39936 research outputs found
Sort by
Fatigue Life (Limit) Analysis Through Infrared Thermography on Flax/PLA Composites with Different Reinforcement Configurations
Get PDFThis paper presents the fatigue limit of flax/PLA composites with different fiber reinforcement architectures. The configurations of the analyzed flax/PLA composites are [0°]8, [0°/90°]s, [+45°/−45°]s, [90°]4, stacking sequences, and basket weave laminates. The methods used to estimate the fatigue limit are the fitting of stress versus number of cycles data using Weibull and Basquin equations, the surface thermographic technique with bilinear and exponential models to analyze the evolution of temperature increment, and volumetric dissipated energy. According to the results found, superficial temperature and the maximum strain reached stabilization over 2000 cycles for σmax/σut < 0.7, which was used to determine cyclic stress–strain curves and the fatigue limit. The cyclic stress–strain shows a nonlinear behavior for all laminates, having a good correlation to the Ramberg–Osgood model. Furthermore, having the stabilized temperature and volumetric dissipated energy, the exponential model was used to evaluate the fatigue limit and compared to the values found by Basquin and bilinear models. The fatigue limit found by Basquin and bilinear models shows conservative values compared to the exponential models. The results also show that temperature measurement using infrared thermography is quite sensitive to the environmental temperature variation, especially at low stress applied, and finally, the comparison of these methods on different reinforcement configurations provides a guide to select a proper technique in each case.This research was funded by Ministerio de Ciencia e Innovación [AEI/10.13039/501100011033]; and FEDER, Spain, grant number [PID2020-118946RB-I00] and [TED2021-130314B-I00]
Reflections on the Potential and Risks of AI for Scientific Article Writing after the AI Endorsement by Some Scientific Publishers: Focusing on Scopus AI
Get PDFThe introduction of ChatGPT3 in 2023 disrupted the field of artificial intelligence (AI). ChatGPT uses large language models (LLMs) but has no access to copyrighted material including scientific articles and books. This review is limited by the lack of access to: (1) prior peer-reviewed articles and (2) proprietary information owned by the companies. Despite these limitations, the article reviews the use of LLMs in the publishing of scientific articles. The first use was plagiarism software. The second use by the American Psychological Association and Elsevier helped their journal editors to screen articles before their review. These two publishers have in common a large number of copyrighted journals and textbooks but, more importantly, a database of article abstracts. Elsevier is the largest of the five large publishing houses and the only one with a database of article abstracts developed to compete with the bibliometric experts of the Web of Science. The third use and most relevant, Scopus AI, was announced on 16 January 2024, by Elsevier; a version of ChatGPT-3.5 was trained using Elsevier copyrighted material written since 2013. Elsevier's description suggests to the authors that Scopus AI can write review articles or the introductions of original research articles with no human intervention. The editors of non-Elsevier journals not willing to approve the use of Scopus AI for writing scientific articles have a problem on their hands; they will need to trust that the authors who have submitted articles have not lied and have not used Scopus AI at all
A quantile neural network framework for two-stage stochastic optimization
No full textTwo-stage stochastic programming is a popular framework for optimization under uncertainty, where decision variables are split between first-stage decisions, and second-stage (or recourse) decisions, with the latter being adjusted after uncertainty is realized. These problems are often formulated using Sample Average Approximation (SAA), where uncertainty is modeled as a finite set of scenarios, resulting in a large “monolithic” problem, i.e., where the model is repeated for each scenario. The resulting models can be challenging to solve, and several problem-specific decomposition approaches have been proposed. An alternative approach is to approximate the expected second-stage objective value using a surrogate model, which can then be embedded in the first-stage problem to produce good heuristic solutions. In this work, we propose to instead model the distribution of the second-stage objective, specifically using a quantile neural network. Embedding this distributional approximation enables capturing uncertainty and is not limited to expected-value optimization, e.g., the proposed approach enables optimization of the Conditional Value at Risk (CVaR). We discuss optimization formulations for embedding the quantile neural network and demonstrate the effectiveness of the proposed framework using several computational case studies including a set of mixed-integer optimization problems.The authors gratefully acknowledge the financial support from MCIN/AEI/10.13039/501100011033, project PID2020-116694GB-I00, and the FPU grant (FPU20/00916). Funding for APC: Universidad Carlos III de Madrid (Agreement CRUE-Madroño 2025
Journalistic AI Codes of Ethics: Analyzing Academia's Contributions to their Development and Improvement
No full textJournalists are expected to make an ethical use of artificial intelligence (AI) systems in their editorial work in order to maintain journalistic principles while adapting to AI. The growing academic interest in the journalistic use of AI and its ethical limits results in research of potential value for newsrooms. The first purpose of this study is to outline emerging ethical codes and guidelines on the use of AI in newsrooms developed and made public by news organizations worldwide. Secondly, it aims at systematizing proposals and recommendations from academic sources that could potentially contribute to develop new ethical codes and guidelines on the use of AI in newsrooms and to improve the existing ones. The third purpose is to assess to what extent such proposals and recommendations appear in ethical codes and guidelines. Ultimately, this allows us to find out whether academic contributions could be better exploited by media practitioners. Documents from 84 media organizations of Europe, America, and Asia, and eight major proposals or recommendations from twelve academic papers were identified and analyzed. Results show that, at present, 40 ethical codes or recommendations around the world are accessible to the public. Proposals coming from academic sources are related to accuracy and credibility, accessibility, relevant contents, diversity, transparency and accountability, data and privacy, human factor, and interdisciplinary teams. One significant finding is that the proposals have a limited presence in media organizations' ethical codes on AI and the ones with greater presence focus on the importance of the human factor, both in editorial decision-making and in the creative part of the journalistic process. Ultimately, the results highlight the need to discuss the professional-academic divide
Working For Democracy: Poll Officers and the Turnout Gender Gap
No full textWhat factors contribute to closing the turnout gender gap after female enfranchisement? In the wake of franchise expansion, we test whether being a poll officer-and hence being exposed to election management-boosted the politicisation and mobilisation of women. In the context of the Spanish Second Republic (1931-1939), we exploit a lottery that assigned recently enfranchised women to be poll officers in the first election women were allowed to vote (1933). We use an original individual-level panel database and show that women randomly selected as polling officers were as likely to participate in subsequent elections than men, while the gender turnout gap persisted among the rest. Further analyses suggest that being poll officers made women more receptive to political organisations mobilisation strategies, and their presence had positive externalities by encouraging other women to participate. Our findings highlight the potential benefits of exposure to election engineering among groups previously excluded or less engaged with democracy.This research was supported by Universitat Pompeu Fabra, Universitat de Barcelona, and Universidad Carlos III–IC3JM
Artificial precursor for alkaline cements
Get PDFOne of the main challenges for the future development of alkaline cements is the availability of precursors. Traditional precursors (such as coal FA and the BFS) have some limitations concerning quality and quantity (the long term supply is not guaranteed). The progressive closure of coal-fired power plants and changes in steel production in many countries exacerbate the problem. The present work addresses the challenge of fabricating an artificial precursor (via thermal treatment) with a chemical composition similar to a type- C fly ash (20 % CaO and SiO2/CaO=3 and SiO2/Al2O3=3). Three temperatures of synthesis were tested: 1000 °C, 1100 °C and 1250 °C. The precursors obtained after thermal treatment of a mixture of chemicals were activated with a 8 M NaOH solution. The temperature of synthesis obviously affected the degree of vitrification. Nevertheless, it can be said that partially amorphous/vitreous precursors, were produced at 1000 °C, developing good mechanical performance. In all cases, compressive strengths above 20 MPa were obtained, after 1 day curing. In cements made with precursors synthesized at 1000 and 1100 °C (amorphous content 70 %), a (N,C)-A-S-H type gel was formed, as the main product of hydration. However, in those cements made with precursors synthesized at 1250 °C (amorphous content 99 %), a mixture of (N)-C-A-S-H and (N,C)-A-S-H gels were observed after the hydration process.This works has been funded by the Spanish Research Agency (AEI), the Spanish Ministry of Science and Innovation and the ERDF (research projects (PID2019-11464RB-100//AEI/10.13039/501100011033, PID2022-138637OB-C31/AEI/10.13039/501100011033/FEDER, UE)), "JIN Projects 2021 PID2020-116738RJ-I00//AEI/10.13039/501100011033. RYC Excellence Contract (RYC2021-032620-I, MCIN/AEI/10.13039/50110001033. The award of the FPI pre-doctoral grant (PRE2020-091909) is also acknowledged. This work has been carried out in the facilities of Instituto Eduardo Torroja de Ciencias de la Construcción (IETcc-CSIC) and the Universidad Carlos III de Madrid (UC3M)
AZTEC+: Long and Short Term Resource Provisioning for Zero-Touch Network Management
Get PDFIn the past few years, network infrastructures have transitioned from prominently hardware-based models to networks of functions, where software components provide the required functionalities with unprecedented scalability and flexibility. However, this new vision entails a completely new set of problems related to resource provisioning and the network function operation, making it difficult to manage the network function lifecycle management with traditional, human-in-the-loop approaches. Novel zero-touch management solutions promise autonomous network operation with limited human interactions. However, modeling network function behavior into compelling variables and algorithm is an aspect that such solutions must take into account. In this paper, we propose AZTEC + a data-driven solution for anticipatory resource provisioning in network slicing scenarios. By leveraging a hybrid and modular deep learning architecture, AZTEC +, not only forecasts the future demands for target services but also identifies the best trade-offs to balance the costs due to the instantiation and reconfiguration of such resources. Our experimental evaluation, based on real-world network data, shows how AZTEC +, can outperform state-of-the-art management solutions for a large set of metrics.This work has been partially supported by the ORIGAMI project, which has received funding from the Smart Networks and Services Joint Undertaking (SNS JU) under the European Union's Horizon Europe research and innovation programme with Grant Agreement No. 101139270 and by the 6G-IRONWARE project (CNS2023-143870) funded by MICIU/AEI /10.13039/501100011033 and the EU NextGenerationEU/PRTR. The work of the University Carlos III of Madrid has also been funded by the Spanish Ministry of Economic Affairs and Digital Transformation and the European Union-NextGenerationEU through the UNICO 5G I+D 6G-CLARION project, and by the TrialsNet Project, which has received funding from the SNS JU under the European Union's Horizon Europe research and innovation programme with Grant Agreement No. 10109587
Design and validation of a low-cost vehicle driver monitoring device
No full textThis paper presents the design and experimental validation of a novel low-cost, wearable device for continuous monitoring of the motion and physiological state of vehicle drivers. The system strategically integrates Inertial Measurement Units (IMUs) on the driver's head, neck, and torso to accurately capture body dynamics while driving. It also incorporates physiological sensors to capture electrocardiography (ECG) and body temperature signals, providing objective indicators of the driver's physical and emotional state. The device's architecture enables real-time data processing, which is essential for early detection of risky behaviors such as distracted or aggressive driving. Validation of the device was conducted through an experimental protocol in a controlled urban driving environment, evaluating common maneuvers such as straight-line driving, braking, roundabouts, and lane changes. The results obtained demonstrate the feasibility of the device to characterize significant changes in posture and physiological parameters, demonstrating its potential application as a low-cost tool for driver monitoring in Advanced Driver Assistance Systems (ADAS), accident prevention campaigns, and epidemiological studies of driving behavior.Funding for APC: Universidad Carlos III de Madrid (Agreement CRUE-Madroño 2025)
Reliability analysis of vehicle semi-active suspension systems under parameter uncertainties in magnetorheological dampers
No full textMagnetorheological dampers (MR) have become essential components in recent years for vibration control in dynamic systems. Their effective application relies on the calibration of models using experimental data. However, these models are often assumed to have deterministic parameters, which can negatively impact the reliability of the control system by neglecting uncertainties. In this work, we propose a methodology to assess the reliability of a vehicle semi-active suspension system under parameter uncertainties in an MR damper model. First, we calibrate the Kwok's MR damper model using Bayesian methods across a wide range of input currents, excitation frequencies, and displacements. Then, we design a procedure to identify the strongest nonlinear response of the semi-active suspension system, enabling the construction of surrogate models for uncertainty propagation without excessive computational cost. Finally, we apply this approach to estimate the probability density functions (PDFs) of the MR damper model parameters for untested input currents. This study demonstrates the application of existing uncertainty quantification techniques to improve the identification of parameter uncertainties in smart semi-active suspension systems.Maria Jesus Lopez Boada reports financial support was provided by State Agency of Research, by the grant [PID2022-136468OB-I00] funded by MCIN/AEI/10.13039/501100011033, by “ERDF A way of making Europe
Adversarial dynamics in centralized versus decentralized intelligent systems
No full textThis article is part of the topic "Building the Socio-Cognitive Architecture of COHUMAIN: Collective Human-Machine Intelligence," Cleotilde Gonzalez, Henny Admoni, Scott Brown and Anita Williams Woolley (Topic Editors).Artificial intelligence (AI) is often used to predict human behavior, thus potentially posing limitations to individuals' and collectives' freedom to act. AI's most controversial and contested applications range from targeted advertisements to crime prevention, including the suppression of civil disorder. Scholars and civil society watchdogs are discussing the oppressive dangers of AI being used by centralized institutions, like governments or private corporations. Some suggest that AI gives asymmetrical power to governments, compared to their citizens. On the other hand, civil protests often rely on distributed networks of activists without centralized leadership or planning. Civil protests create an adversarial tension between centralized and decentralized intelligence, opening the question of how distributed human networks can collectively adapt and outperform a hostile centralized AI trying to anticipate and control their activities. This paper leverages multi-agent reinforcement learning to simulate dynamics within a human-machine hybrid society. We ask how decentralized intelligent agents can collectively adapt when competing with a centralized predictive algorithm, wherein prediction involves suppressing coordination. In particular, we investigate an adversarial game between a collective of individual learners and a central predictive algorithm, each trained through deep Q-learning. We compare different predictive architectures and showcase conditions in which the adversarial nature of this dynamic pushes each intelligence to increase its behavioral complexity to outperform its counterpart. We further show that a shared predictive algorithm drives decentralized agents to align their behavior. This work sheds light on the totalitarian danger posed by AI and provides evidence that decentrally organized humans can overcome its risks by developing increasingly complex coordination strategies.Manuel Cebrian was partially supported by the Ministry of Universities of the Government of Spain, under the program "Convocatoria de Ayudas para la recualificación del sistema universitario español para 2021-2023, de la Universidad Carlos III de Madrid, de 1 de Julio de 2021."Open access funding enabled and organized by Projekt DEAL