39936 research outputs found
Sort by
Improved tangential interpolation-based multi-input multi-output modal analysis of a full aircraft
No full textIn the field of Structural Dynamics, modal analysis is the foundation of System Identification and vibration- based inspection. However, despite their widespread use, current state-of-the-art methods for extracting modal parameters from multi-input multi-output (MIMO) frequency domain data are still affected by many technical limitations. Mainly, they can be computationally cumbersome and/or negatively affected by close-in-frequency modes. The Loewner Framework (LF) was recently proposed to alleviate these problems with the limitation of working with single-input data only. This work proposes a computationally improved version of the LF, or iLF, to extract modal parameters more efficiently. Also, the proposed implementation is extended in order to handle MIMO data in the frequency domain. This new implementation is compared to state-of-the-art methods such as the frequency domain implementations of the Least Square Complex Exponential method and the Numerical Algorithm for Subspace State Space System Identification on numerical and experimental datasets. More specifically, a finite element model of a 3D Euler-Bernoulli beam is used for the baseline comparison and the noise robustness verification of the proposed MIMO iLF algorithm. Then, an experimental dataset from MIMO ground vibration tests of a trainer jet aircraft with over 91 accelerometer channels is chosen for the algorithm validation on a real-life application. Its validation is carried out with known results from a single-input multi-output dataset of the starboard wing of the same aircraft. Excellent results are achieved in terms of accuracy, robustness to noise, and computational performance by the proposed improved MIMO method, both on the numerical and the experimental datasets. The MIMO iLF MATLAB implementation is shared in the work supplementary material.The first author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work has been supported by the Madrid Government (Comunidad de Madrid - Spain) under the Multiannual Agreement with UC3M (IA_aCTRl-CM-UC3M). The second author is supported by the Centro Nazionale per la Mobilità Sostenibile (MOST – Sustainable Mobility Center), Spoke 7 (Cooperative Connected and Automated Mobility and Smart Infrastructures), Work Package 4 (Resilience of Networks, Structural Health Monitoring and Asset Management)
KPI optimization and future landscape in online advertising
No full textLa revolución digital ha transformado casi todos los aspectos de la vida moderna, redefiniendo cómo nos comunicamos, consumimos y realizamos negocios. En particular, el auge de Internet ha desencadenado una ola de innovación sin precedentes, dando lugar a una nueva generación de servicios y empresas que eran inimaginables hace solo unas décadas. Las redes sociales, las plataformas de comercio electrónico y los servicios de streaming han emergido como fuerzas dominantes en el ecosistema digital, revolucionando industrias y alterando profundamente el tejido social. Estas plataformas no solo han conectado a miles de millones de personas en todo el mundo, sino que también han creado formas novedosas para que las empresas interactúen con sus audiencias, lo que ha dado lugar al surgimiento de nuevos modelos de negocio y a la redefinición de sectores enteros.
Uno de los sectores que ha experimentado una transformación más profunda en esta era digital ha sido el marketing. Con la rápida evolución de las redes sociales y las plataformas digitales, las empresas se han visto obligadas a adaptarse rápidamente para alcanzar a sus audiencias en los entornos donde más tiempo pasan: en el mundo digital. Los canales de marketing tradicionales, como la prensa escrita, la radio y la televisión, han cedido paso a estrategias digitales que aprovechan datos en tiempo real, inteligencia artificial y análisis avanzados. El mundo digital ha introducido nuevas oportunidades para la segmentación, el compromiso y la medición, haciendo del marketing no solo una cuestión de creatividad, sino también de precisión basada en datos.
En la era digital, los servicios en línea y las redes sociales han remodelado drásticamente el panorama del marketing, dando lugar a un ecosistema publicitario digital altamente lucrativo. Desde la aparición del primer anuncio digital en 1994, este sector ha experimentado un crecimiento exponencial, impulsado por la accesibilidad generalizada a Internet, el auge de las redes sociales y los avances en tecnologías de análisis de datos. Esta evolución ha otorgado al marketing digital ventajas distintivas sobre la publicidad tradicional, como un mayor control sobre las campañas y la capacidad de optimizar los anuncios en función del comportamiento y las preferencias de los usuarios.
Hoy en día, la industria de la publicidad digital es un gigante en rápida expansión, valorado en miles de millones de dólares y alimentado por la inmensa cantidad de datos de usuarios que recopila. Empresas tecnológicas como Google, Meta y Amazon utilizan estos datos para personalizar la publicidad, optimizar la segmentación y, en última instancia, aumentar las ventas. No obstante, este enfoque centrado en los datos ha suscitado preocupaciones significativas sobre la privacidad y el uso indebido de la información personal. En respuesta, han surgido marcos regulatorios como el Reglamento General de Protección de Datos (RGPD) y la eliminación progresiva de tecnologías de seguimiento como las cookies, reflejando la creciente demanda de transparencia y protección del consumidor. Estos cambios están complicando la tarea de diseñar estrategias de marketing digital efectivas.
Esta tesis explora tres desafíos principales en el marketing digital contemporáneo: cómo medir el rendimiento de las campañas, el equilibrio entre la efectividad publicitaria y la intrusión en la experiencia del usuario, y la privacidad en un entorno sin cookies. El modelado de atribución es crucial para evaluar el retorno de la inversión (ROI) de las campañas de marketing, al identificar cómo los diferentes canales influyen en las acciones de los usuarios. La tesis mejora los modelos existentes para incrementar su precisión y eficacia. El segundo desafío se centra en equilibrar la exposición publicitaria mediante el uso del "frequency capping", que consiste en controlar el número de veces que un usuario es expuesto a un anuncio. Un control efectivo de la frecuencia puede prevenir la fatiga publicitaria, que se produce cuando una sobreexposición a anuncios reduce el compromiso del usuario y afecta negativamente la percepción de la marca. El tercer desafío aborda las preocupaciones sobre la privacidad, con un enfoque particular en la API de Topics de Google, una herramienta diseñada para proteger la privacidad de los usuarios, al tiempo que mantiene la efectividad de la publicidad segmentada en un mundo sin cookies.
El desafío de medir y analizar el rendimiento de las campañas digitales añade un nivel adicional de complejidad. A diferencia del marketing tradicional, los análisis digitales requieren la integración de datos de múltiples fuentes, como sitios web y plataformas de terceros, y deben adaptarse a nuevos obstáculos como el seguimiento en múltiples dispositivos y la atribución multicanal. Además, normativas estrictas de protección de datos como el RGPD y la Ley de Privacidad del Consumidor de California (CCPA) han impuesto límites sobre el alcance y la precisión de la recopilación de datos, lo que exige enfoques innovadores y creativos para resolver las brechas de datos y mantener estrategias de marketing efectivas.
La fatiga publicitaria es otro desafío importante para los anunciantes. Los especialistas en marketing deben equilibrar cuidadosamente la precisión de la segmentación con la necesidad de limitar la repetición de anuncios, ya que una exposición excesiva puede llevar a la desafección de los usuarios o al uso de bloqueadores de anuncios. Al optimizar el "frequency capping", los anunciantes pueden asegurarse de que los usuarios estén expuestos a los anuncios sin sentirse abrumados, preservando así la efectividad de la campaña y mejorando la lealtad a la marca.
A medida que las cookies y otras tecnologías de seguimiento se vuelven obsoletas, la industria de la publicidad digital debe adoptar nuevas soluciones para mantener su rentabilidad y efectividad. La recopilación y el uso de datos personales han sido durante mucho tiempo la base de la publicidad segmentada, pero estas prácticas también han suscitado importantes preocupaciones sobre la privacidad. Con navegadores como Safari y Chrome eliminando gradualmente las cookies de terceros, los anunciantes están recurriendo a alternativas como la API de Topics de Google, que prioriza la privacidad sin dejar de permitir a los especialistas en marketing ofrecer anuncios relevantes a los usuarios.
En resumen, esta tesis ofrece una exploración en profundidad de los principales desafíos y oportunidades dentro del marketing digital moderno, proporcionando soluciones prácticas para garantizar un futuro sostenible y respetuoso con la privacidad para la industria. Al abordar cuestiones relacionadas con la atribución, la privacidad del usuario y el control de la frecuencia publicitaria, esta investigación aporta valiosas perspectivas que ayudarán a los especialistas en marketing a navegar el cambiante panorama de la publicidad digital y aumentar la rentabilidad de las campañas.The digital revolution has transformed nearly every aspect of modern life, reshaping how we communicate, consume, and conduct business. In particular, the rise of the internet has sparked an unprecedented wave of innovation, giving birth to a new breed of services and businesses that were unimaginable just a few decades ago. Social networks, e-commerce platforms, and streaming services have all emerged as dominant forces in the digital ecosystem, revolutionizing industries and altering the very fabric of society. These platforms have not only connected billions of people globally but have also created novel ways for businesses to engage with their audiences, leading to the birth of new business models and the redefinition of entire sectors.
One of the most dramatically altered sectors in this digital age has been marketing. With the rapid evolution of social networks and online platforms, businesses have had to adapt swiftly to reach their audiences where they spend most of their time: online. Traditional marketing channels, such as print, radio, and TV, have given way to digital strategies that leverage real-time data, artificial intelligence, and advanced analytics. The digital world has introduced new opportunities for targeting, engagement, and measurement, making marketing not just about creativity, but also about data-driven precision.
In the digital age, online services and social networks have drastically reshaped the marketing landscape, leading to the emergence of a highly lucrative online advertising ecosystem. Since the first digital ad was displayed in 1994, this sector has experienced exponential growth, driven by the widespread of internet accessibility, the rise of social media, and advancements in data analytics technologies. This evolution has granted digital marketing distinct advantages over traditional advertising, including more precise control over campaigns and the ability to optimize ads based on user behaviour and preferences.
Today, the digital advertising industry is a rapidly growing, multi-billion-dollar behemoth, driven by the vast amounts of user data it collects. Tech giants like Google, Meta, and Amazon use this data to personalize advertising, optimize targeting, and ultimately boost sales. However, this data-centric approach has raised significant concerns about privacy and the exploitation of personal information. In response, regulatory frameworks such as the General Data Protection Regulation (GDPR) and the phasing out of tracking technologies like cookies have emerged, reflecting the growing demand for transparency and consumer protection. These shifts are complicating the task of crafting effective digital marketing strategies.
This thesis explores three major challenges in contemporary digital marketing: how to measure campaigns performance, the balance between ad effectiveness and user intrusion, and privacy in a cookieless environment. Attribution modelling is crucial for evaluating marketing campaigns’ return on investment (ROI) by identifying how various channels influence user actions. The thesis refines existing models to improve their accuracy and efficacy. The second challenge focuses on balancing ad exposure with frequency capping, which is the practice of controlling the number of times a user is exposed to an ad. Effective frequency capping can prevent ad fatigue, where overexposure to ads leads to diminished user engagement and negative brand perceptions. The final challenge addresses privacy concerns, with a specific focus on Google’s Topics API, a tool designed to protect user privacy while maintaining the effectiveness of targeted advertising in a post-cookie world.
Further complicating digital marketing is the challenge of measuring and analyzing campaign performance. Unlike traditional marketing, digital analytics requires the integration of data from multiple sources, such as websites and third-party platforms, while adapting to new obstacles like cross-device tracking and cross-channel attribution. Additionally, stringent data privacy regulations like GDPR and the California Consumer Privacy Act (CCPA) have imposed limits on the scope and detail of data collection, necessitating innovative and creative approaches to address data gaps and maintain effective marketing strategies.
The issue of ad fatigue presents another significant challenge for advertisers. Marketers must carefully balance the precision of their targeting with the need to limit ad repetition, which can lead to user disengagement or the adoption of ad blockers. By optimizing frequency capping, advertisers can ensure that users are exposed to ads without being overwhelmed, thereby preserving campaign effectiveness and enhancing brand loyalty.
As cookies and other tracking technologies become obsolete, the digital advertising industry must embrace new solutions to maintain profitability and effectiveness. The collection and use of personal data have long been the foundation of targeted advertising, but these practices have also raised significant privacy concerns. With browsers such as Safari and Chrome phasing out third-party cookies, advertisers are turning to alternatives like Google’s Topics API, which prioritizes privacy while still allowing marketers to deliver relevant ads to users.
In summary, this thesis provides an in-depth exploration of the key challenges and opportunities within modern digital marketing, offering practical solutions to ensure a sustainable and privacy-compliant future for the industry. By addressing issues related to attribution, user privacy, and frequency capping, this research contributes valuable insights that will help marketers navigate the evolving landscape of online advertising and increase the profitability of the campaigns.Programa de Doctorado en Ingeniería Telemática por la Universidad Carlos III de MadridPresidente: Roberto González Sánchez.- Secretaria: Patricia Callejo Pinardo.- Vocal: Gregorio Ignacio López Lópe
Convex Risk Control with Exact Probabilities: The CVaR-Chance-Constraint Approach
No full textChance-constrained optimization (CCO) offers exact control of failure probabilities but becomes numerically prohibitive for large scenario sets. The buffered failure probability, also known as the Conditional Value-at-Risk (CVaR), is convex and therefore tractable, but it typically leads to overly conservative designs. We introduce a new formulation, the CVaR-Chance-Constraint (CVaR-CC), which preserves the probabilistic guarantee of CCO while leveraging the convex-analytic structure of the superquantile.
We develop three scalable algorithms: (i) a secant root-finding scheme that iteratively adjusts the CVaR right-hand side until the chance constraint is met; (ii) a unit-slope quasi-Newton iteration whose local convergence holds under mild assumptions; and (iii) an active-set procedure that retains only tail scenarios, shrinking the master problem and accelerating convergence for very large scenario sets. For each algorithm we establish convergence and provide explicit sufficient conditions.
Numerical experiments on illustrative examples and energy-portfolio benchmarks show that CVaR-CC attains the required reliability with objective values close to the CCO solution while solving up to an order of magnitude faster than mixed-integer state-of-the-art methods. The framework reconciles risk fidelity with computational efficiency, enabling chance-constrained design in large, data-driven applications
Padre muy amado de mi alma: A Preliminary Study of the Forms of Address in Women’s Letters from Colonial Mexico
No full textEl interés por el estudio de cartas privadas de mujeres ha crecido en los últimos años, tal y como se desprende de las investigaciones que comparan cartas de hombres y mujeres, y de los estudios centrados exclusivamente en cartas femeninas. Este trabajo se enmarca en la segunda categoría, ya que analiza un corpus de 39 cartas misivas del siglo xviii, principalmente de mujeres de Ciudad de México, encontradas en el Archivo General de la Nación. Las cartas incluyen correspondencia entre religiosas y superiores, así como entre mujeres seglares y familiares, con algunas de especial valor histórico, como la carta enviada por una monja indígena. El estudio se centra en determinar si las car-tas reflejan rasgos lingüísticos regionales o influencias del español peninsular, lo que permitirá, en estudios comparativos posteriores, identificar diferencias lingüísticas de género. Para ello, se realiza un análisis preliminar enfocado en el uso de las formas de tratamiento desde una perspectiva sociopragmática que incorpora factores geográficos, diastráticos y diafásicos.Interest in women’s private letters has grown in recent years, with studies comparing the letters of men and women and also focusing exclusively on those of women. This paper falls into the second category; it analyzes a corpus of 39 eighteenth-century letters, from women mainly in Mexico City, found in the Archivo General de la Nación. The letters include correspondence between nuns and superiors, as well as between lay women and family members, with some of special historical value, such as a letter sent by an indigenous nun. The study seeks to determine whether the letters contain regional linguistic features or influences from peninsular Spanish, and to identify linguistic gender differences. To this end, a preliminary analysis focuses on the use of the forms of address from a sociopragmatic perspective, incorporating geographic, diastratic, and diaphasic factors
Learning Through Explanation: Producing and Peer-Reviewing Videos on Electric Circuits Problem Solving
No full textContributions: This article presents the results from a teaching innovation project based on the creation of educational videos by students and their assessment through blind peer review in the context of an electric circuit course. This article also analyses the activity’s impact on learning outcomes by comparing the results of participating students with nonparticipants, as well as with results from the previous years. The study includes surveys completed by students. Background: Electric circuit courses involve a cumulative learning process that advances throughout the course. Students who do not adhere to a regular study-homework routine often struggle to maximize the benefits of their class time and are more prone to test failures. Research Questions (RQs): RQ1) Can peer assessment be relied upon as a grading method in an electrical engineering course? RQ2) Is it possible to enhance students’ study routines and improve their results by incorporating assessment activities different from partial exams, such as creating educational videos and peer review assessments? Methodology: Students create videos, which are then submitted to the designated task through the Moodle workshop tool. Subsequently, peer reviews are conducted using a rubric form. The reliability of peer review is analysed by comparing the grades assigned by students with those assigned by teachers who are introduced as incognito reviewers. Findings: The evaluation system, relying on peer assessments, demonstrated fair reliability. Participants have substantially improved their academic performance while dedicating less time to preparing for the different evaluation tests.This work was supported by the Funding for APC: Universidad Carlos III de Madrid (Agreement CRUE-Madroño 2024
Informe técnico de implementación de e-Archivo4OS
No full textVersión extendida de la presentación “Migrando a Dspace 7: retos y oportunidades desde la perspectiva de cuatro universidades españolas. Experiencia en la UC3M” conferencia organizada por REBIUN con motivo de la Open Access Week 2024. https://hdl.handle.net/10016/47036. Grabación: https://youtu.be/Pu9Qpio5lT
An Iberian Outlook: Report on Social Media Disinformation in Spanish, Portuguese and EU elections and detection tools
Get PDFIn this inaugural report for the SmartVote project, we establish the groundwork for the projects¿ intervention, aiming to provide access, knowledge and skills that allow for technology to enable autonomy and social participation in wider society. We specifically aim to tackle the issue of disinformation and its impact during electoral processes and referendums as a threat to democracy, and the role of news and journalism and technology in the process. On the technology side we will create technological resources which allow for the identification and mitigation of disinformation. The overall objective of the SmartVote project is to ensure that all people have access to, learn about, and use technology to promote their autonomy, increasing their opportunities, strengthening their rights, and encouraging their social participation. More specifically, it addresses the phenomenon of disinformation, especially during referendums and electoral periods, as it distorts the process, values, and purposes of democratic political systems
A KWS System for Edge-Computing Applications with Analog-Based Feature Extraction and Learned Step Size Quantized Classifier
Get PDFEdge-computing applications demand ultra-low-power architectures for both feature extraction and classification tasks. In this manuscript, a Keyword Spotting (KWS) system tailored for energy-constrained portable environments is proposed. A 16-channel analog filter bank is employed for audio feature extraction, followed by a digital Gated Recurrent Unit (GRU) classifier. The filter bank is behaviorally modeled, making use of second-order band-pass transfer functions, simulating the analog front-end (AFE) processing. To enable efficient deployment, the GRU classifier is trained using a Learned Step Size (LSQ) and Look-Up Table (LUT)-aware quantization method. The resulting quantized model, with 4-bit weights and 8-bit activation functions (W4A8), achieves 91.35% accuracy across 12 classes, including 10 keywords from the Google Speech Command Dataset v2 (GSCDv2), with less than 1% degradation compared to its full-precision counterpart. The model is estimated to require only 34.8 kB of memory and 62,400 multiply–accumulate (MAC) operations per inference in real-time settings. Furthermore, the robustness of the AFE against noise and analog impairments is evaluated by injecting Gaussian noise and perturbing the filter parameters (center frequency and quality factor) in the test data, respectively. The obtained results confirm a strong classification performance even under degraded circuit-level conditions, supporting the suitability of the proposed system for ultra-low-power, noise-resilient edge applications.This research was funded by program H2020-MSCA-ITN-2020, grant No. 956601, and by Agencia Estatal de Investigacion (AEI), Spain, under Grant PID2020-118804RB-I00VCO
Spanish works on subscription video-on-demand services: availability and prominence. 2025 Edition
Get PDFReport elaborated by the Audiovisual Diversity Research Group of the Universidad Carlos III de Madrid with the support of the University Institute of Spanish Cinema (IUCE-UC3M).This report analyses the presence of Spanish works in six subscription video-on-demand (SVOD) services in the Spanish market. The report focuses, in order of their arrival in Spain, on Netflix, Prime Video, Max, Apple TV+, Disney+ and SkyShowtime. In addition to the presence of Spanish works, it examines the main mechanisms used to give them prominence in the services. The availability of Spanish titles has been increasing since 2022, when this annual report began. The main conclusion of this 2025 study is that all services include Spanish works in their respective catalogues, although the quantity and characteristics of these works vary considerably. Altogether, the six catalogues offer 1,802 unique Spanish titles, including international co-productions. The figure implies an increase of 211 works compared to January 2024. Prime Video, with 1,032 titles, is the service with the largest amount of Spanish content, followed by Netflix, which offers 674 Spanish works. When these data are compared to the total size of each catalogue, SkyShowtime is the service having the highest percentage of local works, with Spanish titles accounting for 15.8% of its catalogue. Prime Video is close behind, with 15% of its catalogue made up of Spanish titles. On Netflix and Max this percentage is around 8%, while on Disney+ and Apple TV+ they represent less than 3% in each case. The consolidation of certain trends identified in the SVOD market segment regarding the content offered and the way it is marketed, make analysing the services’ interfaces increasingly complex. In this scenario, the services examined here use different mechanisms to give prominence to the Spanish works available in their catalogues. With the exception of Apple TV+, all of them have sections dedicated to Spanish works. Likewise, the search engines of each service vary in the number and type of works they return when keywords related to local works are entered, with Netflix providing the greatest possibility of finding Spanish films, series and other types of content.This report was conceived within the framework of the research project Diversity and subscription video-on-demand audiovisual services (PID2019-109639RB-I00), funded by the Spanish Ministry of Science, Innovation and Universities and the State Research Agency (MCIN/AEI/10.13039/501100011033/)
Experimental discharge analysis of a high-temperature thermal energy storage system made of alumina blocks
Get PDFThermal energy storage (TES) systems working at very high temperatures play a crucial role in the development of more efficient solar thermal power plants. Sensible heat storage in solids is the most mature TES technology. This work presents a novel lab-scale TES system made of stacked alumina blocks, which resist high temperature and thermal shock. The alumina blocks are perforated by hexagonal channels arranged as a honeycomb. With initial temperatures as high as 800 ◦C, discharge tests are conducted for different flow rates of compressed air. Discharge times range from 2 h 1 min (at 480 L/min) to 5 h 9 min (at 120 L/min). Experimental data show the temperature segregation throughout the storage media. The system pressure drops are very low, with the highest measured being 224 Pa, at 1200 L/min. Measurements are compared with results from a 1D transient model, which tends to slightly underestimate the air temperature. The lab-scale experiments demonstrate the
feasibility of the alumina TES system for integration into dispatchable high-temperature Concentrated Solar Power plants.This research was funded by the Spanish Government under the project STORESOL, reference number PID2019-109224RA-100. This work has been also supported by the Madrid Government (Comunidad de Madrid - Spain) under the Multiannual Agreement with UC3M (SOLAROPIA-CM-UC3M)