Divina, Federico

Profesor/a Titular de Universidad
Profile Picture
First Name
Last Name
Universidad Pablo de Olavide
Deporte e Informática
Research Center
Lenguajes y Sistemas Informáticos
Research Group
PhD programs
UPO investigaORCIDScopus Author IDWeb of Science ResearcherIDDialnet IDGoogle Scholar ID

Search Results

Now showing 1 - 6 of 6
  • Publication
    El trabajo autónomo como herramienta didáctica
    (2012) Barranco, Carlos D.; Rodríguez Baena, Domingo Savio; Divina, Federico; Aguilar-Ruiz, Jesús S.
    El objetivo de este artículo es el de presentar tres casos prácticos, en el ámbito de tres asignaturas de la Titulación en Ingeniería Técnica en Informática de Gestión de la Universidad Pablo de Olavide, en los que el trabajo autónomo del alumno ha sido la herramienta utilizada para solventar la problemática provocada por la reducción de horas de clases que deriva de la implantación del EEES que se agravaba más en la modalidad semipresencial de la titulación, modalidad en la que los alumnos, normalmente trabajadores en activo, ven reducidas las horas de presencialidad requerida un 50% para facilitar la compaginación de estudios y actividad laboral. Los resultados obtenidos en términos de tasas de éxito y porcentajes de abandono muestran unamejora de los resultados obtenidos por las asignaturas, corroborando la utilidad de un trabajo autónomo bien planteado.
  • Publication
    A multivariate approach to the symmetrical uncertainty measure: Application to feature selection problem
    (Elsevier, 2019) Sosa Cabrera, Gustavo; García Torres, Miguel; Gómez Guerrero, Santiago; E. Schaerer, Christian; Divina, Federico
    In this work we propose an extension of the Symmetrical Uncertainty (SU) measure in order to address the multivariate case, simultaneously acquiring the capability to detect possible correlations and interactions among features. This generalization, denoted Multivariate Symmetrical Uncertainty (MSU), is based on the concepts of Total Correlation (TC) and Mutual Information (MI) extended to the multivariate case. The generalized measure accounts for the total amount of dependency within a set of variables as a single monolithic quantity. Multivariate measures are usually biased due to several factors. To overcome this problem, a mathematical expression is proposed, based on the cardinality of all features, which can be used to calculate the number of samples needed to estimate the MSU without bias at a pre-specified significance level. Theoretical and experimental results on synthetic data show that the proposed sample size expression properly controls the bias. In addition, when the MSU is applied to feature selection on synthetic and real-world data, it has the advantage of adequately capturing linear and nonlinear correlations and interactions, and it can therefore be used as a new feature subset evaluation method.
  • Publication
    Identifying livestock behavior patterns based on accelerometer dataset
    (Elsevier, 2020) Rodríguez Baena, Domingo Savio; Gómez-Vela, Francisco Antonio; García Torres, Miguel; Divina, Federico; Barranco, Carlos D.; Díaz-Díaz, Norberto; Jiménez, Manuel; Montalvo, Gema
    In large livestock farming it would be beneficial to be able to automatically detect behaviors in animals. In fact, this would allow to estimate the health status of individuals, providing valuable insight to stock raisers. Traditionally this process has been carried out manually, relying only on the experience of the breeders. Such an approach is effective for a small number of individuals. However, in large breeding farms this may not represent the best approach, since, in this way, not all the animals can be effectively monitored all the time. Moreover, the traditional approach heavily rely on human experience, which cannot be always taken for granted. To this aim, in this paper, we propose a new method for automatically detecting activity and inactivity time periods of animals, as a behavior indicator of livestock. In order to do this, we collected data with sensors located in the body of the animals to be analyzed. In particular, the reliability of the method was tested with data collected on Iberian pigs and calves. Results confirm that the proposed method can help breeders in detecting activity and inactivity periods for large livestock farming.
  • Publication
    A multi-GPU biclustering algorithm for binary datasets
    (Elsevier, 2021) López Fernández, Aurelio; Rodríguez Baena, Domingo Savio; Gómez-Vela, Francisco Antonio; Divina, Federico; García Torres, Miguel
    Graphics Processing Units technology (GPU) and CUDA architecture are one of the most used options to adapt machine learning techniques to the huge amounts of complex data that are currently generated. Biclustering techniques are useful for discovering local patterns in datasets. Those of them that have been implemented to use GPU resources in parallel have improved their computational performance. However, this fact does not guarantee that they can successfully process large datasets. There are some important issues that must be taken into account, like the data transfers between CPU and GPU memory or the balanced distribution of workload between the GPU resources. In this paper, a GPU version of one of the fastest biclustering solutions, BiBit, is presented. This implementation, named gBiBit, has been designed to take full advantage of the computational resources offered by GPU devices. Either using a single GPU device or in its multi-GPU mode, gBiBit is able to process large binary datasets. The experimental results have shown that gBiBit improves the computational performance of BiBit, a CPU parallel version and an early GPU version, called ParBiBit and CUBiBit, respectively. gBiBit source code is available at
  • Publication
    Technical Analysis Strategy Optimization using a Machine Learning Approach in Stock Market Indices
    (Elsevier, 2021) Ayala, Jordan; García Torres, Miguel; Vázquez Noguera, José Luis; Gómez-Vela, Francisco Antonio; Divina, Federico
    Within the area of stock market prediction, forecasting price values or movements is one of the most challenging issue. Because of this, the use of machine learning techniques in combination with technical analysis indicators is receiving more and more attention. In order to tackle this problem, in this paper we propose a hybrid approach to generate trading signals. To do so, our proposal consists of applying a technical indicator combined with a machine learning approach in order to produce a trading decision. The novelty of this approach lies in the simplicity and effectiveness of the hybrid rules as well as its possible extension to other technical indicators. In order to select the most suitable machine learning technique, we tested the performances of Linear Model (LM), Artificial Neural Network (ANN), Random Forests (RF) and Support Vector Regression (SVR).As technical strategies for trading, the Triple Exponential Moving Average (TEMA) and Moving Average Convergence/Divergence (MACD) were considered. We tested the resulting technique on daily trading data from three major indices: Ibex35 (IBEX), DAX and Dow Jones Industrial (DJI). Results achieved show that the addition of machine learning techniques to technical analysis strategies improves the trading signals and the competitiveness of the proposed trading rules.
  • Publication
    Evolutionary feature selection on high dimensional data using a search space reduction approach
    (Elsevier, 2023) García Torres, Miguel; Ruiz, Roberto; Divina, Federico
    Feature selection is becoming more and more a challenging task due to the increase of the dimensionality of the data. The complexity of the interactions among features and the size of the search space make it unfeasible to find the optimal subset of features. In order to reduce the search space, feature grouping has arisen as an approach that allows to cluster feature according to the shared information about the class. On the other hand, metaheuristic algorithms have proven to achieve sub-optimal solutions within a reasonable time. In this work we propose a Scatter Search (SS) strategy that uses feature grouping to generate an initial population comprised of diverse and high quality solutions. Solutions are then evolved by applying random mechanisms in combination with the feature group structure, with the objective of maintaining during the search a population of good and, at the same time, as diverse as possible solutions. Not only does the proposed strategy provide the best subset of features found but it also reduces the redundancy structure of the data. We test the strategy on high dimensional data from biomedical and text-mining domains. The results are compared with those obtained by other adaptations of SS and other popular strategies. Results show that the proposed strategy can find, on average, the smallest subsets of features without degrading the performance of the classifier