Persona: Fernández Galán, Severino
Cargando...
Dirección de correo electrónico
ORCID
0000-0003-3767-7480
Fecha de nacimiento
Proyectos de investigación
Unidades organizativas
Puesto de trabajo
Apellidos
Fernández Galán
Nombre de pila
Severino
Nombre
6 resultados
Resultados de la búsqueda
Mostrando 1 - 6 de 6
Publicación Comparative Evaluation of Region Query Strategies for DBSCAN Clustering(Elsevier, 2019-10) Fernández Galán, SeverinoClustering is a technique that allows data to be organized into groups of similar objects. DBSCAN (Density-Based Spatial Clustering of Applications with Noise) constitutes a popular clustering algorithm that relies on a density-based notion of cluster and is designed to discover clusters of arbitrary shape. The computational complexity of DBSCAN is dominated by the calculation of the ϵ-neighborhood for every object in the dataset. Thus, the efficiency of DBSCAN can be improved in two different ways: (1) by reducing the overall number of ϵ-neighborhood queries (also known as region queries), or (2) by reducing the complexity of the nearest neighbor search conducted for each region query. This paper deals with the first issue by considering the most relevant region query strategies for DBSCAN, all of them characterized by inspecting the neighborhoods of only a subset of the objects in the dataset. We comparatively evaluate these region query strategies (or DBSCAN variants) in terms of clustering effectiveness and efficiency; additionally, a novel region query strategy is introduced in this work. The results show that some specific DBSCAN variants are only slightly inferior to DBSCAN in terms of effectiveness, while greatly improving its efficiency. Among these variants, the novel one outperforms the rest.Publicación Comparative Evaluation of the Fast Marching Method and the Fast Evacuation Method for Heterogeneous Media(Taylor & Francis, 2021-08-30) Fernández Galán, SeverinoThe evacuation problem is usually addressed by assuming homogeneous media where pedestrians move freely in the presence of several exits and obstacles. From a more general perspective, this work considers heterogeneous media in which the velocity of pedestrians depends on their location. We use cellular automata with a floor field that indicates promis- ing movements to pedestrians and, in this context, we extend two competitive evacuation methods in order for them to be applied to heterogeneous media: the Fast Marching Method and the Fast Evacuation Method. Furthermore, we evaluate the performance that these two methods exhibit over different simulated scenarios characterized by the presence of hetero- geneous media. The resulting winning method in terms of evacuation effectiveness is greatly influenced by the particular problem being simulated.Publicación Self-Adaptive Polynomial Mutation in NSGA-II(Springer, 2023-08-21) Carles Bou, José Luis; Fernández Galán, SeverinoEvolutionary multi-objective optimization is a field that has experienced a rapid growth in the last two decades. Although an important number of new multi-objective evolutionary algorithms have been designed and implemented by the scientific community, the popular Non-Dominated Sorting Genetic Algorithm (NSGA-II) remains as a widely used baseline for algorithm performance comparison purposes and applied to different engineering problems. Since every evolutionary algorithm needs several parameters to be set up in order to operate, parameter control constitutes a crucial task for obtaining an effective and efficient performance in its execution. However, despite the advancements in parameter control for evolutionary algorithms, NSGA-II has been mainly used in the literature with fine-tuned static parameters. This paper introduces a novel and computationally lightweight self-adaptation mechanism for controlling the distribution index parameter of the polynomial mutation operator usually employed by NSGA-II in particular and by multi-objective evolutionary algorithms in general. Additionally, the classical NSGA-II using polynomial mutation with a static distribution index is compared with this new version utilizing a self-adapted parameter. The experiments carried out over twenty-five benchmark problems show that the proposed modified NSGA-II with a self-adaptive mutator outperforms its static counterpart in more than 75% of the problems using three quality metrics (hypervolume, generalized spread, and modified inverted generational distance).Publicación Fast Evacuation Method: using an effective dynamic floor field based on efficient pedestrian assignment(Elsevier, 2019-12) Fernández Galán, SeverinoThe problem of pedestrian evacuation can be addressed through cellular automata incorporating a floor field that indicates promising movements to pedestrians. The two main types of floor field are the static, which represents the shortest path from each cell to an exit (and is usually combined with dynamic measures such as the density or distribution of pedestrians), and the dynamic, which represents the quickest path from each cell to an exit. The second type has been widely used recently, since it gives rise to more efficient and realistic simulations of pedestrian dynamics. The goal of these two types of floor field is to minimize the travel time for each pedestrian; however, this paper tackles the evacuation problem from a different perspective: The time taken by the whole evacuation process is optimized. For that purpose, a floor field is constructed by assigning pedestrians to exits such that the estimated time for complete evacuation is minimized. An experimental evaluation is conducted to compare the new fast evacuation method with competitive methods using floor fields based on quickest paths: Flood Fill and the Fast Marching Method. The results show that the new method is effective in terms of the number of time steps for complete evacuation and efficient regarding the total simulation runtime.Publicación Minimum Modulus Visualization of Algebraic Fractals(Elsevier, 2023-08) Fernández Galán, SeverinoFractals are a family of shapes formed by irregular and fragmented patterns. They can be classified into two main groups: geometric and algebraic. Whereas the former are characterized by a fixed geometric replacement rule, the latter are defined by a recurrence function in the complex plane. The classical method for visualizing algebraic fractals considers the sequence of complex numbers originated from each point in the complex plane. Thus, each original point is colored depending on whether its generated sequence escapes to infinity. The present work introduces a novel visualization method for algebraic fractals. This method colors each original point by taking into account the complex number with minimum modulus within its generated sequence. The advantages of the novel method are twofold: on the one hand, it preserves the fractal view that the classical method offers of the escape set boundary and, on the other hand, it additionally provides interesting visual details of the prisoner set (the complement of the escape set). The novel method is comparatively evaluated with other classical and non-classical visualization methods of fractals, giving rise to aesthetic views of prisoner sets.Publicación Extending cellular evolutionary algorithms with message passing(Springer, 2021-02-02) Fernández Galán, SeverinoCellular evolutionary algorithms (cEAs) use structured populations whose evolutionary cycle is governed by local interactions among individuals. This helps to prevent the premature convergence to local optima that usually takes place in panmictic populations. The present work extends cEAs by means of a message passing phase whose main effect is a more effective exploration of the search space. The mutated offspring that potentially replaces the original individual under cEAs is considered under message passing cellular evolutionary algorithms (MPcEAs) as a message sent from the original individual to itself. In MPcEAs, unlike in cEAs, a new message is sent from the original individual to each of its neighbors, representing a neighbor’s mutated offspring whose second parent is selected from the neighborhood of the original individual. Thus, every individual in the population ultimately receives one additional candidate for replacement from each of its neighbors rather than having a unique candidate. Experimental tests conducted in the domain of real function optimization for continuous search spaces show that, in general, MPcEAs significantly outperform cEAs in terms of effectiveness. Specifically, the best solution obtained through MPcEAs has an importantly improved fitness quality in comparison to that obtained by cEAs.