Artículos (Ciencias de la Computación e Inteligencia Artificial)
URI permanente para esta colecciónhttps://hdl.handle.net/11441/11302
Examinar
Envíos recientes
Artículo Análisis de la tasa de abandono en un Centro con varios Grados en Ingeniería Informática(2017) Ruiz Cortés, David; Gómez Rodríguez, Francisco de Asís; Ruiz Reina, José Luis; Universidad de Sevilla. Departamento de Lenguajes y Sistemas Informáticos; Universidad de Sevilla. Departamento de Arquitectura y Tecnología de Computadores; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia ArtificialEn este trabajo se muestra el análisis realizado del impacto que sobre la tasa de abandono tiene el cambio de estudios entre los tres Grados en Ingeniería Informática que se imparten en un Centro concreto. Dicho análisis ha sido llevado a cabo por el Equipo de Dirección del Centro a instancia de los informes realizados tras las visitas para la renovación de la acreditación de dichos títulos. Las principales conclusiones a las que hemos llegado son: i) el cambio de estudios entre Grados en Informática siempre tiene un efecto negativo sobre la tasa de abandono, oscilando este entre el 3% y el 20 %; ii) dicho cambio de estudios puede responder a cuestiones académicas en algunos casos, pero también se apuntan cuestiones económicas por el ahorro que puede llegar a suponer; iii) aproximadamente un tercio de nuestros estudiantes abandona los estudios en Ingeniería Informática; iv) la tasa de abandono a lo largo de los últimos 5 años se ha mantenido acorde con lo establecido en las memorias de verificación y conforme a la media nacional en la rama de conocimiento de Ingeniería y Arquitectura; v) los sistemas de indicadores definidos por los distintos sistemas de garantía de calidad de los Títulos en ocasiones no son homogéneos, lo que dificulta realizar cualquier tipo de análisis.Artículo Review of ensembles of multi-label classifiers: Models, experimental study and prospects(Elsevier, 2018-11) Moyano Murillo, José María; Gibaja, E.L.; Cios, K.J.; Ventura, S.; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia Artificial; Ministerio de Economía y Competitividad. España; Ministerio de Educación. EspañaThe great attention given by the scientific community to multi-label learning in recent years has led to the development of a large number of methods, many of them based on ensembles. A comparison of the state-of-theart in ensembles of multi-label classifiers over a wide set of 20 datasets have been carried out in this paper, evaluating their performance based on the characteristics of the datasets such as imbalance, dependence among labels and dimensionality. In each case, suggestions are given to choose the algorithm that fits best. Further, given the absence of taxonomies of ensembles of multi-label classifiers, a novel taxonomy for these methods is proposed.Artículo MLDA: A tool for analyzing multi-label datasets(Elsevier, 2017) Moyano Murillo, José María; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia ArtificialThe objective of this paper is to present MLDA, a tool for the exploration and analysis of multi-labeldatasets with both simple and multiple views. MLDA comprises a GUI and a Java API, providing the userwith a wide set of charts, metrics, methods for transforming and preprocessing data, as well as compari- son of several datasets. The paper introduces the main features of the framework, and introduces its usetoward some illustrative examples.Artículo KEEL 3.0: An Open Source Software for Multi-Stage Analysis in Data Mining(Atlantis Press, 2017) Triguero, Isabel; González, Sergio; Moyano Murillo, José María; García, Salvador; Alcalá Fernández, Jesús; Luengo, Julián; Fernández, Alberto; Jesús, María José del; Sánchez, Luciano; Herrera, Francisco; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia Artificial; Ministerio de Educación y Ciencia (MEC). España; Ministerio de Educación. EspañaThis paper introduces the 3rd major release of the KEEL Software. KEEL is an open source Java framework (GPLv3 license) that provides a number of modules to perform a wide variety of data mining tasks. It includes tools to perform data management, design of multiple kind of experiments, statistical analyses, etc. This framework also contains KEEL-dataset, a data repository for multiple learning tasks featuring data partitions and algorithms’ results over these problems. In this work, we describe the most recent components added to KEEL 3.0, including new modules for semi-supervised learning, multi-instance learning, imbalanced classification and subgroup discovery. In addition, a new interface in R has been incorporated to execute algorithms included in KEEL. These new features greatly improve the versatility of KEEL to deal with more modern data mining problems.Artículo Combining multi-label classifiers based on projections of the output space using Evolutionary algorithms(Elsevier, 2020) Moyano Murillo, José María; Gibaja, E.L.; Cios, K.J.; Ventura, S.; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia ArtificialThe multi-label classification task has gained a lot of attention in the last decade thanks to its good application to many real-world problems where each object could be attached to several labels simultaneously. Several approaches based on ensembles for multi-label classification have been proposed in the literature; however, the vast majority are based on randomly selecting the different aspects that make the ensemble diverse and they do not consider the characteristics of the data to build it. In this paper we propose an evolutionary method called Evolutionary AlGorithm for multi- Label Ensemble opTimization, EAGLET, for the selection of simple, accurate and diverse multi-label classifiers to build an ensemble considering the characteristics of the data, such as the relationship among labels and the imbalance degree of the labels. In order to model the relationships among labels, each classifier of the ensemble is focused on a small subset of the label space, resulting in models with a relative low computational complexity and lower imbalance in the output space. The resulting ensemble is generated incrementally given the population of multi-label classifiers, so the member that best fits to the ensemble generated so far, considering both predictive performance and diversity, is selected. The experimental study comparing EAGLET with state-of-the-art methods in multi-label classification over a wide set of sixteen datasets and five evaluation measures, demonstrated that EAGLET significantly outperformed standard MLC methods and obtained better and more consistent results than state-of-the-art multi-label ensembles.Artículo An evolutionary approach to build ensembles of multi-label classifiers(Elsevier, 2019-10) Moyano Murillo, José María; Gibaja, E.L.; Cios, K.J.; Ventura, S.; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia ArtificialIn recent years, the multi-label classification task has gained the attention of the scientific community given itsability to solve problems where each of the instances of the dataset may be associated with several class labels atthe same time instead of just one. The main problems to deal with in multi-label classification are the imbalance,the relationships among the labels, and the high complexity of the output space. A large number of methodsfor multi-label classification has been proposed, but although they aimed to deal with one or many of theseproblems, most of them did not take into account these characteristics of the data in their building phase. Inthis paper we present an evolutionary algorithm for automatic generation of ensembles of multi-label classifiersby tackling the three previously mentioned problems, called Evolutionary Multi-label Ensemble (EME). Eachmulti-label classifier is focused on a small subset of the labels, still considering the relationships among thembut avoiding the high complexity of the output space. Further, the algorithm automatically designs the ensembleevaluating both its predictive performance and the number of times that each label appears in the ensemble,so that in imbalanced datasets infrequent labels are not ignored. For this purpose, we also proposed a novelmutation operator that considers the relationship among labels, looking for individuals where the labels aremore related. EME was compared to other state-of-the-art algorithms for multi-label classification over a set offourteen multi-label datasets and using five evaluation measures. The experimental study was carried out in twoparts, first comparing EME to classic multi-label classification methods, and second comparing EME to otherensemble-based methods in multi-label classification. EME performed significantly better than the rest of classicmethods in three out of five evaluation measures. On the other hand, EME performed the best in one measure inthe second experiment and it was the only one that did not perform significantly worse than the control algorithmin any measure. These results showed that EME achieved a better and more consistent performance than the restof the state-of-the-art methods in MLC.Artículo On Conditional Axioms and Associated Inference Rules(MDPI, 2024-05-07) Borrego Díaz, Joaquín; Cordón Franco, Andrés; Lara Martín, Francisco Félix; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia Artificial; Universidad de Sevilla. TIC137: Logica, Computacion e Ingenieria del ConocimientoIn the present paper, we address the following general question in the framework of classical first-order logic. Assume that a certain mathematical principle can be formalized in a firstorder language by a set E of conditional formulas of the form α(v) → β(v). Given a base theory T, we can use the set of conditional formulas E to extend the base theory in two natural ways. Either we add to T each formula in E as a new axiom (thus obtaining a theory denoted by T + E) or we extend T by using the formulas in E as instances of an inference rule (thus obtaining a theory denoted by T + E–Rule). The theory T + E will be stronger than T + E–Rule, but how much stronger can T + E be? More specifically, is T + E conservative over T + E–Rule for theorems of some fixed syntactical complexity Γ? Under very general assumptions on the set of conditional formulas E, we obtain two main conservation results in this regard. Firstly, if the formulas in E have low syntactical complexity with respect to some prescribed class of formulas Π and in the applications of E–Rule side formulas from the class Π and can be eliminated (in a certain precise sense), then T + E is ∀B(Π)- conservative over T + E–Rule. Secondly, if, in addition, E is a finite set with m conditional sentences, then nested applications of E–Rule of a depth at most of m suffice to obtain ∀B(Π) conservativity. These conservation results between axioms and inference rules extend well-known conservation theorems for fragments of first-order arithmetics to a general, purely logical framework.Artículo A novel solution for GCP based on an OLMS membrane algorithm with dynamic operators(Springer, 2020) Andreu Guzmán, José A.; Valencia Cabrera, Luis; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia Artificial; Universidad de Sevilla. Departamento de Lenguajes y Sistemas InformáticosGraph coloring problem (GCP) is an NP-complete combinatorial optimization problem. Its computational complexity motivated many efforts to get approximate solutions through different meta-heuristics, such as several variants of evolutionary algorithms. On the other hand, membrane algorithms have appeared as alternative hybrid techniques merging together the structure and operators of membrane systems, along with the capabilities of optimization algorithms inside each membrane. This paper explores the ability of a new variants of one-level membrane systems using a recent variant of evolutionary algorithm dynamically using different genetic operators depending on the best fitness found. The experimental results presented show that this new algorithm, called DOGAPS, outperforms the dynamic evolutionary algorithm, with the extra value provided by the membrane system. Additionally, the role of some parameters involved in our algorithm are analyzed, including the number of membranes, iterations per membrane or mutation rate.Artículo A protocol for solutions to DP-complete problems through tissue membrane systems(MDPI, 2023-06-21) Orellana Martín, David; Ramírez de Arellano Marrero, Antonio; Andreu Guzmán, José A.; Romero Jiménez, Álvaro; Pérez Jiménez, Mario de Jesús; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia Artificial; Universidad de Sevilla. Departamento de Lenguajes y Sistemas InformáticosConsidering a class R comprising recognizer membrane systems with the capability of providing polynomial-time and uniform solutions for NP-complete problems (referred to as a “presumably efficient” class), the corresponding polynomial-time complexity class PMCR encompasses both the NP and co NP classes. Specifically, when R represents the class of recognizer presumably efficient cell-like P systems that incorporate object evolution rules, communication rules, and dissolution rules, PMCR includes both the DP and co DP classes. Here, DP signifies the class of languages that can be expressed as the difference between any two languages in NP (it is worth noting that NP DP and co NP co DP). As DP-complete problems are believed to be more complex than NP-complete problems, they serve as promising candidates for studying the P vs NP problem. This outcome has previously been established within the realm of recognizer P systems with active membranes. In this paper, we extend this result to encompass any class R of presumably efficient recognizer tissue-like membrane systems by presenting a detailed protocol for transforming solutions of NP-complete problems into solutions of DP-complete problems.Artículo Random walk simulation by population dynamics P systems(Springer, 2024) Orellana Martín, David; Andreu Guzmán, José A.; Graciani Díaz, Carmen; Riscos Núñez, Agustín; Pérez Jiménez, Mario de Jesús; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia Artificial; Universidad de Sevilla. Departamento de Lenguajes y Sistemas InformáticosPDP systems have been widely used for real-life applications, such as systems biology, ecosystems, physics or economy, among others. Complex systems related with these areas are simulated in the framework of Membrane Computing using objects and membranes that can represent entities or places in the real-life process. In physics, the study of a particle in different fluids, depending on their composition, is really interesting for several applications. A first approximation to this field is to think that particles move randomly in the available space, without any force that constrains their movements. This behavior is known as random walk, and it is used not only in physics but in economics, genetics, and ecology among other areas. In this paper, we introduce generic PDP systems for simulating the behavior of particles, both for one-dimensional spaces and for two-dimensional spaces, using different simulators to analyze the computational resources consumed.Artículo SpaceRL-KG: Searching paths automatically combining embedding-based rewards with Reinforcement Learning in Knowledge Graphs(Pergamon-Elsevier, 2024) Bermudo Bayo, Miguel; Ayala Hernández, Daniel; Hernández Salmerón, Inmaculada Concepción; Ruiz Cortés, David; Toro Bonilla, Miguel; Universidad de Sevilla. Departamento de Lenguajes y Sistemas Informáticos; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia Artificial; Ministerio de Ciencia, Innovación y Universidades (MICINN). EspañaKnowledge Graph Completion seeks to find missing elements in a Knowledge Graph, usually edges representing some relation between two concepts. One possible way to do this is to find paths between two nodes that indicate the presence of a missing edge. This can be achieved through Reinforcement Learning, by training an agent that learns how to navigate through the graph, starting at a node with a missing edge and identifying what edge among the available ones at each step is more promising in order to reach the target of the missing edge. While some approaches have been proposed to this effect, their reward functions only take into account whether the target node was reached or not, and only apply a single Reinforcement Learning algorithm. In this regard, we present a new family of reward functions based on node embeddings and structural distance, leveraging additional information related to semantic similarity and removing the need to reach the target node to obtain a measure of the benefits of an action. Our experimental results show that these functions, as well as the novel use of more modern Reinforcement Learning techniques, are able to obtain better results than the existing strategies in the literature.Artículo Silence — A web framework for an agile generation of RESTful APIs(Elsevier, 2022) Borrego, Agustín; Bermudo Bayo, Miguel; Ayala Hernández, Daniel; Hernández Salmerón, Inmaculada Concepción; Ruiz Cortés, David; Universidad de Sevilla. Departamento de Lenguajes y Sistemas Informáticos; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia Artificial; Ministerio de Ciencia, Innovación y Universidades (MICINN). EspañaSilence is a web framework made in Python that allows users to automatically generate a RESTcompliant API and a set of test skeletons given a relational database. Silence provides the user with a number of console commands to create, manage and run projects. In contrast with other existing web frameworks, Silence makes it quick and straightforward to publish existing data available online, and it is flexible enough to be used for both research and general-purpose applications. Silence is published as a Python package, making it easy to install and operate, and does not require knowledge of any specific programming language to be used.Artículo Semi-honest subrecursive degrees and the collection rule in arithmetic(Springer, 2024) Cordón Franco, Andrés; Lara Martín, Francisco Félix; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia ArtificialBy a result of L.D. Beklemishev, the hierarchy of nested applications of the Σ1-collection rule over any Π2-axiomatizable base theory extending Elementary Arithmetic collapses to its first level. We prove that this result cannot in general be extended to base theories of arbitrary quantifier complexity. In fact, given any recursively enumerable set of true Π2-sentences, S, we construct a sound (Σ2∨Π2)-axiomatized theory T extending S such that the hierarchy of nested applications of the Σ1-collection rule over T is proper. Our construction uses some results on subrecursive degree theory obtained by L. Kristiansen.Artículo Lipschitz and Wadge binary games in second order arithmetic(Elsevier Science, 2023) Cordón Franco, Andrés; Lara Martín, Francisco Félix; Loureiro, Manuel J.S.; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia Artificial; Ministerio de Economía, Industria y Competitividad, EspañaWe present a detailed formalization of Lipschitz and Wadge games in the context of second order arithmetic and we investigate the logical strength of Lipschitz and Wadge determinacy, and the tightly related Semi-Linear Ordering principle, for the first levels of the Hausdorff difference hierarchy in the Cantor space. As a result, we obtain characterizations of WKL0and ACA0in terms of these determinacy principles.
Artículo Reservoir computing models based on spiking neural P systems for time series classification(PERGAMON-ELSEVIER SCIENCE LTD, 2024) Peng, Hong; Xiong, Xin; Wu, Min; Wang, Jun; Yang, Qiang; Orellana Martín, David; Pérez Jiménez, Mario de Jesús; ; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia ArtificialNonlinear spiking neural P (NSNP) systems are neural-like membrane computing models with nonlinear spiking mechanisms. Because of this nonlinear spiking mechanism, NSNP systems can show rich nonlinear dynamics. Reservoir computing (RC) is a novel recurrent neural network (RNN) and can overcome some shortcomings of traditional RNNs. Based on NSNP systems, we developed two RC variants for time series classification, RC-SNP and RC-RMS-SNP, which are without and integrated with reservoir model space (RMS), respectively. The two RC variants use NSNP systems as the reservoirs and can be easily implemented in the RC framework. The proposed two RC variants were evaluated on 17 benchmark time series classification datasets and compared with 16 state-of-the-art or baseline classification models. The comparison results demonstrate the effectiveness of the proposed two RC variants for time series classification tasks.Artículo Bridges Between Spiking Neural Membrane Systems and Virus Machines(2024) Ramírez de Arellano, Antonio; Orellana Martín, David; Pérez Jiménez, Mario de Jesús; ; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia ArtificialSpiking Neural P Systems (SNP) are well-established computing models that take inspiration from spikes between biological neurons; these models have been widely used for both theoretical studies and practical applications. Virus machines (VMs) are an emerging computing paradigm inspired by viral transmission and replication. In this work, a novel extension of VMs inspired by SNPs is presented, called Virus Machines with Host Excitation (VMHEs). In addition, the universality and explicit results between SNPs and VMHEs are compared in both generating and computing mode. The VMHEs defined in this work are shown to be more efficient than SNPs, requiring fewer memory units (hosts in VMHEs and neurons in SNPs) in several tasks, such as a universal machine, which was constructed with 18 hosts less than the 84 neurons in SNPs, and less than other spiking models discussed in the workArtículo Sparse Spiking Neural-Like Membrane Systems on Graphics Processing Units(World Scientific, 2024) Hernandez Tello, Javier; Martínez del Amor, Miguel Ángel; Orellana Martín, David; Cabarle, Francis George C.; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia Artificial; Universidad de Sevilla. Departamento de Lenguajes y Sistemas Informáticos; Ministerio de Ciencia, Innovación y Universidades (MICINN). España; Junta de Andalucía; Grupo de Investigación en Computación Natural TIC193; I3US; SCORE LabThe parallel simulation of Spiking Neural P systems is mainly based on a matrix representation, where the graph inherent to the neural model is encoded in an adjacency matrix. The simulation algorithm is based on a matrix-vector multiplication, which is an operation efficiently implemented on parallel devices. However, when the graph of a Spiking Neural P system is not fully connected, the adjacency matrix is sparse and hence, lots of computing resources are wasted in both time and memory domains. For this reason, two compression methods for the matrix representation were proposed in a previous work, but they were not implemented nor parallelized on a simulator. In this paper, they are implemented and parallelized on GPUs as part of a new Spiking Neural P system with delays simulator. Extensive experiments are conducted on high-end GPUs (RTX2080 and A100 80GB), and it is concluded that they outperform other solutions based on state-of-the-art GPU libraries when simulating Spiking Neural P systems.Informe WebSnapse Tutorial: A Hands-On Approach for Web and Visual Simulations of Spiking Neural P Systems(2023) Ren Tristan A. De La Cruz; Ko, Daryll; Cabarle, Francis George C.; Tristan De La Cruz, Ren; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia Artificial; Grupo de Investigación en Computación Natural TIC193Spiking neural P (SN P) systems were introduced as a special class of P systems. Traditional P systems involve nested membranes through which objects can be transported. This makes natural the idea of distributed or parallel computing, as object transportation may happen simultaneously across different membranes [2, 3]. SN P systems approach distributed computing using different constructs: neurons and synapses. Interactions between these constructs mimic the operation of the human brain: neurons send signals (or spikes) to other neurons via synapses. Investigations on SN P systems have revealed that their computing power is equiv- alent to that of a Turing machine and, at the cost of space, these systems are able to solve NP-hard problems in polynomial time. Understanding theoretical concepts like SN P systems can be made easier by going through examples and manual simulations of such systems. A seminal exam- ple of this is Paun's tutorial. Examples like the even natural number generator in the tutorial help the reader form an adequate mental model of the central con- cepts. The present work follows in a similar fashion, applied to the simulator Web- Snapse. The origin of WebSnapse comes from the Snapse simulator with the following raison d'être: an easy to use and visual simulator for learning about SN P systems. The simulator WebSnapse v1 (that is, version 1) improves on this reason for being by allowing users to access the software by using only their web browser, including more animations and features. Unlike Snapse which requires specific Spiking neural P (SN P) systems were introduced as a special class of P systems. Traditional P systems involve nested membranes through which objects can be transported. This makes natural the idea of distributed or parallel computing, as object transportation may happen simultaneously across different membranes .Informe Analyses and Implementation of Homogenisation Algorithms for Spiking Neural P Systems in the WebSnapse Tool(2024) Kristian Llanto, Tim; Amador, Joshua; Cabarle, Francis George C.; Tristan De La Cruz, Ren; Ko, Daryll; Universidad de Sevilla. Departamento de Ciencias de la Computación e Inteligencia Artificial; Grupo de Investigación en Computación Natural TIC193An area of computer science called Membrane Computing (MC) was introduced by Gheorghe P˘aun in. It is a form of computing that takes inspiration from how living cells work. Its system, called the P system, mimics the structure of a cell and how it communicates with its neighbor cells. This is different from the traditional way of computing since cells can communicate in a distributed and parallel manner.Artículo Applicability of Deep Learning to Dynamically Identify the Different Organs of the Pelvic Floor in the Midsagittal Plane(Springer, 2024-05-01) García Mejido, José Antonio; Solís Martín, David; Martín Morán, Marina; Fernández-Conde, Cristina; Fernández Palacín, Fernando; Sáinz Bueno, José Antonio; Universidad de Sevilla. Departamento de CirugíaIntroduction and Hypothesis The objective was to create and validate the usefulness of a convolutional neural network (CNN) for identifying different organs of the pelvic floor in the midsagittal plane via dynamic ultrasound. Methods This observational and prospective study included 110 patients. Transperineal ultrasound scans were performed by an expert sonographer of the pelvic floor. A video of each patient was made that captured the midsagittal plane of the pelvic floor at rest and the change in the pelvic structures during the Valsalva maneuver. After saving the captured videos, we manually labeled the different organs in each video. Three different architectures were tested—UNet, FPN, and LinkNet—to determine which CNN model best recognized anatomical structures. The best model was trained with the 86 cases for the number of epochs determined by the stop criterion via cross-validation. The Dice Similarity Index (DSI) was used for CNN validation. Results Eighty-six patients were included to train the CNN and 24 to test the CNN. After applying the trained CNN to the 24 test videos, we did not observe any failed segmentation. In fact, we obtained a DSI of 0.79 (95% CI: 0.73 – 0.82) as the median of the 24 test videos. When we studied the organs independently, we observed differences in the DSI of each organ. The poorest DSIs were obtained in the bladder (0.71 [95% CI: 0.70 – 0.73]) and uterus (0.70 [95% CI: 0.68 – 0.74]), whereas the highest DSIs were obtained in the anus (0.81 [95% CI: 0.80 – 0.86]) and levator ani muscle (0.83 [95% CI: 0.82 – 0.83]). Conclusions Our results show that it is possible to apply deep learning using a trained CNN to identify different pelvic floor organs in the midsagittal plane via dynamic ultrasound.