Área de Enxeñaría e Arquitectura

Permanent URI for this collectionhttps://hdl.handle.net/10347/2292

Browse

Now showing 1 - 20 of 410

La infraestructura verde como instrumento de transformación territorial en los sistemas socioecológicos contemporáneos
(2026) Losada Iglesias, Rocío; Díaz Varela, Emilio Rafael; García Lamparte, Andrés Manuel; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS); Miranda Barrós, David
La Infraestructura Verde es una herramienta de planificación territorial fundamental para la protección de la biodiversidad y la adaptación al cambio climático de manera integrada y sostenible a largo plazo. Sin embargo, su planificación efectiva enfrenta barreras conceptuales, técnicas y sociales. Esta tesis desarrolla tres metodologías orientadas a transformar sus principios fundamentales, conectividad ecológica, multifuncionalidad y gobernanza, en herramientas prácticas que apoyan la toma de decisiones y la delimitación espacial de sus elementos, para garantizar su integración real. Los resultados subrayan la necesidad de combinar la optimización de la información disponible, el uso de modelos de apoyo y la implementación de una gobernanza inclusiva para alcanzar una planificación territorial transformadora.
Distributed quantum computing integrated into high-performance computing environments
(2026) Cardama Santiago, Francisco Javier; Fernández Pena, Anselmo Tomás; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
High-Performance Computing (HPC) has historically served as the primary engine for scientific advancement, enabling complex simulations ranging from meteorological prediction to protein folding or drug discovery. For decades, the exponential growth in computational power was sustained by the empirical axioms of Moore’s Law and Dennard scaling. However, in the last decade, these principles have begun to exhibit unambiguous signs of saturation. The fundamental physical, thermodynamic, and quantum limitations of silicon transistors have compelled the industry to abandon frequency scaling in favor of massive parallelism and architectural heterogeneity. Consequently, contemporary top-tier supercomputers are no longer homogeneous machines, but complex clusters that integrate multicore Central Processing Units (CPU) with specialized accelerators such as Graphics Processing Units (GPU) or FPGAs.
Potencial de técnicas no destructivas en la estimación de propiedades mecánicas de la madera de Quercus robur a lo largo de su cadena de transformación
(2026) Portela Barral, María; Guaita Fernández, Manuel; Gonçalves, Raquel; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
El objetivo general de esta tesis doctoral fue evaluar la viabilidad, precisión y aplicabilidad de diversas Técnicas No Destructivas (TND), incluyendo métodos acústicos (ultrasonidos, ondas de impacto, vibración), inspección visual y Correlación Digital de Imágenes (DIC), para estimar las propiedades mecánicas de la madera de Quercus robur. Para ello, el estudio cubrió de manera integral las distintas fases de procesamiento de la madera: desde el árbol en pie, pasando por la troza, hasta su transformación en productos de ingeniería. Los objetivos específicos abordaron la determinación de la fase más fiable para la estimación (árbol, troza o tabla), la influencia de factores ambientales y de configuración del ensayo, y la capacidad predictiva de las técnicas acústicas y DIC en láminas con uniones dentadas.
Machine learning algorithms and software to quantify immunohistochemical images in biomedicine
(2026) Al Tarawneh, Zakarya Abdullah Ayid; Cernadas García, Eva; Gándara Vila, Pilar; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
Oral cancer can develop from potentially malignant disorders such as oral leukoplakia. Immunohistochemical staining is a complementary technique to improve the diagnosis and prognosis of oral leukoplakia. Cell counting in these images was traditionally performed manually, which was very laborious. Accurate computer-aided cell detection in immunohistochemistry images of different tissues is essential for advancing digital pathology and enabling large-scale quantitative analysis. This thesis focus on the analysis of immunohistochemical (IHC) images of mouth tissue, developping software and proposing image analysis algorithms, in order to provide new skills to the diagnosis and prognosis of oral cancer. OralImmunoAnalyser is presented, which is a new reliable and easy-to-use software tool to estimate the oral leukoplakia from the IHC images of mouth tissues, which is available from the CiTIUS repository for research purposes.
Adaptação da Vitis vinífera às alterações climáticas na Região dos Vinhos Verdes
(2026) Silva, Simão Pedro Alves da; Cancela Barrio, Javier José; Valín Sanjiao, María Isabel; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
As alterações climáticas constituem uma ameaça crescente na viticultura atlântica, na qual se insere a Região dos Vinhos Verdes, sobretudo pelo impacto negativo que exercem na disponibilidade de água para rega. Nesse contexto complexo, que são as alterações climáticas, a implementação de estratégias de rega que visem o uso eficiente da água constitui uma abordagem essencial para mitigar os efeitos do défice hídrico e garantir a produtividade e qualidade das uvas. Para avaliar o impacto de diferentes estratégias de rega, estudou-se a produtividade da água e a resposta fisiológica da Vitis vinifera L., cv. Loureiro e cv. Alvarinho, em condições de sequeiro (sem rega), rega deficitária e rega total. Os estudos foram conduzidos em duas vinhas comerciais com solos franco-arenosos ambas caracterizadas por um clima atlântico.
Federated Learning over sets of heterogeneous devices
(2026) Fernández Criado, Marcos; Ladra González, Manuel; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
The objective of this PhD thesis is to face the issue of data heterogeneity when training a Federated Learning model across several devices, or participants. There exist lot of previous approaches that aim to solve this kind of situation. In contrast with them, this dissertation starts with a bird's eye view of the whole data heterogeneity casuistry. Performing this analysis, data heterogeneity can be rigorously classified, making specific problems more approachable. Two main branches of problems are identified: the ones caused purely by data skewness across the devices, and the ones were data evolves over time. At the same time, each of these two branches can be further split according to the kind of data probability that is affected. This split is based on mathematical properties of the data probability. After classifying the data heterogeneity in such a way, two different algorithms are proposed, each of them dedicated to improve previous state-of-the-art results in one of the main branches mentioned before. On the one hand, we present an algorithm designed for handling time-evolving sets of data. On the other hand, we display an algorithm that focuses on identifying conflicting data distributions and manages to reach an FL model that obtains groundbreaking results in this kind of situations.
Single-Shot Fourier iToF Camera with On-Chip Multifrequency Demodulation in CMOS Technology
(2025) Fayyaz Shahandashti, Seyyed Peyman; López Martínez, Paula; Heredia Conde, Miguel; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
This thesis presents a novel iToF camera that enables single-shot, on-chip multifrequency demodulation using periodic ultrashort-pulse illumination in combination with computationally efficient algorithms. By exploiting optimal Fourier-domain sensing patterns, the sensor captures the temporal response of a scene, which is typically sparse or compressible, while maintaining a practical trade-off in spatial resolution. The Fourier iToF camera was implemented in a standard 180 nm CMOS image sensor (CIS) technology and validated experimentally. The proposed sensing concept improves the accuracy and robustness of depth recovery by leveraging multifrequency measurements, enabling real-time, power-efficient 3D imaging with potential applications in robotics, the Internet of Things, autonomous systems, and beyond.
Novel Methods for Bayesian Networks Construction and Explanation using Natural Language
(2025) Babakov, Nikolay; Bugarín-Diz, Alberto; Reiter, Ehud; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
My PhD research advances Explainable AI (XAI) for Bayesian Networks (BNs) by improving their construction, inference explanation, and user trust. I conducted the first systematic review of BN reusability, revealing major gaps in reusability. To address structure learning, I developed CausalGraphBench, a benchmark evaluating Large Language Models-driven BN construction. For inference explanation, I introduced the Factor Argument framework, which enhances natural language explanations and was validated in the medical domain. Finally, in a case study, I applied LLM-driven BN construction, generated explanations for real users, and collected feedback.
A CMOS Frame Difference Event Sensor with Dynamic Range Extension
(2025) Jaklin, Marko; Brea Sánchez, Víctor Manuel; Kolb, Andreas; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
The growth of Internet of Things pushes the demand for low power imagers that are used for machine vision applications. Dynamic vision systems (DVS) are event cameras that are well suited for these applications as they offer high temporal resolution, high dynamic range (DR) and low power consumption. Another way to reduce redundant data and save power, storage, and computation is by frame differencing. Frame difference is broadly used for motion detection and feature extraction. Like DVS sensors, FD sensors outperform conventional cameras in power consumption and speed. When compared to DVS sensors, the frame difference cameras are outclassed in terms of DR and speed. We aim to narrow the gap between FD and DVS sensors by increasing the dynamic range of FD sensors by implementing a lateral overflow capacitor.
Portable heterogeneous computing on distributed HPC systems
(2025) Rodríguez Alcaraz, Silvia; López Vilariño, David; Fernández Rivera, Francisco; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
Heterogeneous hardware has been increasingly adopted at all scales of computer systems to meet the high and specific demands of current applications. Given the wide diversity of accelerators available, cross-platform programming models such as SYCL have emerged to offer a unified abstraction layer. This thesis addresses the development of portable heterogeneous applications using Intel's implementation of SYCL, Intel oneAPI, and evaluates their performance on distributed HPC systems. To this end, a microbenchmark was developed to solve iterative problems, applying dynamic workload balancing and considering CPU+GPU and CPU+FPGA configurations. Furthermore, the degree of automation provided by SYCLomatic for migrating CUDA to SYCL was examined, analysing the performance portability of native and migrated code.
Advancing Agricultural Knowledge Systems: Leveraging Ontology Matching, Query Expansion, and Synonym Substitution with Large Language Models
(2025) Arideh, Mohammad Ibrahim Ismail; Taboada Iglesias, María Jesús; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
This PhD research introduces a comprehensive framework for enhancing agricultural knowledge systems by integrating pretrained language models PLMs and large language models LLMs with advanced techniques such as ontology matching query expansion and synonym substitution The work addresses major challenges in agricultural text mining particularly the complexity of domain specific terminology and the limitations of traditional annotators Key components of the proposed framework include 1 ZeroShot Prompting for Fine Grained Annotation Utilizes AGROVOC subgraphs and AgricultureBERT to annotate texts in the domain of animal welfare without prior training data The method achieved up to 84 F1score and enables contextrich precise entity recognition 2 Query Expansion and Synonym Generation Enhances AGROVOC by automatically generating and validating multiword synonyms through hierarchical relationships and semantic filtering using AgricultureBERT
Object Detection with Minimal Supervision: From Few-Shot to Open-Vocabulary
(2025) García Fernández, Pablo; Mucientes Molina, Manuel; Cores Costa, Daniel; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
Object detection has achieved remarkable progress with deep learning, yet remains heavily reliant on large annotated datasets, fixed category vocabularies, and stable domains. This thesis explores object detection under minimal supervision, progressively relaxing these constraints across three settings. First, in few-shot detection, we propose a pseudo-label mining framework to expand training data with high-quality, diverse annotations from unlabeled images. Second, in cross-modal open-vocabulary detection, we introduce a training-free adaptation pipeline for transferring RGB-trained models to X-ray imagery, supported by a new benchmark. Finally, in the category-free scenario, we develop a caption-driven approach for discovering and grounding novel categories without predefined taxonomies. Together, these contributions advance towards adaptable, data-efficient detection systems capable of operating in diverse, open-world environments.
Deep Generative Models in Process Mining
(2025) Gamallo Fernández, Pedro; Lama Penín, Manuel; Vidal Aguiar, Juan Carlos; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
Nos últimos anos, a aprendizaxe profunda consolidouse como unha ferramenta prometedora para abordar diversas tarefas no ámbito da minería de procesos. Con todo, a maioría dos enfoques existentes teñen prestado pouca atención ao xeito en que se representan os eventos e as trazas de execución, o que limita a súa capacidade para capturar a complexidade inherente aos procesos reais. Nesta tese, exploramos o potencial dos modelos xerativos profundos, especialmente da familia dos autoencoders e as súas variantes, para aprender representacións latentes adaptadas ao contexto dos procesos, capaces de reflectir tanto as dependencias estruturais como as relacións contextuais presentes nos rexistros de eventos. Propoñemos tres arquitecturas innovadoras que permiten xerar representacións compactas e informativas, imputar valores faltantes de maneira eficiente e detectar anomalías nas execucións dos procesos.
Towards the industrial-scale valorization of food industry effluents into volatile fatty acid
(2025) Cassero Díaz, Tamara; Carballa Arcos, Marta; Mauricio Iglesias, Miguel; Castro Barros, Celia María; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
A valorización de correntes residuais e subprodutos para a obtención de recursos permite, de forma simultánea, a minimización dos residuos e a xeración de produtos con valor engadido, fomentando un sistema baseado na economía circular onde as correntes residuais se transforman en novos produtos. Este enfoque encaixa coa produción de ácidos graxos volátiles de cadea curta - AGV- a partir de correntes residuais da industria alimentaria para a súa aplicación no sector industrial. Os AGV, como o ácido acético, propiónico, butírico e valérico e os seus isómeros ramificados, presentan unha elevada demanda de mercado e poden empregarse como axentes microbianos e/ou reactivos químicos de ampla aplicación en sectores como o químico, cosmético ou alimentario, onde o ácido propiónico e o butírico se utilizan como conservantes tanto en alimentos como en pensos. Ademais, poden ser transformados en compostos de maior valor engadido, como os ácidos orgánicos de cadea media.
Antibióticos de consumo humano: adsorción-desorción en suelos y materiales residuales
(2025) Cela Dablanca, Raquel; Álvarez Rodríguez, Esperanza; Núñez Delgado, Avelino; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
El descubrimiento de los antibióticos supuso un hito en la medicina moderna, al reducir notablemente la mortalidad y morbilidad. No obstante, su uso intensivo y, a menudo, indiscriminado ha incrementado su presencia en diversos entornos. Tras su administración, pueden excretarse parcialmente sin metabolizar y alcanzar el medio a través de aguas residuales, lodos, purines o estiércoles, generando un riesgo ambiental al aplicarse sobre suelos. Allí, pueden ser absorbidos por plantas o llegar a masas de agua, facilitando su entrada en la cadena trófica y afectando la salud humana y ambiental. Además, su persistencia favorece la aparición de bacterias resistentes, un problema de escala global. Entre los antibióticos más usados en medicina humana están la amoxicilina (AMX) y la cefuroxima (CFX), dos Beta lactámicos, y la azitromicina (AZM), un macrólido.
Caracterización agromorfológica, genotípica y ecogeográfica de poblaciones silvestres y variedades locales de Phaseolus vulgaris L. del noroeste de Argentina
(2025) Ferreyra, Mariana Jimena; Mosquera Losada, María Rosa; Ron Pedreira, Antonio Miguel de; Galván, Marta Zulema; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
La judía o poroto común, Phaseolus vulgaris L., es la especie del género Phaseolus de mayor importancia a nivel mundial. Es una leguminosa anual, diploide, predominantemente autógama y de gran valor nutricional. Las provincias del Noroeste de Argentina (NOA) representan el límite sur del centro Andino de domesticación de la judía y albergan un gran número poblaciones silvestres y de variedades locales de P. vulgaris, que contienen una importante diversidad genética de la especie, y muchas de ellas se encuentran bajo el riesgo de erosión genética. Por esta razón, es importante contar con colecciones conservadas y evaluadas que puedan resguardar la mayor cantidad de diversidad posible y que se encuentre disponible para su uso.
Caracterización de novas especies de leguminosas anuais en rotacións con millo forraxeiro aproveitadas mediante ensilado para as explotacións de leite de Galicia
(2025) Valladares Alonso, Juan Enrique; Flores Calvete, Gonzalo; Resch Zafra, César; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
Esta Tese está dirixida a mellora do sistema de produción agrícola das explotacións lácteas galegas e se orienta á produción de leite mediante ensilaxes de elevada calidade nutricional incorporando novas especies de leguminosas forraxeiras anuais como cultivo invernal en rotacións de dous cultivo por ano con millo forraxeiro como cultivo de verán. Mediante ensaios en pequena parcela e en distintos ambientes se avaliou o valor nutritivo das leguminosas anuais e a súa produtividade en monocultivo ou mesturadas con raigrás, en laboratorio se comprobou o efecto de presecado e inoculantes no seu ensilado, en granxas colaboradores se verificaron os datos obtidos en pequena parcela, en gran parcela se avaliou o seu efecto na rotación con millo e por ultimo en rabaño experimental de vacas leiteiras da raza Holstein se estudou o seu efecto na introdución da dieta sobre a produción, composición e calidade do leite.
Efficient alignment-based conformance checking of procedural and declarative process models
(2025) Casas Ramos, Jacobo; Lama Penín, Manuel; Mucientes Molina, Manuel; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
O campo da Minaría de Procesos acadou unha atención significativa nos últimos anos debido á súa capacidade para proporcionar información valiosa sobre os procesos empresariais. Un dos principais desafíos na Minaría de Procesos é a comprobación da conformidade, que se centra en comparar o comportamento observado nos rexistros de eventos cun modelo de proceso, co fin de identificar e avaliar desviacións e incongruencias. Entre as técnicas máis empregadas e que ofrecen unha información máis útil para esta tarefa atópase o cálculo de aliñamentos óptimos, xa que permite relacionar o comportamento real recollido nos rexistros de eventos co comportamento esperado definido polos modelos de proceso, mesmo cando non existe unha coincidencia perfecta. A comprobación da conformidade é crucial para mellorar a eficiencia do proceso e garantir o cumprimento das regulamentacións.
Insights and trends of resources and waste in the total environment: applications to natural, industrial, and energy systems
(2025) Francisco López, Ahinara; Bello Bugallo, Pastora María; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
The environment is made up of interconnected dynamic systems that can be altered by human activities. It is also a source of resources and a waste sink. Effective management of resources and waste, developed by EU policies, supports the transition to a regenerative economy. The main objective of this thesis is to conduct a comprehensive investigation into the insights and trends related to resources and waste, considering their interactions within the total environment. Based on the lessons learned from previous and current approaches, existing strategies are analyzed with the aim of identifying areas for improvement and consolidating sustainable practices through a multidisciplinary approach that includes natural, industrial, and energy systems. Sustainable techniques are applied and and different tools and approaches are used and analysed to generate knowledge that supports policy development and contributes to the effective planning and management of these systems.
Environmental assessment and selection of indicators for urban and industrial wastewater treatment within a circular economy framework
(2025) Estévez Rivadulla, Sofía; Moreira Vilar, María Teresa; Feijoo Costa, Gumersindo; Universidade de Santiago de Compostela. Escola de Doutoramento Internacional (EDIUS)
The growing concern about climate change and resource scarcity has driven technological progress in the wastewater treatment. sector towards the recovery of wastewater constituents rather than their removal. While the implementation of valorisation practices is generally a sound strategy for promoting a more sustainable and circular sector, robust assessment methodologies are essential to validate their performance and prevent greenwashing. Therefore, the aim of this thesis was to analyse the environmental and economic sustainability of low-maturity technologies applied to various types of wastewater, treatment scales and intended objectives. Additionally, a novel methodological framework using indicators was developed for application in wastewater treatment plants.

Browse

Recent Submissions