposed, based upon the results of these experiments. The algorithm combines a tabu list within a multi-start GLS approach, with an efficient feasibility-repairing heuristic. Experimental tests on a set of 24 well-known service network design benchmark instances have shown that the proposed algorithm is superior to a previously proposed tabu search method, reducing the computation time by over a third. In addition, we also show that far better results can be obtained when a faster linear program solver is adopted for the sub-problem solution. The contribution of this paper is an efficient algorithm, along with detailed analyses of effective mechanisms which can help to increase the speed of the GLS algorithm for the SNDP. [Copyright &y& Elsevier]