Search

This paper proposes a hybrid heuristic that combines Variable Neighborhood Search (VNS) with Ant Colony Optimization (ACO) to solve the scheduling problem of nonrelated parallel machines with sequence dependent setup times in order to minimize the makespan. The Variable Neighborhood Search is proposed to solve the scheduling problem with a descending scheme in a first phase, with an ACO algorithm, which successively reorder the jobs in the machine with the largest makespan in a second phase. An experimental study was performed using test problems from the literature showing that the second phase of the algorithm improves the solution obtained in the first phase. The results obtained are also compared with other methods in the literature proving to be a competitive method. [ABSTRACT FROM AUTHOR]

In this paper, we present a modification of the repetitive random search algorithm. In our proposal, we change the fixed parameters which values are set by the user, by deterministic values following a chaotic map. The proposed algorithm is used to optimize four well known test functions for 10, 20 and 30 dimensions. For all cases, our proposal converges to better points that the optimal points obtained using the traditional version with fixed values in the parameters. The obtained results encourage us to continue the development and testing of the proposal algorithm with a major suite of test functions, and to compare it with other well established heuristic algorithms. [ABSTRACT FROM AUTHOR]

Often data that should be written so they are not identical due to misspellings and typos, variations in word order, use of prefixes and suffixes, among others. Phonetic techniques for duplicate detection are not geared toward the Spanish language, which makes the identification and correction of problems such as spelling errors in texts written in this language. In this research paper we propose an algorithm called PhoneticSpanish to detect duplicate text strings which considers the presence of spelling errors in Spanish. The proposed algorithm was compared with nine techniques to detect duplicates. The results were satisfactory and the algorithm that performed better than the other techniques and demonstrate opportunities for improved analysis of information in Spanish. [ABSTRACT FROM AUTHOR]

Databases are tools to store and manage information, the better they are distributed, the easier and more efficient will its access be. At present, thanks to technological development of fields such as computer science and electronics, most of them have an electronic format, since they offer a wide rank of solutions to the store data problem. In this paper, we present the design and development of a database for range images, based on the relational model and a design methodology consisting of three steps: conceptual design, logical design, and physical design. The data base has a WEB access. A multi-resolution algorithm was developed to improve the time cost in visualization and renderization, by using decimation parameters based on the curvature. A WEB interface was developed in order to provide an easy access, consequently, it is possible to enter, update, and eliminate models on a reliable and simple way. At present, the database counts on 40 models divided in 3 categories: Faces (22), Museum (4) and objects of free form (14). The use of the multi-resolution algorithm allows saving up to 96% of computational cost. [ABSTRACT FROM AUTHOR]

To obtain a segment image is still not possible. Typically, results obtained by computer programs show over-segmentation and not well-defined edges. Most of these difficulties are believed to be due to similarity criteria used by segmentation algorithms. In this paper, there is an analysis of similarity criteria most used in literature and an analysis of criteria based on Bayesian decision theory. [ABSTRACT FROM AUTHOR]