Computer security is delving into new fields, and short-distance communication by light is one of them due to its advantages over traditional means that use radio frequency. A few advantages are not reading from a close distance, operating only when the user requires it, and short-range communication that is ideal for security applications and access control. This paper presents the development in the access control prototype with smartphones using light communication. Diverse sensors and flash response speeds were evaluated with physical tests in several cell phones; the same minimum response frequency of 60 Hz was obtained for different devices, and depending on the device, a maximum response of 140 Hz was obtained. Using this information, an app was developed on the Flutter platform to send data frames implementing protocols like infrared, where the signal is encoded with the flash camera’s pulse-width modulation. Using an embedded system known as ESP8266 WeMos D1 Mini, data bytes could be sent, received, synchronized, and verified from the cell phone at a frequency of 30 Hz modulation per bit.
Predicting financial budgets remains an open challenge. Most studies focus on high-frequency markets, such as stock price prediction, leaving the budget domain aside. This work addresses that gap. We evaluate four artificial intelligence techniques for budget prediction: decision trees, random forests, linear regression, and multilayer perceptron. Additionally, we implement a hybrid version, MLP_GA, where the network hyperparameters are optimized using a genetic algorithm. For training and validation, we use two original datasets not publicly available: quarterly financial statements from Mexican entities (2010-2024) and monthly financial statements from Ecuadorian organizations (2019-2022). Both present characteristics typical of emerging economies, including episodes of high inflation, currency devaluations, fiscal policy changes, and the COVID-19 crisis. The results show that MLP_GA achieves superior predictive performance compared to the standard multilayer perceptron and the other techniques. Genetic algorithm optimization enables exploring multimodal and non-convex error surfaces, finding hyperparameter configurations (number of hidden neurons, learning rate, L2 regularization coefficient, maximum iterations, and early stopping tolerance) that significantly reduce overfitting, a critical problem when dealing with short time series of only forty to sixty quarterly observations. Statistical tests confirm that the differences are significant. The contribution is threefold. First, regarding the application domain, we present the first systematic and comparative evaluation of hybrid models combining genetic algorithms with multilayer perceptrons specifically applied to financial budget prediction, distinguishing ourselves from existing studies that focus almost exclusively on high-frequency stock market prediction. Second, regarding the geographical context, we validate our models in two Latin American economies that the literature has systematically ignored: Mexico and Ecuador, a region representing less than one percent of datasets used in the literature according to recent meta-analyses. Third, regarding the methodological justification, we demonstrate that genetic optimization is particularly effective for moderate-sized data, where more complex architectures like LSTMs or transformers suffer from overfitting due to their high parameter count. The combination of monthly and quarterly analysis improves model robustness, allowing accurate performance in both short-term scenarios and broader projections. These findings suggest that the MLP_GA approach can improve financial budgeting in data-constrained contexts. The paper concludes with a discussion of limitations and future research lines.
Los semilleros de investigación universitarios combinan objetivos pedagógicos con desarrollo real de software, introduciendo tensiones organizacionales que el SDLC estándar no contempla. Este trabajo presenta un análisis retrospectivo del desarrollo de la versión 1.0 del Aplicativo del Banco de Ideas (ABI), desarrollado por el semillero Master Digital de la Universidad de Investigación y Desarrollo (UDI), Bucaramanga, Colombia. A partir de evidencia documental —reportes semanales, artefactos de diseño, registros Git y actas— se identificaron cinco dimensiones de desviación recurrente: gobernanza y dirección académica, gestión de proyecto y coordinación, coherencia metodológica y definición temprana del stack tecnológico, documentación y gestión del conocimiento, y responsabilidad individual. La ausencia de mecanismos formales de gobernanza generó un atraso acumulado de entre siete y nueve meses, concentrando la carga operativa en un subconjunto reducido de integrantes. Se proponen recomendaciones organizacionales adaptadas a entornos formativos para fortalecer la trazabilidad, continuidad técnica y estabilidad operativa de proyectos similares en semilleros universitarios latinoamericanos.
Actualmente, hospitales en Huila-Colombia necesitan sistemas de información hospitalaria para garantizar la integridad y disponibilidad de la información de pacientes; además, tener disponible la información de pacientes en línea. Aunque, los sistemas son necesarios para muchas actividades, los hospitales del Huila no cuentan con sistemas de información. En consecuencia, personal de salud utiliza métodos no estructurados para acceder a los datos; que ocasiona la toma de decisiones inadecuadas. Por lo tanto, este trabajo presenta el diseño e implementación de un sistema de información para optimizar el proceso de gestión de pruebas de análisis de orina en hospitales del Huila, que mejore el acceso a la información. El sistema implementado utiliza el estándar HL7-FHIR, sobre plataformas Open-Source, y basado en la Web, escalable y compatible con otros sistemas de información hospitalaria. Finalmente, la solución propuesta minimiza tiempos de acceso y gestión de las pruebas integrales realizadas por el equipo médicos.
Este artículo presenta una investigación orientada a explorar la integración de inteligencia artificial (IA) en entornos de pair-programming apoyados en desarrollos tecnológicos dentro de la educación superior. En una primera fase, se realizó una revisión sistemática de literatura siguiendo la metodología PRISMA, a partir de 35 fuentes académicas indexadas en Springer, ACM y ScienceDirect, con el fin de identificar las tendencias, beneficios y limitaciones en el uso de asistentes conversacionales, tutores virtuales y modelos de lenguaje de gran escala (LLMs) aplicados a la enseñanza de la programación. Los resultados de esta revisión evidencian que la IA favorece la motivación, reduce la ansiedad y contribuye a mejorar el rendimiento académico, aunque no reemplaza la interacción social propia del aprendizaje colaborativo, además de mostrar vacíos en la integración de estas soluciones en sistemas de gestión como Moodle y en contextos latinoamericanos. A partir de estas conclusiones, se diseñó un estudio exploratorio en el aula con una herramienta prototipo que combina pair-programming, IA y un desarrollo tecnológico, cuyo objetivo fue analizar la percepción estudiantil sobre su uso, con la participación de 94 estudiantes de pregrado. Los hallazgos preliminares muestran una valoración positiva, destacando la utilidad del soporte automatizado para la retroalimentación inmediata y la reducción de la carga cognitiva, aunque los estudiantes enfatizaron la necesidad de un acompañamiento docente constante y un andamiaje progresivo para consolidar el aprendizaje. Se concluye que la articulación entre pair-programming, IA y Desarrollos tecnológicos constituye una innovación pedagógica prometedora, cuya adopción efectiva depende de políticas institucionales claras y de una implementación responsable.
Los conjuntos de herramientas de Inteligencia Artificial (IA) están siendo utilizados para desarrollar software innovador con gran rapidez, aunque seguir estos atajos podría generar Deuda Técnica de la IA (DTIA). Este tipo de deuda puede tener consecuencias tanto en lo técnico, como en lo social. Sin embargo, los desarrolladores no siempre son conscientes de si están incurriendo en DTIA durante sus proyectos. Peor aún, desconocen sus impactos sociotécnicos y cómo mitigarlos. Para reducir esta brecha, se desarrolló una Revisión Sistemática de la Literatura (RSL) para averiguar cuáles son los tipos y prácticas de DTIA, así como cuáles son sus repercusiones sociotécnicas. Los resultados arrojaron variedad de tipos y antipatrones de DTIA, así como prácticas para identificarla y mitigarla. También, se encontró que existen daños a la calidad, seguridad y mantenibilidad del producto, además de retos como sesgos y discriminaciones. Esta información puede apoyar a científicos y practicantes para detectar y gestionar la deuda técnica y social en contextos de desarrollo de software habilitado por la IA.
Ante la imperante necesidad de salvaguardar la seguridad alimentaria mundial, diversos sectores académicos, gubernamentales y empresariales han dedicado esfuerzos para optimizar la producción agrícola a pequeña escala mediante tecnologías como la agricultura de precisión. No obstante, la creciente brecha tecnológica, explicada principalmente por las capacidades digitales y la deficiencia de infraestructura propia de las zonas rurales, entorpece la implementación de dichas tecnologías en dicho sector. Este artículo propone una causa adicional relacionada con las características de diseño empleadas en el manejo campesino-solución, resultado de una revisión sistemática de la literatura de 101 artículos científicos publicados entre 2015 y 2021, así como de una vigilancia tecnológica de cuatro empresas reconocidas a nivel mundial por el desarrollo de soluciones para el riego. Se identificaron ocho características de interacción en las distintas soluciones seleccionadas para el estudio, lo que facilitó la comparación de los diseños implementados en el ámbito académico y en la industria. Dichas características contribuyeron a identificar las deficiencias en el desarrollo de innovaciones agrícolas digitales para el sector agrícola de pequeña escala y a analizar su influencia en las barreras de apropiación tecnológica.
Las proyecciones del precio de la energía eléctrica son un insumo relevante para la evaluación de los resultados financieros de empresas en operación, la evaluación de nuevos proyectos de inversión y la estructuración de portafolios de cobertura. En este artículo realizamos un análisis de la eficacia de métodos de proyección del precio spot de la energía eléctrica en Colombia, basados en Deep Learning y su capacidad para vencer una estructura de tipo ARIMA. Los resultados son insumo para investigadores y analistas de mercados eléctricos a nivel global que estén interesados en realizar proyecciones de la energía eléctrica y también algunas variables que posean características de reversión a la media, alta volatilidad y saltos. En general, los métodos de Deep Learning considerados logran capturar la tendencia de la serie en periodos out of sample, pero no logran vencer el desempeño de pronósticos realizados al aplicar métodos de Box-Jenkins.
There are several proposals in the literature on the application of Natural Language Processing – NLP to address activities and challenges in requirements engineering – RE. In recent years, Generative Artificial Intelligence – AI implemented with Large Language Models – LLM has gained great recognition due to the improvements contributed to NLP tasks. This work proposes a systematic literature review – SLR to collect research that presents some use of LLMs to solve problems and improve the RE process. The results show promising proposals aimed mainly at different model creation and requirements classification tasks. However, these proposals need more development and empirical validation to be widely accepted and applied in software development environments. Therefore, it is necessary to continue researching the applications of LLM in the RE process.
Project-based learning is a work strategy that facilitates the development of real projects within an educational setting, enabling students to develop various skills in preparation for their professional careers. This approach is widely used in technological fields, including software engineering, with the goal of developing quality software systematically. This research presents a comparative study of groups utilizing project-based learning for software development in a team-based format. It involves work teams that replicate the roles found in software development. The experiment was conducted with two groups of Software Engineering students over a semester. The results assess the performance of both groups and highlight the generic and specific competencies of the career that are enhanced by applying project-based learning. This study aims to identify the characteristics that groups undertaking project-based learning should possess for effective team-based software development.
This document presents a road map proposal to integrate the scientific community in Latin America, for sharing computational resources and implementing climatic applications to forecast events related to global warming in the region. Climatic applications are commonly developed in distributed and parallel architectures because of the amount of data to process. As an optimization procedure, we are also proposing implementing Artificial Intelligence methods as the evaluation of the application performance as the improvement of the results of climatic simulations. An advantage of using this kind of algorithm is based on the possibility of identifying relationships between different weather events, with the development of models that integrate different types of phenomena, including extreme events. Climate forecasting models use various data sources, mainly collecting images and time series. A climate prediction model can be implemented with a network of remote sensors, and it involves a geographical distribution that could be extrapolated to processing tasks. The scientific community in Latin America can share computing resources and has the talent to develop the implementation of these algorithms. Moreover, we invite the scientific community members to join us to face this challenge.
This work focused on the evaluation of some machine learning – ML models and their application in e-health within Internet of Things – IoT platforms used for the detection of seizures or epileptic episodes. The evaluation was based on two groups of metrics; the first group consists of statistical validation and the second group sought to measure the cost and computational complexity of the models; the two groups of metrics were applied in the training and validation phases. The results obtained can determine relevant factors for the selection of ML models, either based on the statistical and intrinsic efficiency of ML models, or on their suitability to be implemented in IoT under the criteria of cost and computational complexity, which are directly related to their energy consumption. The evaluation scenario was defined under an architecture with Edge, Fog and Cloud – EFC layers, where the models were implemented, initially in the cloud layer, then in the fog and edge layer. The results highlight that GBC and XGBC models present better performance when run from the cloud; LR, NB and SNN models can be trained from fog nodes and, finally, SLR and MLP can be implemented and used from edge nodes. MLP especially presents a good balance between low computational cost and high accuracy in seizure detection.
High-Performance Computing – HPC platforms based in post-Moore architectures integrate multiple specific-purpose chips in heterogenous architectures, focusing on embedded and low-power computational infrastructures – i.e., RISC architecture processors – to improve energy efficiency and low cost. This characteristic allows us to consider a certain search for computing sustainability. Currently, these platforms are very useful to implement from miniclusters to edge/cloud exchange nodes because they present an adequate relationship in terms of performance, scalability, and sustainability in addition to their low cost. This work discusses the performance and sustainability of deploying system operative images and applications in embedded post-Moore architectures addressed to HPC sustainable platforms.
Although heterogeneous systems based on hardware accelerators are a trending topic in the HPC community, exploring the trade-offs of reconfigurable hardware-based ones in linear algebra libraries for high-performance systems, has not been deeply studied. Therefore, in this research, we aim to take advantage of FPGAs' reconfigurability, adaptability, and capacity to reduce power consumption to generate FPGA-based kernels in Ginkgo, a specialized high-performance linear algebra library for many-core systems. We generated 3 FPGA-based kernels for the CSR, SELLP, and SELL SpMV formats, and obtained speedups of at least 10x concerning CPU-based kernels. Furthermore, we demonstrated via a performance characterization study that FPGAs outperform general-purpose processors in terms of compute time.
Costa Rica’s River systems play a pivotal role in providing valuable resources for society. Given that the country is exposed to a water dense, tropical climate, it is crucial to assess flooding risk and plan for extreme events. In this paper, the first step towards a river simulation pipeline is established using the Reventazón River as case study. To achieve this, an HPC portable shallow water equation solver was implemented. Boundary and initial conditions were set using QGIS and Python, and a simple Manning model was considered for friction. No rainfall, infiltration, or subsurface modeling was implemented in this work. The simulation yielded good results qualitatively on water flow for the whole Reventazón River. More complex simulations are enabled with these results given that an initial condition for water flow in the river was established.
This preliminary study describes the renewable energy production of the Fenicia district in the city of Bogota, Colombia. Based on urban distribution and population, it focuses on electrical production through solar panels and biogas production from waste recycling through methanation and its conversion into electricity. The objective of studying this energy mix is to size and design a micro datacenter for the district, following the principles of a circular economy and so powered exclusively by locally produced renewable energy. The study reveals an electricity production preference towards solar panels by an order of magnitude greater than methanation, although the interest in methanation remains significant due to its much lower fluctuations.
Given the critical nature of their missions, space systems such as satellites, probes, spacecraft, etc., are commonly embedded with specific hardware and software solutions. While this approach has led to numerous achievements, it has also limited the available capacities of a spacecraft and the potential integration between multiple units. In this work, we propose the concept of Space Mission as a Service (SMaaS), a set of strategies for deploying space systems capable of general-purpose computing, embedded Artificial Intelligence, user transparency, and flexibility towards the integration between multiple spacecraft. Such strategies will include evaluating standard operating systems and embedded companion computers, such as the NVIDIA® Jetson Series, under space conditions, common AI frameworks, High-Performance Embedded Computing, and cloud computing as an integrator between space computing devices and earth ground stations. As a demonstration, we intend to evaluate such strategies within the scope of a project consisting of devising the computing on-board system of a nanosatellite belonging to the Colombian Air Force. The possibilities are endless if a spacecraft were embedded with a companion computer with the necessary hardware and software components to execute general-purpose computing and artificial intelligence software. On-board data preprocessing, optimized space-earth download bandwidth, vision-based navigation, autonomous collision avoidance, and overall higher levels of autonomy are a few examples of the potential of this approach that could lead to the implementation of a supercomputer in space.
Evaluation and monitoring of vegetation in urban areas is used for the management of natural resources and urban planning. This information has become more important than ever due to climate change. This work proposes the use of drones or Unmanned Aerial Vehicle (UAV) to make these vegetation inventory and process the acquired images using the photogrammetry software OpenDroneMap (ODM) in a high performance computer under Singularity and Snakemake. These tools can generate a big image map of the interested zone which can be used to study the future condition and health of vegetation.
Deep learning (DL) has advanced computer vision, delivering impressive performance on intricate visual tasks. Yet, the need for accurate uncertainty estimations, particularly for out-of-distribution (OOD) inputs, persists. Our research evaluates uncertainty in Convolutional Neural Networks (CNN) and Vision Transformers (ViT) using the MNIST and ImageNet-1K datasets. Using High-Performance (HPC) platforms, including the traditional Polaris supercomputer and AI accelerators like Cerebras CS-2 and SambaNova DataScale, we assessed the computational merits and bottlenecks of each platform. This paper delineates key considerations for using HPC in uncertainty estimations in DL, offering insights that guide the integration of algorithms and hardware for robust DL applications, especially in computer vision.
In today’s era of exascale machines, energy efficiency is more crucial than ever. This study explores the potential of V-nets, initially tested on small-scale machines, to be scaled up for larger systems that support parallelism. By capturing real-time data as sequences of discrete events, this project investigates how V-nets can effectively analyze these event sequences to diagnose system behavior in High-Performance Computing (HPC) systems. The focus is on constructing temporal patterns to assess the energy performance of scalable computing systems. While no specific system is tested, the analysis emphasizes the significance of this innovative formalism. It showcases V-nets ability to identify simultaneous event occurrences, detect partial sequences, and mitigate false positives. This research aims to bridge the gap between theoretical analysis and practical implementation in Industry 4.0, ultimately advancing the optimization of scalable computing systems.
