Actas y comunicaciones de congresos

URI permanente para esta colección

hdl:123456789/22

Examinar

Mostrando 1 - 20 de 51

A data driven approach for person name disambiguation in web search results
(2014-08-23) Víctor Fresno, Víctor; Montalvo, Soto; Delgado Muñoz, Agustín Daniel; Martínez Unanue, Raquel
This paper presents an unsupervised approach for the task of clustering the results of a search engine when the query is a person name shared by different individuals. We propose an algorithm that calculates the number of clusters and establishes the groups of web pages according to the different individuals without the need to any training data or predefined thresholds, as the successful state of the art systems do. In addition, most of those systems do not deal with social media web pages and their performance could fail in a real scenario. In this paper we also propose a heuristic method for the treatment of social networking profiles. Our approach is compared with four gold standard collections for this task obtaining really competitive results, comparable to those obtained by some approaches with supervision.
A Scorewriter Application using Electrooculography-based Human-Computer Interface
(IEEE, 2022) Pérez–Roa, Enrique M.; Mañoso Hierro, María Carolina; Pérez de Madrid y Pablo, Ángel; Romero Hortelano, Miguel
At present, many projects are being developed with human-computer interfaces in different areas but few are related to music. In this work we present a scorewriter application that uses electrooculography as input interface. For one side, the hardware used to record the electrooculogram consists mainly of a low-cost Arduino based microcontroller board that will receive the signal from the electrodes, collect it and send it via USB to the computer. On the other hand, we use free software to implement the application running on the computer. This application is in charge of processing, classifying (using a neural network) and translating the signal into commands to finally build the song and play it. The modularity of the application allows it to be easily modified for other tasks using the same interface. Due to the nature of the application it is very suitable for entertainment. Furthermore, due to the characteristics of its interface it is also suitable for people with reduced mobility who want to easily perform simple music composition tasks.
A simple measure to assess non-response
(2011-06-19) Peñas Padilla, Anselmo; Rodrigo Yuste, Álvaro
There are several tasks where is preferable not responding than responding incorrectly. This idea is not new, but despite several previous attempts there isn’t a commonly accepted measure to assess non-response. We study here an extension of accuracy measure with this feature and a very easy to understand interpretation. The measure proposed (c@1) has a good balance of discrimination power, stability and sensitivity properties. We show also how this measure is able to reward systems that maintain the same number of correct answers and at the same time decrease the number of incorrect ones, by leaving some questions unanswered. This measure is well suited for tasks such as Reading Comprehension tests, where multiple choices per question are given, but only one is correct.
Analyzing information retrieval methods to recover broken web links
(2011-06-19) Martínez Romo, Juan; Araujo Serna, M. Lourdes
In this work we compare different techniques to automatically find candidate web pages to substitute broken links. We extract information from the anchor text, the content of the page containing the link, and the cache page in some digital library.The selected information is processed and submitted to a search engine. We have compared different information retrievalmethods for both, the selection of terms used to construct the queries submitted to the search engine, and the ranking of the candidate pages that it provides, in order to help the user to find the best replacement. In particular, we have used term frequencies, and a language model approach for the selection of terms; and cooccurrence measures and a language model approach for ranking the final results. To test the different methods, we have also defined a methodology which does not require the user judgments, what increases the objectivity of the results.
Aplicaciónde técnica de inteligencia artificial y tratamiento de señales en fusión
(CEA-IFAC, 2005, 2005-01-01) Farias Castro, Gonzalo Alberto; Santos, Matilde
Assessing Feature Selection Techniques for AI-based IoT Network Intrusion Detection
(Springer, 2025-06) García Merino, José Carlos; Tobarra Abad, María de los Llanos; Robles Gómez, Antonio; Pastor Vargas, Rafael; Vidal Balboa, Pedro; Dionisio Rocha, André; Jardim Gonçalves, Ricardo
The widespread adoption of Internet of Things (IoT) technology in rural areas has led to qualitative leaps in fields such as agriculture, livestock farming, and transportation, giving rise to the concept of Smart Rural. However, Smart Rural IoT ecosystems are often vulnerable to cyberattacks. Although Artificial Intelligence (AI) based intrusion detection systems offer an effective solution to protect these environments, IoT devices are typically constrained in terms of memory and computation capabilities, making it essential to optimise the computational burden of AI models. This work explores different feature selection techniques to develop compact and fast Random Forest models for anomaly detection in IoT environments. The obtained results demonstrate that appropriate feature selection can reduce model size and inference time by at least 45% and 8%, respectively, without compromising predictive performance.
Automated IoT vulnerability classification using Deep Learning
(2025-07) Sernández Iglesias, Daniel; Enrique Fernández Morales,; Garcia Merino, Jose Carlos; Tobarra Abad, María de los Llanos; Pastor Vargas, Rafael; Robles Gómez, Antonio; Sarraipa, Joao
Technological advancements in the development of low-power chips have enabled everyday objects to connect to the Internet, giving rise to the concept known as the Internet of Things (IoT). It is currently estimated that there are approximately 16 billion IoT connections worldwide, a figure expected to double by 2030. However, this rapid growth of the IoT ecosystem has introduced new vulnerabilities that could be exploited by malicious actors. Since many IoT devices handle personal and sensitive information, threats to these devices can have severe consequences. Moreover, a series of cybersecurity incidents could undermine public trust in IoT technology, potentially delaying its widespread adoption across various sectors.Common Vulnerabilities and Exposures records (also known by their acronym as CVEs) is a public cataloging system designed to identify and list known security vulnerabilities in software and hardware products. This system is developed and maintained by MITRE with the support of the cybersecurity community and sponsored by the U.S. Department of Homeland Security (DHS) through the Cybersecurity and Infrastructure Security Agency (CISA). CVE provides a reference database that enables security researchers, manufacturers, and organizational security managers to more effectively identify and address security issues.In our study, we have focused on CVEs exclusively oriented towards IoT systems, with the aim of analyzing the main vulnerabilities detected from 2010 to nowadays as a basis for detecting the main attack vectors in IoT systems. As part of this effort we have created the following dataset. CVEs records include various metrics such as: - Common Weakness Enumeration (CWE), mainly focused on technical classification of vulnerabilities. - Common Vulnerability Scoring System (CVSS), which reports about different metrics such as the attack vector, the severity of the vulnerability or the impact level of the exploitation of the vulnerability. This is one of the most informative metric. - Stakeholder-Specific Vulnerability Categorization (SSVC), oriented towards help cybersecurity team to handle properly the vulnerability. These metrics allow security teams on the one hand to prioritize, such vulnerabilities within their security program, evaluating efforts to mitigate them. But according to our analysis of our dataset, around the 14% of CVEs records do not contain any metric. Around the 83% of CVEs registries contain CWE metric (an ID or its textual description). This metric, as it is explained before, only reports about the type of vulnerability from a technic point of view. Only the 10% of CVEs registries contain SSVC metrics. And CVSS, in its different versions, appears only in the 40% of the studied CVEs registries. Additionally, most of studied records includes metrics a retrospectively, several weeks or months later the vulnerability is disclosed. Thus, cybersecurity teams must trust their previous knowledge in order to distinguish which vulnerabilities are relevant and which not.To tackled this situation, our proposal is focused in the application of Deep Learning techniques in order to classify the severity of CVE records from its textual description. Textual description is a mandatory field that is present in all CVEs records. To achieve this objective, we trained the BiLSTM algorithm using the CVE records with CVSS metrics and its description field; and performed a comparative study of different hyperparameter configurations to find the optimal configuration. The metrics for model evaluation that have been studied are accuracy, loss and F1-score.
Circuit Testing Based on Fuzzy Sampling with BDD Bases
(University of Hawaiʻi at Mānoa, 2023) Pinilla, Elena; Fernández Amoros, David José; Heradio Gil, Rubén
Fuzzy testing of integrated circuits is an established technique. Current approaches generate an approximately uniform random sample from a translation of the circuit to Boolean logic. These approaches have serious scalability issues, which become more pressing with the ever-increasing size of circuits. We propose using a base of binary decision diagrams to sample the translations as a soft computing approach. Uniformity is guaranteed by design and scalability is greatly improved. We test our approach against five other state-of-the-art tools and find our tool to outperform all of them, both in terms of performance and scalability.
Composición fotográfica mediante el uso de un dron
(Comité Español de Automática, 2024-07-15) Sánchez García, Juan Miguel; Sánchez Moreno, José; Moreno Salinas, David
La composición fotográfica, conocida como mosaicos, es crucial en aplicaciones donde no es posible capturar toda la extensión de grandes superficies en una sola toma. Por ende, se requiere fotografiar secciones más pequeñas para luego componerlas y lograr una reproducción lo más precisa posible de la realidad. En este trabajo se presenta el resultado de aplicar los principios de las distintas etapas necesarias para crear un mosaico, complementado con el uso de un dron para la captura de las imágenes. La creación del mosaico implica técnicas avanzadas de procesamiento de imágenes que facilitan la detección de características, la transformación geométrica y la alineación de píxeles. Sin embargo, la experimentación con diferentes algoritmos ha revelado que no siempre es viable encontrar una transformación geométrica que produzca un mosaico de calidad, especialmente cuando las características de la fotografía no son óptimas, lo cual puede ser atribuible, en parte, a la resolución de los dispositivos fotográficos utilizados.
Correlation-Aware Averaging for Federated Learning in Remote Sensing Data Classification
(IEEE, 2024) Moreno Álvarez, Sergio; han, lirong; Paoletti, Mercedes Eugenia; Haut, Juan Mario; https://orcid.org/0000-0002-8613-7037; https://orcid.org/0000-0003-1030-3729; https://orcid.org/0000-0001-6701-961X
The increasing volume of remote sensing (RS) data offers substantial benefits for the extraction and interpretation of features from these scenes. Indeed, the detection of distinguishing features among captured materials and objects is crucial for classification purposes, such as in environmental monitoring applications. In these algorithms, the classes characterized by lower correlation often exhibit more distinct and discernible features, facilitating their differentiation in a straightforward manner. Nevertheless, the rise of Big Data provides a wide range of data acquired through multiple decentralized devices, where its susceptibility to be shared among various users or clients presents challenges in safeguarding privacy. Meanwhile, global features for similar classes are required to be learned for generalization purposes in the classification process. To address this, federated learning (FL) emerges as a privacy efficient decentralized solution. Firstly, in such scenarios, proprietary data is held by individual clients participating in the training of a global model. Secondly, clients may encounter challenges in identifying features that are more distinguishable within the data distributions of other clients. In this study, in order to handle these challenges, a novel methodology is proposed that considers the least correlated classes (LCCs) included in each client data distribution. This strategy exploits the distinctive features between classes, thereby enhancing performance and generalization ability in a secure and private environment.
Dataset Generation and Study of Deepfake Techniques
(Springer, 2023) Falcón López, Sergio Adrián; Robles Gómez, Antonio; Tobarra Abad, María de los Llanos; Pastor Vargas, Rafael
The consumption of multimedia content on the Internet has nowadays been expanded exponentially. These trends have contributed to fake news can become a very high influence in the current society. The latest techniques to influence the spread of digital false information are based on methods of generating images and videos, known as Deepfakes. This way, our research work analyzes the most widely used Deepfake content generation methods, as well as explore different conventional and advanced tools for Deepfake detection. A specific dataset has also been built that includes both fake and real multimedia contents. This dataset will allow us to verify whether the used image and video forgery detection techniques can detect manipulated multimedia content.
Deep Attention-Driven HSI Scene Classification Based on Inverted Dot-Product
(Institute of Electrical and Electronics Engineers Inc., 2022) Paoletti, Mercedes Eugenia; Tao, Xuanwen; han, lirong; Wu, Zhaoyue; Moreno Álvarez, Sergio; Haut, Juan M.; https://orcid.org/0000-0003-1030-3729; https://orcid.org/0000-0003-1093-0079; https://orcid.org/0000-0002-8613-7037; https://orcid.org/0000-0002-6797-2440; https://orcid.org/0000-0001-6701-961X
Capsule networks have been a breakthrough in the field of automatic image analysis, opening a new frontier in the art for image classification. Nevertheless, these models were initially designed for RGB images and naively applying these techniques to remote sensing hyperspectral images (HSI) may lead to sub-optimal behaviour, blowing up the number of parameters needed to train the model or not correctly modeling the spectral relations between the different layers of the scene. To overcome this drawback, this work implements a new capsule-based architecture with attention mechanism to improve the HSI data processing. The attention mechanism is applied during the concurrent iterative routing procedure through an inverted dot-product attention
Deep Robust Hashing Using Self-Distillation for Remote Sensing Image Retrieval
(IEEE, 2024) han,lirong; Paoletti, Mercedes Eugenia; Moreno Álvarez, Sergio; Haut, Juan Mario; Plaza, Antonio; https://orcid.org/0000-0002-8613-7037; https://orcid.org/0000-0003-1030-3729; https://orcid.org/0000-0001-6701-961X; https://orcid.org/0000-0002-9613-1659
This paper presents a novel self-distillation based deep robust hash for fast remote sensing (RS) image retrieval. Specifically, there are two primary processes in our proposed model: teacher learning (TL) and student learning (SL). Two transformed samples are produced from one sample image through nuanced and signalized transformations, respectively. Transformed samples are fed into both the TL and the SL flows. To reduce discrepancies in the processed samples and guarantee a consistent hash code, the parameters are shared by the two modules during the training stage. Then, a resilient module is employed to enhance the image features in order to ensure more dependable hash code production. Lastly, a three-component loss function is developed to train the entire model. Comprehensive experiments are conducted on two common RS datasets: UCMerced and AID. The experimental results validate that the proposed method has competitive performance against other RS image hashing methods.
Detection of Cerebral Ischaemia using Transfer Learning Techniques
(IEEE) Antón Munárriz, Cristina; Haut, Juan M.; Paoletti, Mercedes E.; Benítez Andrades, José Alberto; Pastor Vargas, Rafael; Robles Gómez, Antonio
Cerebrovascular accident (CVA) or stroke is one of the main causes of mortality and morbidity today, causing permanent disabilities. Its early detection helps reduce its effects and its mortality: time is brain. Currently, non-contrast computed tomography (NCCT) continues to be the first-line diagnostic method in stroke emergencies because it is a fast, available, and cost-effective technique that makes it possible to rule out haemorrhage and focus attention on the ischemic origin, that is, due to obstruction to arterial flow. NCCT are quantified using a scoring system called ASPECTS (Alberta Stroke Program Early Computed Tomography Score) according to the affected brain structures. This paper aims to detect in an initial phase those CTs of patients with stroke symptoms that present early alterations in CT density using a binary classifier of CTs without and with stroke, to alert the doctor of their existence. For this, several well-known neural network architectures are implemented in the ImageNet challenges (VGG, NasNet, ResNet and DenseNet), with 3D images, covering the entire brain volume. The training results of these networks are exposed, in which different parameters are tested to obtain maximum performance, which is achieved with a DenseNet3D network that achieves an accuracy of 98% in the training set and 95% in the test set
Determinación de parámetros de la transfomada Wavelet para la clasificación de señales del diagnóstico scattering Thomson
(Jornadas de Automática 2004, 2004-01-01) Farias Castro, Gonzalo Alberto; Santos, Matilde; Fernández Marrón, José Luis; Dormido Canto, Sebastián
Distributed reconfiguration of distance-based formations with virtual surface constraints
(IEEE, 2024) Guinaldo Losada, María; Sánchez Moreno, José; S. Zaragoza; Mañas Álvarez, Francisco José
This paper proposes a method to recover from the failure or loss of a subset of agents in a distance-based formation problem, where the system is initially deployed forming a virtual shield embedded in the 3D space. First, a distributed algorithm is proposed to restore the topology, which is a Delaunay triangulation. After that, the nodes execute a distance-based distributed control law that considers adaptive target distances. These values are computed in parallel by the nodes, which try to reach an agreement with some constraints, given by the desired shield shape. The updating policy is based on events. The results are illustrated through simulation examples.
Distributed targeted distance-based formation control for mechanical systems
(IEEE Xplore, 2020-07-20) Aranda Escolástico, Ernesto; Colombo, Leonardo; Guinaldo Losada, María
This paper studies the problem of distributed targeted distance-based formation control for mechanical systems. The problem consists on finding a distributed control law such that if each agent observes a convex set as a targeted set, and also the relative position of their nearest neighbors, then the agents must achieve the desired formation in these sets while its velocities are driven to zero. We study the problem for agents with a time-delay communication in the measurements of the relative positions and where the motion of each agent is determined by a Lagrangian function. Simulation are given to validate the theoretical result.
Error dependent sampling to reduce transient events in event-based control
(Institute of Electrical and Electronics Engineers, 2025-03-25) Miguel Escrig, Oscar; Romero Pérez, Julio Ariel; Sánchez Moreno, José; Dormido Bencomo, Sebastián
One of the objectives of event-based control is the reduction of generated events to perform an appropriate process control. Among the event generation techniques used to this end, those quantifying the error signal can be found more frequently for its simplicity of implementation. However, they entail a drawback in the form of trade-off between event generation and performance. Aiming to reduce the number of events, specifically in the transient response, without degrading the performance, an error dependent sampling scheme is studied in this work.
Estimación Automática del Coste de Comunicación de Aplicaciones Paralelas en Plataformas Heterogéneas
(Universidad Extremadura, 2018) Moreno Álvarez, Sergio; Rico Gallego, Juan A.; Díaz Martín, Juan Carlos; https://orcid.org/0000-0002-4264-7473; https://orcid.org/0000-0002-8435-3844
Optimizar el tiempo de ejecución de aplicaciones paralelas en plataformas heterogéneas de altas prestaciones es un problema complejo. Estas aplicaciones cient´ıficas normalmente se componen de kernels que implementan algoritmos como la multiplicación de matrices, ecuaciones en derivadas parciales o Transformadas de Fourier. Los kernels son ejecutados por los procesos desplegados en los diferentes recursos de cómputo de una plataforma, por ejemplo, en procesadores multi-core o aceleradores (GPUs, Xeon PHIs, etc.). El volumen de datos del kernel se distribuye entre los procesos de forma proporcional a su capacidad de cómputo, de forma que se equilibra la carga computacional global. Este equilibrado de carga no homogéneo tiene un impacto importante en el coste de las comunicaciones. La optimización del coste de las comunicaciones de éstas aplicaciones se aborda habitualmente mediante pruebas exhaustivas en la plataforma destino. Sin embargo, estas pruebas consumen recursos y tiempo, y a menudo se basan en la extrapolación de los resultados obtenidos con la ejecución de una versión reducida de la aplicación en la plataforma. Los Modelos Anal´ıticos de Rendimiento de Comunicaciones ofrecen una alternativa factible y prometedora en este sentido. Estos modelos representan el coste de las comunicaciones de un kernel en una plataforma heterogénea, ofreciendo una estimación precisa de su tiempo de comunicación de forma no invasiva, esto es, sin utilizar recursos de cómputo HPC en la estimación. Este trabajo contribuye ofreciendo una herramienta de estimación que permite representar y evaluar expresiones de coste de comunicaciones que siguen el modelo t- Lop. Adem´as, permite incluir el c´alculo de coste de las comunicaciones de forma autom´atica en algoritmos de particionamiento y optimización de comunicaciones. En este documento se proporcionan ejemplos tanto de uso b´asico como avanzado. Se incluyen tres casos de ejemplo de modelado de comunicaciones en kernels representativos utilizando la herramienta: la solución de una ecuación diferencial utilizando la técnica de elementos finitos, un algoritmo paralelo de multiplicación de matrices densas, y una simulación N-Body. Estos kernels utilizan diferentes patrones de comunicación y particionamiento del espacio de datos.
Evaluación de Rendimiento del Entrenamiento Distribuido de Redes Neuronales Profundas en Plataformas Heterogéneas
(Universidad de Extremadura, 2019) Moreno Álvarez, Sergio; Paoletti, Mercedes Eugenia; Haut, Juan Mario; Rico Gallego, Juan Antonio; Plaza, Javier; Díaz Martín, Juan Carlos; Vega Rodriguez, Miguel ángel; Plaza Miguel, Antonio J.; https://orcid.org/0000-0003-1030-3729; https://orcid.org/0000-0001-6701-961X; https://orcid.org/0000-0002-4264-7473; https://orcid.org/0000-0002-8908-1606; https://orcid.org/0000-0002-8435-3844
Asynchronous stochastic gradient descent es una tecnica de optimizacion comunmente utilizada en el entrenamiento distribuido de redes neuronales profundas. En distribuciones basadas en particionamiento de datos, se entrena una replica del modelo en cada unidad de procesamiento de la plataforma, utilizando conjuntos de muestras denominados mini-batches. Este es un proceso iterativo en el que al nal de cada mini-batch, las replicas combinan los gradientes calculados para actualizar su copia local de los parametros. Sin embargo, al utilizar asincronismo, las diferencias en el tiempo de entrenamiento por iteracion entre replicas provocan la aparicion del staleness, esto es, las replicas progresan a diferente velocidad y en el entrenamiento de cada replica se utiliza una vers on no actualizada de los parametros. Un alto gradde staleness tiene un impacto negativo en la precision del modelo resultante. Ademas, las plataformas de computacion de alto rendimiento suelen ser heterogeneas, compuestas por CPUs y GPUs de diferentes capacidades, lo que agrava el problema de staleness. En este trabajo, se propone aplicar t ecnicas de equilibrio de carga computacional, bien conocidas en el campo de la Computaci on de Altas Prestaciones, al entrenamiento distribuido de modelos profundos. A cada r eplica se asignar a un n umero de mini-batches en proporci on a su velocidad relativa. Los resultados experimentales obtenidos en una plataforma hete-rog enea muestran que, si bien la precisi on se mantiene constante, el rendimiento del entrenamiento aumenta considerablemente, o desde otro punto de vista, en el mismo tiempo de entrenamiento, se alcanza una mayor precisi on en las estimaciones del modelo. Discutimos las causas de tal incremento en el rendimiento y proponemos los pr oximos pasos para futuras investigaciones.

Examinar

Examinando Actas y comunicaciones de congresos por Centro "E.T.S. de Ingeniería Informática"

Resultados por página

Opciones de ordenación