Belén Melián-Batista

The Parallel Variable Neighborhood Search for the p -Median Problem

The Variable Neighborhood Search (VNS) is a recent metaheuristic that combines series of random and improving local searches based on systematically changed neighborhoods. When a local minimum is reached, a shake procedure performs a random search. This determines a new starting point for running an improving search. The use of interchange moves provides a simple implementation of the VNS algorithm for the p-Median Problem. Several strategies for the parallelization of the VNS are considered and coded in C using OpenMP. They are compared in a shared memory machine with large instances.

Parallelization of the scatter search for the p -median problem

This article develops several strategies for the parallelization of the metaheuristic called scatter search, which is a population-based method that constructs solutions by combining others. Three types of parallelization have been proposed to achieve either an increase of efficiency or an increase of exploration. The procedures have been coded in C using OpenMP and compared in a shared memory machine with large instances. The obtained algorithms are tested on the p-median problem.

Pre-Marshalling Problem: Heuristic solution method and instances generator

The Pre-Marshalling Problem consists in reshuffling containers in a port yard taking into account that a container with high priority cannot be placed below a container with low priority. The objective of the problem is to minimize the number of movements required to arrange all the containers so that further relocations are not necessary. In this work a heuristic solution method to solve the Pre-Marshalling Problem that significantly outperforms other methods from the literature is proposed. Moreover, an instances generator for this problem with which instances with varying degrees of difficulty can be created is developed. In order to obtain instances with degrees of difficulty that range from low difficulty up to high difficulty, two features that consider both the occupancy rate of the bay of containers and the percentage of containers with high priority that are located below containers with low priority are considered. The computational experiments carried out in this work corroborate the good performance of both the heuristic and the instances generator.

Artificial intelligence hybrid heuristic based on tabu search for the dynamic berth allocation problem

This paper considers the Dynamic Berth Allocation Problem, in which vessels are assigned to discrete positions in berths. This problem, whose goal is to minimize the total time the vessels stay at the port, constitutes one of the most important processes at any containers terminal. We propose a hybrid metaheuristic that combines Tabu Search with Path Relinking, T^2S^@?+PR. The results reached by this hybrid algorithm are compared with the optimal values given by the best mathematical model that appears in the literature for this problem, GSPP, and with a tabu search algorithm from the literature, T^2S. For small instances, the algorithm T^2S^@?+PR is able to obtain most of the optimal solutions in an amount of computational time that is lower than the time required to solve the GSPP model. For medium and large size instances, GSPP cannot be solved to optimality, whereas the proposed hybrid algorithm outperforms T^2S. Moreover, the computational experiments carried out in this paper confirm the robustness of the proposed algorithm with respect to both the parameters governing the procedure and the problem size.

Biased random key genetic algorithm for the Tactical Berth Allocation Problem

The Tactical Berth Allocation Problem (TBAP) aims to allocate incoming ships to berthing positions and assign quay crane profiles to them (i.e. number of quay cranes per time step). The goals of the TBAP are both the minimization of the housekeeping costs derived from the transshipment container flows between ships, and the maximization of the total value of the quay crane profiles assigned to the ships. In order to obtain good quality solutions with considerably short computational effort, this paper proposes a biased random key genetic algorithm for solving this problem. The computational experiments and the comparison with other solutions approaches presented in the related literature for tackling the TBAP show that the proposed algorithm is applicable to efficiently solve this difficult and essential container terminal problem. The problem instances used in this paper are composed of both, those reported in the literature and a new benchmark suite proposed in this work for taking into consideration other realistic scenarios.

High-dimensional feature selection via feature grouping

We introduce the concept of predominant group based on the idea of Markov blanket to identify groups of correlated features.We propose a greedy strategy (GreedyPGG) that groups features based on the concept of predominant groups.We propose a VNS metaheuristic that uses the GreedyPGG strategy to reduce the dimensionality in high-dimensional data.Results show that VNS finds smaller subsets of features without degrading the predictive model. In recent years, advances in technology have led to increasingly high-dimensional datasets. This increase of dimensionality along with the presence of irrelevant and redundant features make the feature selection process challenging with respect to efficiency and effectiveness. In this context, approximate algorithms are typically applied since they provide good solutions in a reasonable time. On the other hand, feature grouping has arisen as a powerful approach to reduce dimensionality in high-dimensional data. Recently, some authors have focused their attention on developing methods that combine feature grouping and feature selection to improve the model. In this paper, we propose a feature selection strategy that utilizes feature grouping to increase the effectiveness of the search. As feature selection strategy, we propose a Variable Neighborhood Search (VNS) metaheuristic. Then, we propose to group the input space into subsets of features by using the concept of Markov blankets. To the best of our knowledge, this is the first time in which the Markov blanket is used for grouping features. We test the performance of VNS by conducting experiments on several high-dimensional datasets from two different domains: microarray and text mining. We compare VNS with popular and competitive techniques. Results show that VNS is a competitive strategy capable of finding a small size of features with similar predictive power than that obtained with other algorithms used in this study.

A bi-objective vehicle routing problem with time windows: A real case in Tenerife

This work is motivated by a real problem posed to the authors by a company in Tenerife, Spain. Given a fleet of vehicles, daily routes have to be designed in order to minimize the total traveled distance while balancing the workload of drivers. This balance has been defined in relation to the length of the routes, regarding to the required time. A bi-objective mixed-integer linear model for the problem is proposed and a solution approach, based on the scatter search metaheuristic, is developed. An extensive computational experience is carried out, using benchmark instances with 25, 50 and 100 customers, to test several components of the proposed method. Comparisons with the exact Pareto fronts for instances up to 25 customers show that the proposed methods obtain good approximations. For comparison purposes, an NSGA-II algorithm has also been implemented. Results obtained on a real case instance are also discussed. In this case, the solution provided by the method proposed in this paper improves the solution implemented by the company.

Variable Neighborhood Search for a Dynamic Rich Vehicle Routing Problem with time windows

This work considers a real-world Dynamic Rich VRPTW with multiple objectives.We develop a metaheuristic algorithm based on Variable Neighborhood Search.An extensive computational experience has been carried out with several purposes.The proposed algorithm is compared with the best algorithms from the literature for the Dynamic VRPTW.The developed software has been embedded into the fleet management system of a company in Spain. A Dynamic Rich Vehicle Routing Problem with Time Windows has been tackled as a real-world application, in which customers requests can be either known at the beginning of the planning horizon or dynamically revealed over the day. Several real constraints, such as heterogeneous fleet of vehicles, multiple and soft time windows and customers priorities, are taken into consideration. Using exact methods is not a suitable solution for this kind of problems, given the fact that the arrival of a new request has to be followed by a quick re-optimization phase to include it into the solution at hand. Therefore, we have proposed a metaheuristic procedure based on Variable Neighborhood Search to solve this particular problem. The computational experiments reported in this work indicate that the proposed method is feasible to solve this real-world problem and competitive with the best results from the literature. Finally, it is worth mentioning that the software developed in this work has been inserted into the fleet management system of a company in Spain.

Hybrid Estimation of Distribution Algorithm for the Quay Crane Scheduling Problem

The competitiveness of a container terminal is highly conditioned by the time that container vessels spend on it. The proper scheduling of the quay cranes can reduce this time and allows a container terminal to be more attractive to shipping companies. The goal of the Quay Crane Scheduling Problem (QCSP) is to minimize the handling time of the available quay cranes when performing the tasks of loading and unloading containers onto/from a container vessel. This paper proposes a hybrid Estimation of Distribution Algorithm with local search to solve the QCSP. This approach includes a priori knowledge about the problem in the initialization step to reach promising regions of the search space as well as a novel restarting strategy with the aim of avoiding the premature convergence of the search. Furthermore, an approximate evaluation scheme is applied in order to reduce the computational burden. Moreover, its performance is statistically compared with the best optimization method from the literature. Numerical testing results demonstrate the high robustness and efficiency of the developed technique. Additionally, some relevant components of the scheme are individually analyzed to check their effectiveness.

A domain-specific knowledge-based heuristic for the Blocks Relocation Problem

A knowledge-based heuristic to solve the Blocks Relocation Problem is presented.This heuristic method significantly outperforms other methods from the literature.The computational tests show the effectiveness and efficiency of the heuristic. The Blocks Relocation Problem consists in minimizing the number of movements performed by a gantry crane in order to retrieve a subset of containers placed into a bay of a container yard according to a predefined order. A study on the mathematical formulations proposed in the related literature reveals that they are not suitable for its solution due to their high computational burden. Moreover, in this paper we show that, in some cases, they do not guarantee the optimality of the obtained solutions. In this regard, several optimization methods based on the well-known A? search framework are introduced to tackle the problem from an exact point of view. Using our A? algorithm we have corrected the optimal objective function value of 17 solutions out of 45 instances considered by Caserta et al. (2012) 4. In addition, this work presents a domain-specific knowledge-based heuristic algorithm to find high-quality solutions by means of short computational times. It is based on finding the most promising positions into the bay where to relocate those containers that are currently located on the next one to be retrieved, in such a way that, they do not require any additional relocation operation in the future. The computational tests indicate the higher effectiveness and efficiency of the suggested heuristic when solving real-world scenarios in comparison with the most competitive approaches from the literature.