Browsing by Author "Puris, Amilkar"
Now showing 1 - 2 of 2
Results Per Page
Sort Options
Item Algoritmos meta heurísticos para el aprendizaje de redes bayesianas(Corporación Universitaria Lasallista, Editorial Lasallista, 2018) Oviedo, Byron; Puris, Amilkar; Zhuma, EmilioIntroduction This article aims to obtain models based on probabilistic case analysis to help decision-making in the education and learning of UTEQ students. To obtain the final product, the development process has been distributed in several stages. Objective To create a probabilistic model to evaluate and diagnose students based on a set of characteristics, which must be learned automatically through a generalization of the AutoClass model allowing the existence of hidden variables, each of them affecting a set different from observable variables (students’ answers to questions raised by an automatic learning system). Materials and methods. Our study will be carried out to define another form of structural learning based on the search of structures through evolutionary metaheuristic models. Results This model will allow the authorities of the UTEQ to identify inconveniences and setbacks in the teaching-learning process. At the same time, the results obtained will allow immediate decision-making to solve the problems detected and thus fulfill the institutional mission of training professionals with a scientific and humanistic vision capable of developing research, creating technologies, maintaining and disseminating our ancestral knowledge and culture, for the construction of solutions to the problems of the region and the country. Conclusions were metaheuristic variable mesh optimization (VMO) to structural learning of Bayesian network classifiers (BVMO).Item Uso de optimización de mallas variables para el “viajante de comercio”(Corporación Universitaria Lasallista, Editorial Lasallista, 2018) Oviedo, Byron; Zambrano-Vega, Cristian; Puris, AmilkarIntroduction This paper presents a proposal to apply the meta-heuristic Optimization Based on Variable Meshes (VMO) to the discreet problem of the Traveler Seller (TSP); This model explores the search space to Objective from a population of solutions called mesh that expands and contracts in order to find good quality solutions. In this context, the expansion operator is modified in such a way that it is applicable in a discrete domain, making combinations among the solutions in order to obtain new nodes. Another element that is modified is the clearing operator, which is responsible for maintaining the diversity of the mesh in each interaction. Methodology. A study of VMO model parameters is summarized in this paper, using a set of TSP instances with different characteristics; In addition, it can be observed that the proposal of this work obtains competitive results when compared with other international reference algorithms mentioned in the state of the art. The work is structured as follows: Section 1 describes the fundamental aspects of the TSP study problem. Then, in the second, the general functioning of VMO is explained, in the third one, each of the expansion and contraction operators is defined for the study problem. Later in the fourth section a study of the parameters of the proposal and an experimental comparative analysis with the results obtained with other algorithms mentioned in the state of the art is made. Conclusions Other operators of generation of new nodes were applied in the expansion process, where a combination of solutions is made in order to comply with the restrictions imposed by the problem.