Latin American Journal of Computing
Not a member yet
191 research outputs found
Sort by
Green software development using Carbon-Aware Scheduling techniques and energy efficiency metrics throughout the SDLC
Get PDFThe objective of the present paper is to systematize contemporary approaches of green software development through the prism of carbon-aware scheduling methodologies and energy efficiency metrics at all stages of the software development life cycle (SDLC). The study will analyze English-language, peer-reviewed articles published between 2020 and 2025. The following four carbon-intensive scheduling strategies have been identified: temporal task shifting, geographic load migration, electricity price consideration, and dynamic resource scaling. Experimental data indicates the potential for a 30–70% reduction in the carbon footprint of applications, with only a moderate impact on latency and cost. The metrics employed for evaluating energy efficiency span from low-level measures such as code complexity and measured power consumption to higher-level metrics addressing infrastructure and integration. It has been established that disregarding the initial phases of the SDLC results in an underestimation of the aggregate carbon footprint. The analysis showed that cutting emissions can conflict with maintaining high service quality. It also highlighted problems with standardizing metrics and ensuring accurate carbon-intensity forecasts, especially when significant task shifting is involved. Further unification of metrics, integration of energy monitoring at all stages of the SDLC, and consideration of economic factors are recommended.
Aplicación de Redes Neuronales Convolucionales en la Detección Automática de Melanoma Cutáneo
Get PDFEarly diagnosis of melanoma is crucial for improving survival rates, which has driven the development of deep learning models for its automated detection. This research aims to evaluate the performance of a convolutional neural network (CNN) in classifying dermoscopic images of skin lesions, comparing its accuracy with that of dermatology experts. To achieve this, a CNN was trained using a set of images that were preprocessed to improve the generalization ability of the model. The evaluation was carried out by means of quality metrics such as accuracy, precision, sensitivity, and F1-score. In addition, the ROC curve and confusion matrix were used to analyze the balance between false positives and false negatives in the classification. The results showed that the CNN outperformed dermatologists in terms of specificity and sensitivity, with an area under the curve (AUC) close to 1, indicating high discriminatory power. The confusion matrix revealed that the classification was correct in most cases, minimizing type I and type II errors. In conclusion, the implementation of neural networks in melanoma diagnosis represents a promising tool for medical care. However, opportunities for improvement were identified, such as adjusting decision thresholds and optimizing image preprocessing, which will increase the accuracy of the model in future clinical applications.El diagnóstico temprano del melanoma es crucial para mejorar la tasa de supervivencia, lo que ha impulsado el desarrollo de modelos de aprendizaje profundo para su detección automatizada. Esta investigación tiene como objetivo evaluar el rendimiento de una red neuronal convolucional (CNN) en la clasificación de imágenes dermoscópicas de lesiones en la piel, comparando su precisión con la de expertos en dermatología. Para lograr esto, se entrenó una CNN utilizando un conjunto de imágenes que fueron preprocesadas para mejorar la capacidad de generalización del modelo. La evaluación se llevó a cabo mediante el uso de métricas de calidad como exactitud, precisión, sensibilidad y F1-score. Además, se utilizó la curva ROC y la matriz de confusión para analizar el equilibrio entre los falsos positivos y falsos negativos en la clasificación. Los resultados mostraron que la CNN superó el rendimiento de los dermatólogos en términos de especificidad y sensibilidad, con un área bajo la curva (AUC) cercana a 1, lo que indica una gran capacidad discriminativa. La matriz de confusión reveló que la clasificación fue correcta en la mayoría de los casos, minimizando los errores de tipo I y II. En conclusión, la implementación de redes neuronales en el diagnóstico de melanoma representa una herramienta prometedora para la asistencia médica. No obstante, se identificaron oportunidades de mejora, como el ajuste de umbrales de decisión y la optimización del preprocesamiento de imágenes, lo que permitirá incrementar la precisión del modelo en aplicaciones clínicas futuras
Malware Detection with CNNs on Entropy and Greyscale Images
Get PDFThis study investigates whether convolutional neural networks (CNNs) trained on visual representations of Portable Executable (PE) files can rival traditional machine learning classifiers trained on engineered features. A dataset of over 200,000 PE files [1] was used to derive two feature sets (Basic and Ember-Lite) [2] and to generate 256x256 greyscale and entropy images [3],[4]. Three CNNs (SimpleCNN, ResNet-18 [5], EfficientNet-B0 [6]) were trained and evaluated against five baselines (Random Forest, XGBoost [7], CatBoost [8], LightGBM, Logistic Regression). Tree-based models with enriched features achieved the highest scores, with CatBoost reaching a ROC-AUC of 0.990. The best CNN, EfficientNet-B0 on entropy images, obtained a ROC-AUC of 0.954. Although CNNs did not surpass feature-based models, they showed competitive results when feature engineering was constrained. These findings indicate that visual approaches offer a promising alternative for static malware detection, particularly when combined with entropy-based representations [9]
La ecuación bio-calor de Pennes con derivada fraccionaria de Caputo aplicada al tratamiento térmico del cáncer ductal de mama
Get PDFThis article examines the Pennes bioheat equation in both its classical form and its extension using the Caputo fractional derivative to model tumor heating through magnetic hyperthermia with SPIONs. In the classical model (α = 1.0), simulations reach and maintain temperatures above 42 °C, consistent with the clinical and experimental results of Caizer et al., where nanoparticles raise and stabilize tissue within the therapeutic range. When incorporating the fractional derivative (α < 1.0), thermal memory effects emerge, allowing a more realistic description of tissue dynamics. Although the explicit L1 method exhibits numerical instability, the implicit L1 method provides stable and physically coherent solutions, showing slower and more localized heating for fractional orders, as expected in tissues with delayed diffusion. These fractional results computationally correspond to the three-dimensional simulations of Rahpeima & Lin, which report non-monotonic temperature patterns and diffusion dependent on SPION concentration. Overall, the implicit L1 method validates both the experimental behavior observed by Caizer and the numerical dynamics reported by Rahpeima & Lin, demonstrating that the fractional approach is promising for modeling tumor hyperthermia when stable numerical schemes are employed.Este artículo examina la ecuación bio-calor de Pennes tanto en su forma clásica como en su extensión mediante la derivada fraccionaria de Caputo para modelar el calentamiento tumoral mediante hipertermia magnética con SPIONs. En el modelo clásico (α = 1.0), las simulaciones alcanzan y mantienen temperaturas superiores a 42 °C, en concordancia con los resultados clínicos y experimentales de Caizer et al., donde las nanopartículas elevan y estabilizan el tejido dentro del rango terapéutico. Al incorporar la derivada fraccionaria (α < 1.0), emergen efectos de memoria térmica que permiten una descripción más realista de la dinámica del tejido. Aunque el método L1 explícito presenta inestabilidad numérica, el método L1 implícito proporciona soluciones estables y físicamente coherentes, mostrando un calentamiento más lento y localizado para órdenes fraccionarios, como se espera en tejidos con difusión retardada. Estos resultados fraccionarios corresponden computacionalmente a las simulaciones tridimensionales de Rahpeima & Lin, quienes reportan patrones de temperatura no monótonos y una difusión dependiente de la concentración de SPIONs. En conjunto, el método L1 implícito valida tanto el comportamiento experimental observado por Caizer como la dinámica numérica reportada por Rahpeima & Lin, demostrando que el enfoque fraccionario es prometedor para modelar la hipertermia tumoral cuando se emplean esquemas numéricos estables
Editorial
Get PDFWith the publication of Volume 13, Number 1, the Latin-American Journal of Computing (LAJC) continues to consolidate its role as a regional forum for rigorous and relevant research in computing. This issue brings together eight articles that reflect the evolving landscape of the computing discipline, where methodological soundness, practical relevance, and contextual awareness converge.The works included in this volume address a broad range of problems that arise at the intersection of computation, data, and real-world applications. Several contributions focus on software-centric solutions, proposing architectures, tools, and evaluation approaches aimed at improving system performance, reliability, and usability. These studies are grounded in concrete use cases, offering insights that are directly transferable to professional and academic settings.Another set of articles emphasizes data processing and intelligent analysis, illustrating how computational techniques can be applied to extract meaning from complex data sources. Through experimental studies and applied methodologies, these contributions highlight the growing importance of analytics and intelligent systems in supporting decision-making across diverse domains.This issue also features research related to infrastructure, security, and technological adoption, where authors examine current challenges and propose solutions that respond to the demands of modern digital environments. These works are particularly relevant in contexts where scalability, trust, and efficient resource management are critical concerns.Taken together, the articles in this issue illustrate not only technical progress, but also a sustained effort to align computing research with societal, institutional, and regional needs. The diversity of perspectives and approaches presented here underscores the dynamic nature of the field and the value of interdisciplinary and context-aware research.We would like to express our sincere appreciation to the authors for their contributions and to the reviewers for their careful and constructive evaluations, which are fundamental to maintaining the scientific quality of the journal. We invite readers to engage with the articles in this issue and trust that they will find them informative and inspiring for future research directions.Con la publicación del Volumen 13, Número 1, la revista Latin-American Journal of Computing (LAJC) continúa consolidándose como un espacio regional para la difusión de investigación rigurosa y pertinente en el área de la computación. Este número reúne ocho artículos que reflejan la evolución constante de esta disciplina, en la que convergen la solidez metodológica, la relevancia práctica y la atención al contexto.Los trabajos incluidos en este volumen abordan una amplia variedad de problemáticas que surgen en la intersección entre la computación, los datos y las aplicaciones del mundo real. Varias contribuciones se centran en soluciones orientadas al software, proponiendo arquitecturas, herramientas y enfoques de evaluación destinados a mejorar el rendimiento, la confiabilidad y la usabilidad de los sistemas. Estos estudios se sustentan en casos de uso concretos, ofreciendo aportes directamente transferibles a entornos académicos y profesionales.Otro conjunto de artículos pone énfasis en el procesamiento de datos y el análisis inteligente, mostrando cómo las técnicas computacionales pueden aplicarse para extraer conocimiento a partir de fuentes de datos complejas. Mediante estudios experimentales y metodologías aplicadas, estos trabajos evidencian la creciente importancia de la analítica y los sistemas inteligentes como apoyo a la toma de decisiones en diversos dominios.Este número también incluye investigaciones relacionadas con infraestructura, seguridad y adopción tecnológica, en las que se analizan desafíos actuales y se proponen soluciones acordes a las exigencias de los entornos digitales modernos. Estas contribuciones resultan especialmente relevantes en contextos donde la escalabilidad, la confianza y la gestión eficiente de los recursos constituyen preocupaciones clave.En conjunto, los artículos publicados en este número no solo evidencian avances técnicos, sino también un esfuerzo sostenido por alinear la investigación en computación con las necesidades sociales, institucionales y regionales. La diversidad de perspectivas y enfoques presentados resalta el carácter dinámico del campo y el valor de la investigacióninterdisciplinaria y contextualizada.Expresamos nuestro sincero agradecimiento a los autores por compartir sus trabajos y a los revisores por sus evaluaciones cuidadosas y constructivas, fundamentales para mantener la calidad científica de la revista. Invitamos a los lectores a explorar los artículos de este número y confiamos en que encontrarán en ellos aportes valiosos e inspiradores para futuras líneas de investigación
Synthesizing the Future of AI-Blockchain Integration: A Pathway for Adaptive, Ethical, and Efficiency.
Get PDFThis study systematically examines the transformative role of Artificial Intelligence (AI) in addressing the persistent challenges of blockchain technology across protocols, smart contracts, and distributed ledger management. Although blockchain offers decentralization, immutability, and transparency, its broader adoption remains constrained by scalability limitations, security vulnerabilities, inefficient consensus mechanisms, and the complexity of contract design and auditing. The findings of this review demonstrate that AI provides promising solutions to these barriers. Reinforcement learning (RL) applied to Proof-of-Stake reduced consensus latency by 30-50%, while NLP-based smart contracts lowered vulnerabilities by up to 40%, though both approaches introduced new concerns related to energy overheads and auditability. In addition, intelligent algorithms enhance ledger efficiency and data analytics, supporting more scalable and secure transaction processing. Drawing on 28 peer-reviewed studies published between 2018 and 2024, and guided by the PRISMA 2020 framework, this paper synthesizes state-of-the-art research, maps sector-specific applications in finance, healthcare, and supply chain management, and highlights unresolved gaps in ethics, reproducibility, and regulatory compliance. Notably, only 12% of the reviewed studies validated their approaches on live networks underscoring the gap between simulation-driven research and real-world deployment. The discussion culminates in the AI–Blockchain Interaction Model (AIBIM), a conceptual framework that systematizes synergies across consensus, contract, and application layers. By integrating empirical insights with critical evaluation, this work emphasizes the interdisciplinary nature of AI–blockchain research and provides actionable directions for advancing decentralized, scalable, and ethically aligned systems. This synthesis provides actionable insights for developers, regulators, and researchers in deploying AI-blockchain systems across finance, healthcare, and supply chains
Desarrollo ágil y evaluación de usabilidad de un prototipo de aplicación educativa para fomentar el alfabetismo tradicional y digital
Get PDFTraditional and digital illiteracy continue to hinder social integration and equitable access to educational and employment opportunities. In response, PixelABC was developed as an interactive application prototype based on Windows Forms, designed to strengthen basic literacy and digital skills. The development process followed the agile Scrum methodology and was guided by a Systematic Mapping Study (SMS) to identify technological strategies in vulnerable contexts. PixelABC integrates educational games, thematic modules, videos, and quizzes to facilitate interactive learning. Usability was evaluated through structured interviews with users, which helped identify key areas for improvement in interface design, instructional clarity, and system performance. Results highlight the potential of the prototype to promote educational inclusion, although enhancements are needed in visual design, loading speed, and adaptability. This study concludes that PixelABC is a viable tool for fostering traditional and digital literacy. Future improvements will focus on interface optimization, integration of multimedia resources, and adaptation to mobile platforms, thereby increasing its reach and effectiveness in reducing digital divides.El analfabetismo tanto tradicional y digital continúa limitando la integración social y el acceso equitativo a oportunidades educativas y laborales. En respuesta, se desarrolló PixelABC, un prototipo de aplicación interactiva basado en Windows Forms, orientado a fortalecer habilidades básicas de lectoescritura y competencias digitales. El desarrollo se estructuró utilizando la metodología ágil Scrum, y su diseño se fundamentó en un mapeo sistemático de literatura (SMS) que identificó estrategias tecnológicas en contextos vulnerables. PixelABC integra recursos como juegos educativos, módulos temáticos, videos y cuestionarios, facilitando el aprendizaje interactivo. La evaluación de usabilidad se realizó mediante entrevistas estructuradas con usuarios, lo que permitió identificar aspectos clave para mejorar la interfaz, la claridad de instrucciones y el rendimiento del sistema. Los resultados destacan el potencial del prototipo para promover la inclusión educativa, aunque se identificaron mejoras necesarias en diseño visual, velocidad de carga y adaptabilidad. Este trabajo concluye que PixelABC es una herramienta viable para fomentar el alfabetismo tradicional y digital. Se proyecta su evolución mediante optimizaciones de interfaz, inclusión de recursos multimedia y adaptación a plataformas móviles, lo cual incrementará su alcance y efectividad en la reducción de brechas digitales
Evaluating the accuracy of manual classification in satellite images using supervised algorithms
Get PDFThis research evaluated manual land-cover classification using images extracted from Sentinel-2. Supervised algorithms were applied to validate and enhance this process. Three algorithms were selected based on their computational efficiency: K-Nearest Neighbors (KNN), Random Forest (RF), and Support Vector Machine (SVM).The results show that KNN achieved optimal performance, demonstrating a strong balance between accuracy, F1 score, and runtime compared to RF. RF, in turn, obtained higher precision and F1 scores, indicating its superior ability to correctly identify classes; however, it required greater computational resources. SVM exhibited lower performance in the evaluated metrics but achieved a shorter runtime. Nevertheless, it was identified as the algorithm with the greatest limitations in separating classes within this dataset derived from the different study areas.Overall, the comparison confirmed that the manual classifications developed in QGIS are supported and validated by the application of these supervised methods. The use of such algorithms contributes to improving accuracy, consistency, and efficiency in geospatial classification tasks
Sentiment and Linguistic Analysis of Epidemic Outbreak Data from Official and Alternative Sources
Get PDFInformation on epidemic outbreaks is a key input for health surveillance, as it allows for the assessment of the spread and associated social perception. This study examines emotional and linguistic patterns in narratives disseminated by international organizations (WHO, UN, CDC) and digital platforms (Google News and Reddit) over a three-month period. The KDD process was applied in R Studio (selection, preprocessing, transformation, modeling, and evaluation), using Bing and NRC lexicons and a supervised Naive Bayes model to enhance the detection of emotional nuances. A total of 12,340 texts (3,100 from official sources, 4,240 from Google News, and 5,000 from Reddit) were analyzed using standardized queries in English: pandemic, confinement, epidemic, and HMPV. Official sources showed a greater presence of positive emotions linked to cooperation and security; Google News concentrated negative narratives with terms such as risk and dangerous; Reddit combined fear and sadness with appearances of hope. The analysis included t-tests and ANOVA with 95% confidence intervals. The work is exploratory and preliminary in nature and suggests that surveillance systems should integrate the monitoring of social networks and digital media, along with public policy measures to improve communication in health crisis situations
Cuando la Luz Encuentra el Sonido: Análisis de Señales de Agujeros Negros
Get PDFWhen light meets sound, a new dimension of analysis unfolds. This work explores black hole observations through the lens of signal theory and acoustic wave mechanics, revealing a resonant bridge between electromagnetic and mechanical waves. Using Event Horizon Telescope EHT data, black hole imagery is treated as a three-dimensional digital signal, where the analytic Hilbert envelope and normalized Discrete Fourier Transform DFT expose hidden structures.
The gravitational shadow is interpreted not as silence, but as a measurable energy dip—an imprint of absorption rather than absence. Euler’s identity is employed to map signal phase and symmetry into polar and complex domains, providing an intuitive mathematical pathway toward the event horizon.
By applying foundational acoustic concepts such as resonance, interference, and entropy, the field surrounding the black hole is reinterpreted as a complex communication signal. This interdisciplinary framework unifies digital signal processing, electromagnetic theory, and acoustics into a novel methodology for astronomical analysis. Notably, when a full noise assessment is conducted, EHT images exhibit a significant enhancement in resolution and information transmissionCuando la luz se encuentra con el sonido, emerge una nueva dimensión de análisis. Este trabajo examina las observaciones de agujeros negros a través de la teoría de señales y la mecánica de ondas acústicas. Utilizando datos del Telescopio del Horizonte de Sucesos EHT, las imágenes de agujeros negros se tratan como señales digitales tridimensionales, donde la envolvente analítica de Hilbert y la Transformada Discreta de Fourier (DFT, sigla en inglés) normalizada revelan estructuras y simetrías ocultas.
La sombra gravitacional se interpreta no como silencio, sino como una caída medible de energía—una huella de absorción en lugar de una simple ausencia. La identidad de Euler se emplea para mapear la fase y la simetría de la señal en planos polares y complejos, ofreciendo un camino matemático intuitivo hacia el horizonte de eventos.
Al aplicar conceptos acústicos fundamentales como la resonancia, la interferencia y la entropía, el campo que rodea al agujero negro se convierte en una señal comunicativa. Este enfoque interdisciplinario unifica el procesamiento digital de señales, la teoría electromagnética y la acústica en una metodología innovadora para el análisis astronómico. Cabe destacar que, al realizar una evaluación completa del ruido, se logra una mejora significativa en la resolución y transmisión de información de las imágenes publicadas por el EHT