ABSTRACT The continuous evolution of techniques for generating manipulated audio, known as voice deepfakes, and the widespread availability of tools that produce convincing forgeries have created an urgent need for reliable detection methods. This work considers the dimensionality of Mel‐Frequency Cepstral Coefficients (MFCCs) as a core design variable for practical, deployable systems. The aim is to identify the smallest number of coefficients that preserve detection performance across heterogeneous models while reducing computational cost, a critical factor for mobile and edge deployment. This study evaluates a hybrid setting on the ASVspoof 2019 Logical Access dataset, in which the same feature family serves as input to five traditional machine learning algorithms (Random Forest, k‐Nearest Neighbours, Linear Support Vector Classification, Extreme Gradient Boosting and Support Vector Machine with radial basis function kernel) and five deep learning models (Convolutional Neural Network, Recurrent Neural Network, Convolutional Recurrent Neural Network, Xception and ResNet). Results indicate that deep models reach near‐peak performance with a small number of coefficients, whereas classical methods require a larger number to achieve stable performance (except Linear Support Vector Classification, which consistently underperforms). Accordingly, 32 coefficients are considered an effective operating point for hybrid deployments. Overall, the results provide evidence to guide the selection of the number of MFCC coefficients in voice deepfake detection, aiming for efficient, reproducible and explainable systems.

The scope of this work focuses on the digital assets of citizens and their personal and private inheritance. The aim is to protect such digital assets after your death, for example, personal email accounts, not those within the corporate domain, which are neither inheritable nor owned by the employee. It is based on blockchain technology due to the intrinsic need to govern hereditary information. These needs include a public and secure record of transactions that cannot be altered, confidence in the integrity of stored data, decentralization of information, the immutability of records, efficiency, and speed. These features empower users with greater control over their data and transactions. Users, as beneficiaries of their digital assets, will be able to establish specific, planned, and transparent conditions for their digital assets after death safely and transparently, ensuring that the records of these transactions are secure and auditable. By adopting these solutions and planning their digital estate in advance, users can ensure that it is managed securely and effectively for future generations.

Hyperspectral unmixing addresses the challenge of mixed pixels in hyperspectral images by identifying the number of pure pixels (endmembers), extracting their spectral signatures, and estimating their proportions (abundances) in each pixel composing the scene. Traditional hyperspectral unmixing methods often struggle with scalability and computational efficiency when dealing with gigabyte-scale datasets. In this paper, we propose a distributed parallel geometric distance (DPGD) method for hyperspectral unmixing, exploring the computational power and benefits of distributed parallel processing within a distributed computing framework. The proposed DPGD leverages geometric distance measurements to accurately identify endmembers and estimate their abundances, taking into account the intrinsic similarities within hyperspectral images. This provides a clearer representation of the data structure, leading to improved unmixing accuracy. By using the Spark programming model, the computational workload is efficiently distributed across multiple nodes, significantly reducing processing time. Experimental results on real hyperspectral datasets demonstrate that DPGD scales effectively up to 32 nodes and 290.9 GB of data, achieving competitive accuracy and efficiency compared to state-of-the-art methods. The code is available at https://github.com/ccaadaro/DPDG.

Cancer is a widespread global health problem, claiming millions of lives each year, and skin cancer represents a significant threat as it is one of the most common types. Early tumor detection via medical imaging is critical for effective treatment. Leveraging artificial intelligence, particularly novel models like Transformers, presents promising avenues for improved diagnosis. This paper explores the efficacy of a Collective Intelligence approach using AI in classifying cancerous and non-cancerous tumors, aiming to reduce classification errors and support clinical decision-making. We created five different configurations using various datasets to compare the results. The results show solid performance for the CI in the evaluated tasks, reaching up to 75.89% accuracy. The lack of images in certain classes significantly contributes to overfitting. It is suggested to explore data expansion strategies and improve consistency in image capture for future work.

The widespread availability of remotely sensed datasets establishes a cornerstone for comprehensive image retrieval within the realm of remote sensing (RS). In response, the investigation into hashing-driven retrieval methods garners significance, enabling proficient image acquisition within such extensive data magnitudes. Nevertheless, the used datasets in practical applications are invariably less desirable and with long-tailed distribution. The primary hurdle pertains to the substantial discrepancy in class volumes. Moreover, commonly utilized RS datasets for hashing tasks encompass approximately two–three dozen classes. However, real-world datasets exhibit a randomized number of classes, introducing a challenging variability. This article proposes a new centripetal intensive attention hashing (CIAH) mechanism based on intensive attention features for long-tailed distribution RS image retrieval. Specifically, an intensive attention module (IAM) is adopted to enhance the significant features to facilitate the subsequent generation of representative hash codes. Furthermore, to deal with the inherent imbalance of long-tailed distributed datasets, the utilization of a centripetal loss function is introduced. This endeavor constitutes the inaugural effort toward long-tailed distributed RS image retrieval. In pursuit of this objective, a collection of long-tail datasets is meticulously curated using four widely recognized RS datasets, subsequently disseminated as benchmark datasets. The selected fundamental datasets contain 7, 25, 38, and 45 land-use classes to mimic different real RS datasets. Conducted experiments demonstrate that the proposed methodology attains a performance benchmark that surpasses currently existing methodologies.

In recent years, the rapid development of remote sensing (RS) technology has led to a drastic increase in the availability of RS images. This calls for the need to develop new methods able to effectively and efficiently retrieve the required instances from a massive amount of RS imagery. In retrieval tasks, finding the nearest-neighbor sample of the retrieval query is a fundamental research topic. Exhaustive comparison is the simplest method to accomplish this task. However, due to the involved computational complexity and memory limitations, this solution is no longer feasible in large data retrieval tasks. As an important branch of approximate nearest-neighbor retrieval (NNR), hash algorithms transform high-dimensional data into low-bit expressions (hash codes) with elements of 0 and 1 to reduce storage and computational costs. Hash algorithms aim to preserve the same nearest-neighbor relationship between the learned hash codes and the original data. Existing hash algorithms are divided into two classes: shallow and deep methods. Furthermore, deep hash algorithms can be divided into (semi-) supervised and unsupervised algorithms. In this article, representative hash-based RS image retrieval (HBRSIR) methods are reviewed, studying the application of hashing in other areas of the RS community and introducing available datasets and evaluation metrics for RS image retrieval (RSIR). The performance of representative and cross-modal hashing methods is validated using two common RSIR datasets (UCMerced and AID) and a cross-modal dataset (DSRSID). Prospects of future work summarizing HBRSIR are also provided.

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.

Classifying remotely sensed hyperspectral images (HSIs) became a computationally demanding task given the extensive information contained throughout the spectral dimension. Furthermore, burgeoning data volumes compound inherent computational and store challenges for data processing and classification purposes. Given their distributed processing capabilities, cloud environments have emerged as feasible solutions to handle these hurdles. This encourages the development of innovative distributed classification algorithms that take full advantage of the processing capabilities of such environments. Recently, computational-efficient methods have been implemented to boost network convergence by reducing the required training calculations. This paper develops a novel cloud-based distributed implementation of the Extreme Learning Machine (CC-ELM) algorithm for efficient HSI classification. The proposal implements a fault-tolerant and scalable computing design, whilst avoiding traditional batch-based back-propagation. CC-ELM has been evaluated over state-of-the-art HSI classification benchmarks, yielding promising results and proving the feasibility of cloud environments for large remote sensing and HSI data volumes processing. Code available on: https://github.com/mhaut/scalable-ELM-HSI.

Eating disorders (ED) are among the most widespread mental illnesses in our society today. This research work presents the study of deep learning models applied to the domain of eating disorders. For this purpose, a collection of messages from the social network Twitter was compiled using web scraping techniques. After collecting a total amount of 1,085,957 tweets, a subset of 2,000 tweets was manually classified. This classification made it possible to differentiate tweets written by people who suffer or have suffered from an ED from those written by people who have not suffered from an ED. After this, 6 predictive models based on Bidirectional Encoder Representations from Transformers (BERT) were created and a comparison was made by evaluating which model scored the best. The best scoring model was RoBERTa using the pre-trained roberta-base model with an accuracy of 87.5%.

This paper proposes a new recommendation system in the context of the HOPE project with the aim of providing medical bibliographic references in a simple, up-to-date and immediate way. In addition, these references are catalogued according to the patient information and symptoms, offering a ranking mechanism which sorts them from most interesting to least relevant ones according to the feedback provided by health professionals. The proposed system has been extensively trained and validated with a set of widely used machine learning models, particularly Random Forest (RF), Multinomial Logistic Regression (MLR) and Support Vector Machines (SVMs). The results obtained over real medical data from HOPE project are quite promising, exhibiting a high precision. In particular, RF is the algorithm which the best behavior with a 89.9% of precision. It is closely followed by the SVM, which reaches great results with a 89.4% of precision, performing quite accurately with false negative cases.

The instructional design of experimental subjects with a distance methodology in Engineering disciplines is nowadays a challenge. One of the most useful applications is the use of remote and virtual laboratories technologies. The learning process must transparently integrate these technologies. It is also essential to be able to analyze the students learning and acceptance of this technology. In particular, the IoT paradigm involves the acquisition of practical skills and knowledge from the use of devices/sensors to the management of cloud environments, by including data storage and processing. A Laboratory of Things platform at UNED (called LoT@UNED) has been developed to cover all phases of a life-cycle development, both hardware infrastructures and software applications. This work focuses on studying the influence among several acceptance indicators during the learning/teaching process of students for the cyber-security topic, and with respect to their intention of employing LoT@UNED for other learning purposes. To achieve this, a new structural equation models (SEM) has been proposed, as well as an statistical analysis is included from a confirmatory point of view.

Despite its great potential in a wide range of human activities, hyperspectral remote sensing imaging (HSI) exhibits several challenges that prevent full exploitation of its data. In particular, land-cover classification based on HSI data suffers significant degradation due to problematic data variability. Convolutional Neural Networks (CNNs) ability to extract spectral-spatial features has enabled the development of powerful classifiers, which achieve not yet seen accuracy results. To enhance the feature extraction procedure, this paper presents a novel HSI-CNN model (DKDCNet) which combines adaptive deforming kernels (DK) and convolutions (DC) with the aim of pinpointing the effective receptive field (ERF) on the challenging input data. Experimental results on the University of Houston benchmark show that DKDCNet is able to obtain a more accurate classification than traditional strategies with similar computational cost for HSI classification. Source code: https://github.com/mhaut/DKDCNet.

The inclusion of practical activities to train critical thinking skills is very relevant in the case of Engineering subjects. In particular, it becomes a key element in the context of cybersecurity, as our case is. Therefore, this paper analyses a gambling experience in a cybersecurity subject within a Computer Engineering degree, in order to assess whether the spaced learning methodology contributes to the improvement of learning. The gamification experience is carried out through a Flag Capture Competition (CTF) among the enrolled students in the course. The use of the platform has also been evaluated in an exploratory way.

The MUREE project addresses the needs of Jordanian Universities to modernize the learning of Engineering related to renewable energies in collaboration with European Universities. After the implementation of several courses, which combined online learning with traditional teaching in most cases, it is time to analyze and verify the benefits o f t his approach in the Jordanian society. In particular, this paper outlines the results obtained with this project, from three points of view: educational impact, research impact, and socio-economic impact. Results detail the students' participation, their profile in terms of training and required skills, and some aspects about recruitment of future professionals in renewable energies.

SNOLA is a Thematic Network of Excellence recognized by the Spanish Ministry of Economy and Competitiveness composed of the main Spanish researchers in the field of learning analytics. This network emerged in 2013 focusing mainly on the technology underlying learning analytics development, but also interested on the integration of other views and disciplines that give the network a wider scope. During these years, SNOLA has carry out numerous actions with the goal of promoting the collaboration on this field, the diffusion of knowledge and initiatives, the collection of resources and the provision of broad support. Remarkable examples of these efforts have been LASI Spain events, workshops at TEEM conferences or the numerous webinars. In this case, the EDUCON 2018 has given us the opportunity to organize this special session where 5 papers have been selected for presentation.

This work deals with the teaching of the innovative technology, named cloud computing, using the Web of Things (WoT) platform model based on web services. These services are designed and programmed by the students to handle embedded hardware devices (things) on Internet. The course is carried out within a makerspace where our students can take advantage of valuable on-line tools which are available in a collaborative learning environment. The introduction of these innovative technological elements improves the students' interest and engagement leading to achieve better learning results.

One of the key competences for Europe in the 21st Century is the technological competence understood as an instrument for development, aimed at encouraging society to use and like science. This competence can only be fully achieved by using practical experiments, which present use cases in a realistic, creative, and revolutionary way. This paper presents our work on the development of practical experiments in the context of several pre-university levels, from pre-school to elementary school, aimed at enhancing the appeal of Science, Technology, Engineering and Mathematics (STEM) among students.

MUREE project aims at the development of courses for training specialists about renewable energy production by combining face-to-face learning with on-line attendance. In this sense, remote laboratories are nowadays essential for distance education, even more within MUREE project, since Jordan students are not able to use face-to-face traditional laboratories due to their different physical location. These remote laboratories can be employed by instructors within their virtual classrooms, so that students can carry out their on-line experiments from anywhere and at anytime. For MUREE project, remote laboratories are seen as pedagogical elements that must be fully-integrated into a the learning/teaching process. Therefore, this work focuses its attentions toward the integration of remote laboratories into Learning Management Systems (LMSs) and, additionally, discusses the advantages and disadvantages of this approach.

The creation and publication of utilities as services (the most widely known being Infrastructure as a Service, IaaS, Platform as a Service, PaaS, and Software as a Service, SaaS) has been a hot topic of research and development for the recent years. They allow easy creation and deployment of infrastructures and applications which increase the versatility and usefulness of the Information Technology (IT) budgets of institutions that implement them. Therefore, this paper proposes the development of Laboratories as a Service (LaaS), which allow users of remote laboratories create versatile experiments adapted to their needs. These will be based on the deconstruction of remote laboratories, creation of clients, and selection of a container. An aeolian laboratory based on the Lego Mindstorms robotic kit is used as an example.

Development of virtual/remote laboratories is a common task involved in the design of course's assessments (in engineering disciplines). The deployment of these laboratories must manage several aspects related to the real use of them, as user's authorization and access, tracking of usage information, loading/saving data from experiments and so on. These features must be implemented by developers in a particular way for each laboratory. This paper shows how these services can be used automatically, with no development, using the RELATED management services. The deployment task consists of making a connector to the RELATED framework in order to consume these services. A full example is shown, describing the steps followed to integrate a previously developed low cost laboratory (based on a Lego system).

In the field of Engineering Education, performing practical experiments is essential, as an accompaniment to magisterial classes, in order to enforce theoretical and practical concepts. This task is much more difficult in distance education since students attend mainly to virtual classes. Therefore, the use of remote laboratories can help minimize this inconvenient. In addition, remote laboratories may help teachers to prepare suitable evaluation on-line experiments, in similar conditions if students were physically in the real laboratory, on the other hand, remote laboratories can be a good complement to real experimentation, as a preparatory step, preparing them to face similar tasks in the immediate future. Accordingly, the principal objective of the European TEMPUS project entitled "Modernizing Undergraduate Renewable Energy Education: EU Experience for Jordan", which supports this work is the development, integration, accreditation, and evaluation of a renewable energy course in the context of engineering degrees from several universities in Jordan. This project follows the guidelines proposed in the Bologna process, and considers the previous experimentation with low-cost renewable energy equipment in order to allow us to study the best approximation of remote laboratories. This is a previous step before addressing this task with complex and expensive equipment.

Practical knowledge is increasingly getting more attention in the higher education, especially in the field of engineering education. Engineering is a discipline which has, by its own definition, a large amount of practical contents. Allowing students to interact with real equipment can give to them a qualitative knowledge that cannot be obtained in other way. In this situation, remote laboratories are especially useful. This paper presents a way of reusing remote labs code by its XML representation. The idea is that students learn not only the basic subjects of the course, even more, that students do not only manipulate lab equipment. The idea is to give the students a practical way to check the potential of work organization and to give a taste of the powerfulness of code reusing, that is, in the end, a way to improve work efficiency.

On-line laboratories (virtual or remote) are widely used in experimental engineering subjects as part of the learning process. In order to develop these laboratories, a development framework called RELATED (Remote Laboratories exTendED) is used by the Communications and Control System Department of the Spanish University for Distance Education of Spain (UNED). This framework defines a structured and methodological development procedure, allowing the students the generation of their own laboratories. Once the laboratory is developed (based in its components), students have to configure their own computing resources in order to make their labs available. However, several problems must be faced by students in the “deployment” of their labs: network configuration, hardware availability, and so on. So, in order to solve these problems, an automatic system based on cloud providers is defined to allow students having their own cloud network/resources for their developed labs. This system simplifies the lab deployment and avoids common errors/mistakes in the development of laboratories with RELATED.

This paper presents a new educational system to automatically adapt the evaluation activities to the students' needs in the context of Higher Engineering Education. As an example, a subject focused on the configuration of network services has been chosen to implement our proposal. Therefore, the system will be able to guide each student through the learning process based on his/her particular knowledge-level. In addition to this, specific techniques are needed to dynamically evolve the system depending on the students' progress. In our case, this is analyzed by using data mining techniques. Finally, we show survey results, which illustrate the ease of use and usefulness of the system.

Merging a Learning Management System (LMS) and a social network has social advantages for students. Among others, students can show off their achievements among their acquaintances, which in turn improves their social leadership and self-esteem. This is specially true in the case of distance learning, where there are no physical interactions between students, so their student life is totally isolated from their everyday life (e.g. students do not go out together, since they may live in different cities or even countries although they study the same subjects). All these points suggest that the use of social networks for distance teaching has clear advantages that can be harnessed. In order to achieve the aforementioned benefits, we are working on integrating a commonly used LMS, Moodle, into one of the most widely-used social network, Facebook. This integration allows students and faculty to use Facebook as a communication tool that improves the learning process and social life of students in the ways presented before.

Nowadays, the use of remote laboratories is a common feature in order to get students involved in practical experiences. This is particularly important in distance learning environment. The regular practice is to develop a particular laboratory, required by the student competence's curricula. Usually, this involves a lack of methodology or a standard procedure so, new developments must be done to include new laboratories, additionally so many “manual” procedures are required. RELATED (REmote LABoratory Extended) it's used to solve these issues. In this paper, a practical experience of the structured development process of a virtual laboratory, using RELATED framework, will be presented from the point of view of students. This virtual laboratory is a simple signal generator. This signal generator will be used to generate a reference position for the height of the ball of an electromagnetic levitation system in a control experiment. Finally, responses of surveys made by students will be presented, in order to get satisfaction results of using/development of virtual/remote laboratories using RELATED.

The development of cloud technologies allows the implementation of scalable, versatile, and customized systems, constructed on-demand. This allows more efficient use of computing resources, improving the revenue of the system and enhancing the Quality of Service (QoS) received by users while minimizing the power consumption of the machines. Several research works conclude that in order to efficiently manage a cloud-based infrastructure (meaning, deploy computing resources when needed without affecting negatively to the QoS perceived by users), accurate predictions on the load of machines should be made. Thanks to this, resources can be ready to use when users need them, and shutdown when they are not needed - thus reducing the power consumption and enhancing the revenue of the system. This paper presents algorithms to perform forecasts of the load of machines based on Exponential Smoothing (ES), so that the machines of the technological infrastructure of our University can be efficiently managed. Furthermore, algorithms to perform monitoring and provision of resources based on load forecasts are presented. The usefulness of these algorithms is illustrated by means of a use case based on the e-learning facilities of our University. This use case shows that thanks to the use of cloud technologies, enhanced with the developed algorithms for load forecasting and provision of resources, better use of resources and lower power consumption can be achieved, without affecting the QoS received by user.

Use of multimedia services has become very important in learning process. This is due to the traditional approach of document based assignments not provide any feature which allows to evaluate some basic competences (oral communications skill, for example). In this paper, the UNED (Spanish University for Distance Education) solution will be presented focusing on user's interaction: lecturers and students. In the first case, the lecturer has to define a task and propose a solution (available only at the end of task availability). This solution is also video/audio based, so the student can learn about it and compare it with his/her solution. The student has several attempts, in order to try better responses and decides which of these attempts will be selected as final response. The whole interaction schema will be presented and how is integrated in the e-learning platform as a usual task (like document based tasks). This integration permits the use of all features of the platform evaluation tool, simplifying the evaluation process and tasks grading (no special procedures are required).

This work presents a new system for the automatic assessment of practical activities in the context of the EHEA. A subject focused on the configuration of network services has been chosen to implement the new automatic evaluation platform. Unlike traditional platforms based on theoretical contents to evaluate the students' knowledge, the proposed system is able to immediately evaluate students' practical skills and, additionally, provide them with feedback about the correctness of their activities. Therefore, faculty can dynamically follow the students' progress in order to adjust the learning process to their needs. Since the UNED University serves a large number of students via a distance-learning methodology, the use of scalable assessment platforms within the students' learning process is crucial to keep it working efficiently. Automatic assessment systems are also of particular interest when practical activities are performed within higher engineering courses.

Thanks to the development of cloud technologies, the way how computing is understood has evolved from "computer guided" to "user guided" systems. That is, initially, computers had static software features in which users sharing them had to "fit", but now there is a shift to dynamic systems in which it is the computer which has to fit into the users' needs. This allows more efficient use of computing resources, improving the revenue and enhancing the Quality of Service (QoS) received by users. In order to deploy computing resources when needed without affecting negatively to the QoS perceived by users, accurate predictions on the load of machines should be made. Thanks to this, resources can be ready to use when users need them, and shutdown when they are not needed. This reduces the power consumption and enhances the revenue of the system. This paper presents an algorithm to perform resource provisioning on the machines of the technological infrastructure of our University, so that they can be efficiently managed. This algorithm is based on load forecasts created using Exponential Smoothing.

This work presents a new e-learning architecture for the automatic assessment of practical activities in the context of Higher Engineering Education. As an example, a subject focused on the configuration of network services has been chosen. Unlike traditional platforms based on theoretical contents, the proposed architecture will be able to immediately evaluate students' practical skills and provide them with feedback about the correctness of their activities. Therefore, lecturers will dynamically follow the students' progress.

Personal Learning Environments (PLEs) have gained importance over recent years thanks to the wide use of the Web 2.0. An interesting functionality is the creation of customized PLEs, to improve the learning process. This work proposes a new paradigm to create customized PLEs by deconstructing a Virtual Learning Environment (VLE).