Category
3,773 papers
FERIAL ZAYED THE WORLD'S FIRST OPEN-SOURCE DIAGNOSTIC ARTIFICIAL INTELLIGENCE FOR HUMANITY A Production-Ready Architecture for Global Health Integration
FERIAL ZAYED THE WORLD'S FIRST OPEN-SOURCE DIAGNOSTIC ARTIFICIAL INTELLIGENCE FOR HUMANITY A Production-Ready Architecture for Global Health Integration
Generative Artificial Intelligence (GAI) has rapidly transformed higher education by introducing new possibilities for personalized learning and pedagogical innovation. It has emerged as a powerful support system capable of information processing and decision-making tasks in higher education. Unlike traditional educational technologies, models such as ChatGPT enable dynamic interaction with learners. Students engage with knowledge, complete academic tasks, and regulate their performance. Building upon previous studies, this article incorporates an expanded theorical framework, and a deeper ethical analysis of student perceptions and emotions, emphasizing its role as a support tool in their learning processes. The results suggest that while ChatGPT enhances efficiency, accessibility, and learner autonomy, it also introduces significant risks related to overreliance, academic integrity, and the loss of critical thinking skills. The study concludes that should be integrated as an academic mediator, emphasizing the need for ethical guidelines and pedagogical institutional redesign. Spanish-language metadata / Metadatos en españolTítulo en español:Inteligencia Artificial Generativa: desafíos éticos e implicaciones pedagógicas en contextos educativos emergentes Resumen:La Inteligencia Artificial Generativa (IAG) ha transformado rápidamente la educación superior al introducir nuevas posibilidades para el aprendizaje personalizado y la innovación pedagógica. Se ha consolidado como un potente sistema de apoyo capaz de realizar tareas de procesamiento de información y toma de decisiones en el ámbito de la educación superior. A diferencia de las tecnologías educativas tradicionales, modelos como ChatGPT permiten una interacción dinámica con los estudiantes. Estos interactúan con el conocimiento, realizan tareas académicas y regulan su propio desempeño. A partir de estudios previos, este artículo incorpora un marco teórico ampliado y un análisis ético más profundo de las percepciones y emociones de los estudiantes, haciendo énfasis en el papel de la IAG como herramienta de apoyo en sus procesos de aprendizaje. Los resultados sugieren que, aunque ChatGPT mejora la eficiencia, la accesibilidad y la autonomía del estudiante, también introduce riesgos significativos relacionados con la dependencia excesiva, la integridad académica y la pérdida de habilidades de pensamiento crítico. El estudio concluye que la IAG debería integrarse como mediadora académica, destacando la necesidad de establecer directrices éticas y de llevar a cabo un rediseño pedagógico e institucional. Palabras Claves:Inteligencia Artificial Generativa, entorno educativo, procesos de aprendizaje Smart citations: SciteAI. Dimensions.Open Alex.
Reach audiences
Advertise in front of researchers, engineers, and readers.
Executive Smmary The tracing of communications is used to fight many kinds of crime, but its effectiveness depends on the resources that law enforcement agencies (LEAs) can dedicate to acting upon intelligence supplied by communication providers (CPs). China has extradited many thousands of scammers from overseas call centres despite not asking foreign countries to build new systems to facilitate traceback. The USA’s strategy concentrates on technology, including heavy expenditure on systems that facilitate traceback. However, the resulting number of prosecutions has been modest despite large volumes of scam calls originating on major US networks and a continued rise in the amounts lost by American consumers to scams instigated by a phone call. Europe can draw lessons from the different approaches adopted by China and the USA, including their respective strengths and limitations. In turn, Europe has much to teach the rest of the world about its own successes in tackling scams. There is some ambiguity in what different parties mean when they discuss traceback. This leads to confusion about what is required of a traceback system. Solutions are designed incorrectly if there is no clear statement of requirements. Several respondents to our survey noted that traceback essentially involves asking CPs where a communication came from, per the records they retained. Countries have been tracing communications for many years. So, when considering how to implement a new system, we should ask what we are trying to trace, where we expect to trace it to, and what is deficient with the systems to be replaced. The notion that a global traceback system stems from work on an allied problem: the authentication of communications.‘Authentication’ is a way of saying work has been done to prevent impersonation. Much of this work is imperfect. This report links the questions of how to tackle spoofing with how to trace communications. Tracing is more general in application. And if we know the origin of a communication with certainty, because it has been authenticated and the sender was subject to rigorous know-your-customer (KYC) checks, then ‘tracing’ that communication is just a peculiar way of saying we look at the sender’s address per the communication. Chapter 5 of this report evaluates 15 methods for authentication or tracing (or both). Considering all the options helps to clarify the strengths and weaknesses of each one. The effectiveness of traceback methods based on hop-by-hop tracing may be limited by criminal adaptation and cross-border implementation challenges. Hop-by-hop tracing is easy to defeat on a technical level because criminals will find alternate routes for their traffic to prevent it being traced. They have decades of experience of routing international communications traffic to evade detection. Hop-by-hop tracing can also be obstructed by laws that prevent the international exchange of relevant data. Even if all countries could be persuaded to support a global hop-by-hop method of tracing dialled voice calls, it would still be rendered redundant before it entered service because criminals would stop using dialled voice calls and start using other ways to transmit their voice.Phone users are switching to over the top services that allow them to speak and send messages to friends, family and businesses without dialling telephone numbers. So are scammers. We should be clear about whether we want to speed up the way tracing has been done in the past – by making enquiries of each CP in turn – or if LEAs want to identify the sources of crime as directly as possible. Transmitting tracing and authentication data in-band within telecoms networks means it follows the same route as the communication, and hence incurs the risk of being interrupted at any hop. Transmitting data out of band avoids those risks because it only requires cooperation between the authorities and CPs at the origin and destination of the communication. Out of band methods are less mature at present but have a more realistic prospect of eventual success. Some individual nations are setting up out of band data exchanges between their telcos without trying to standardise for international cooperation. Meanwhile, the limitations identified in current STIR/SHAKEN implementations have prompted renewed interest in alternative approaches, including out of band SHAKEN and Open Verified Communication. The EU has an opportunity to become a technological leader in this domain by connecting the threads of this work to its eIDAS roadmap for digital ID. European LEAs could look to the future by seeking general-purpose authentication and tracing systems that can also be applied to the newer generations of services that deliver voice and messaging communications through over-the-top and hybrid routes. This would involve out of band transfers of authentication and traceback data so the flow of data is independent of the specifics of the communication protocols implemented by the private sector. The benefit would be that a future-proofed tracing and authentication framework would ignore the increasingly artificial distinction between a voice call instigated using a dialled number or an app, or between text messages sent via SMS, RCS, WhatsApp and iMessage. For all the focus on novel network technologies, it is easy to overlook the fact that Europe is already a world leader in tackling spoofing. The widespread adoption of ECC Recommendation (23)03 has greatly reduced the number of inbound international calls that spoof the domestic phone numbers of Europeans.Where figures are available, it is not unusual to see this control being credited for reductions in scam calls of around 70 or 80 percent. Its effectiveness greatly alters the cost-benefit argument for other controls that require different countries to cooperate. Sender ID registries are similarly effective at reducing the most common text message scams. Some of the most effective anti-scam methods covered by this report have been put into effect in countries that have advanced anti-scam policies, such as Singapore and Australia. But Europe does not need to look to other continents for leadership. Ireland provides an example of a national strategy that has combined relatively simple measures with reported reductions in telecom-related scams. The EU is well-placed to coordinate efforts to tackle networked crime, just as it led the telecoms and tech industries through the adoption of the GSM standards for mobile telephony, the introduction of mobile roaming, and the data privacy rules enshrined in GDPR. ECC Recommendation (23)03 is an example of recent European leadership in consumer protection; eIDAS establishes a positive future for digital ID that serves the needs of EU citizens. By first concentrating on methods that obstruct the flow of scam communications, and then implementing bilateral exchanges of data with countries that want to aid the prosecution of criminals they harbour, Europe can make the most tangible gains in reducing scams. Europe can do this by favouring the authentication and tracing technologies that already prioritise the goal of consumer protection while respecting European standards for security and privacy. This report was prepared by the External Expert Erik Priezkalns, while the topic of the report was requested by Europol’s European Cybercrime Centre (EC3) to analyse existing and operational technical solutions to mitigate Caller-ID and Sender-ID spoofing. We’re collecting feedback on this report through the EU Survey Platform, if you’d like to share your thoughts please click on the link below. https://ec.europa.eu/eusurvey/runner/enact-report-feedback
Executive Smmary The tracing of communications is used to fight many kinds of crime, but its effectiveness depends on the resources that law enforcement agencies (LEAs) can dedicate to acting upon intelligence supplied by communication providers (CPs). China has extradited many thousands of scammers from overseas call centres despite not asking foreign countries to build new systems to facilitate traceback. The USA’s strategy concentrates on technology, including heavy expenditure on systems that facilitate traceback. However, the resulting number of prosecutions has been modest despite large volumes of scam calls originating on major US networks and a continued rise in the amounts lost by American consumers to scams instigated by a phone call. Europe can draw lessons from the different approaches adopted by China and the USA, including their respective strengths and limitations. In turn, Europe has much to teach the rest of the world about its own successes in tackling scams. There is some ambiguity in what different parties mean when they discuss traceback. This leads to confusion about what is required of a traceback system. Solutions are designed incorrectly if there is no clear statement of requirements. Several respondents to our survey noted that traceback essentially involves asking CPs where a communication came from, per the records they retained. Countries have been tracing communications for many years. So, when considering how to implement a new system, we should ask what we are trying to trace, where we expect to trace it to, and what is deficient with the systems to be replaced. The notion that a global traceback system stems from work on an allied problem: the authentication of communications.‘Authentication’ is a way of saying work has been done to prevent impersonation. Much of this work is imperfect. This report links the questions of how to tackle spoofing with how to trace communications. Tracing is more general in application. And if we know the origin of a communication with certainty, because it has been authenticated and the sender was subject to rigorous know-your-customer (KYC) checks, then ‘tracing’ that communication is just a peculiar way of saying we look at the sender’s address per the communication. Chapter 5 of this report evaluates 15 methods for authentication or tracing (or both). Considering all the options helps to clarify the strengths and weaknesses of each one. The effectiveness of traceback methods based on hop-by-hop tracing may be limited by criminal adaptation and cross-border implementation challenges. Hop-by-hop tracing is easy to defeat on a technical level because criminals will find alternate routes for their traffic to prevent it being traced. They have decades of experience of routing international communications traffic to evade detection. Hop-by-hop tracing can also be obstructed by laws that prevent the international exchange of relevant data. Even if all countries could be persuaded to support a global hop-by-hop method of tracing dialled voice calls, it would still be rendered redundant before it entered service because criminals would stop using dialled voice calls and start using other ways to transmit their voice.Phone users are switching to over the top services that allow them to speak and send messages to friends, family and businesses without dialling telephone numbers. So are scammers. We should be clear about whether we want to speed up the way tracing has been done in the past – by making enquiries of each CP in turn – or if LEAs want to identify the sources of crime as directly as possible. Transmitting tracing and authentication data in-band within telecoms networks means it follows the same route as the communication, and hence incurs the risk of being interrupted at any hop. Transmitting data out of band avoids those risks because it only requires cooperation between the authorities and CPs at the origin and destination of the communication. Out of band methods are less mature at present but have a more realistic prospect of eventual success. Some individual nations are setting up out of band data exchanges between their telcos without trying to standardise for international cooperation. Meanwhile, the limitations identified in current STIR/SHAKEN implementations have prompted renewed interest in alternative approaches, including out of band SHAKEN and Open Verified Communication. The EU has an opportunity to become a technological leader in this domain by connecting the threads of this work to its eIDAS roadmap for digital ID. European LEAs could look to the future by seeking general-purpose authentication and tracing systems that can also be applied to the newer generations of services that deliver voice and messaging communications through over-the-top and hybrid routes. This would involve out of band transfers of authentication and traceback data so the flow of data is independent of the specifics of the communication protocols implemented by the private sector. The benefit would be that a future-proofed tracing and authentication framework would ignore the increasingly artificial distinction between a voice call instigated using a dialled number or an app, or between text messages sent via SMS, RCS, WhatsApp and iMessage. For all the focus on novel network technologies, it is easy to overlook the fact that Europe is already a world leader in tackling spoofing. The widespread adoption of ECC Recommendation (23)03 has greatly reduced the number of inbound international calls that spoof the domestic phone numbers of Europeans.Where figures are available, it is not unusual to see this control being credited for reductions in scam calls of around 70 or 80 percent. Its effectiveness greatly alters the cost-benefit argument for other controls that require different countries to cooperate. Sender ID registries are similarly effective at reducing the most common text message scams. Some of the most effective anti-scam methods covered by this report have been put into effect in countries that have advanced anti-scam policies, such as Singapore and Australia. But Europe does not need to look to other continents for leadership. Ireland provides an example of a national strategy that has combined relatively simple measures with reported reductions in telecom-related scams. The EU is well-placed to coordinate efforts to tackle networked crime, just as it led the telecoms and tech industries through the adoption of the GSM standards for mobile telephony, the introduction of mobile roaming, and the data privacy rules enshrined in GDPR. ECC Recommendation (23)03 is an example of recent European leadership in consumer protection; eIDAS establishes a positive future for digital ID that serves the needs of EU citizens. By first concentrating on methods that obstruct the flow of scam communications, and then implementing bilateral exchanges of data with countries that want to aid the prosecution of criminals they harbour, Europe can make the most tangible gains in reducing scams. Europe can do this by favouring the authentication and tracing technologies that already prioritise the goal of consumer protection while respecting European standards for security and privacy. This report was prepared by the External Expert Erik Priezkalns, while the topic of the report was requested by Europol’s European Cybercrime Centre (EC3) to analyse existing and operational technical solutions to mitigate Caller-ID and Sender-ID spoofing. We’re collecting feedback on this report through the EU Survey Platform, if you’d like to share your thoughts please click on the link below. https://ec.europa.eu/eusurvey/runner/enact-report-feedback
The increasing share of solar photovoltaics (PV) in power grids and buildings is reshaping the energy landscape but also introducing operational challenges due to the inherent variability of solar irradiance. Rapid cloud movements can cause short-term fluctuations in PV output, making it difficult to ensure grid stability, manage power imbalances, and optimize energy use within Building Energy Management Systems (BEMS). This thesis, titled Towards Solar Nowcasting, contributes to overcoming these challenges by advancing short-termsolar forecasting techniques, with a particular emphasis on real-time, image-based forecasting also known as solar nowcasting. To this end, the research begins with a comprehensive review of solar forecasting techniques, highlighting the growing importance of Artificial Intelligence (AI) methods in capturing complex irradiance patterns across diverse time horizons, from ultra-short (1 minute) to 24 hours ahead. A particular focus is placed on the potential of neural networks and hybrid AI models, as well as the critical need for standardized datasets and benchmarking practices to ensure accurate model evaluation and performance. This review forms the foundation for the development of innovative nowcasting solutions. Building on these insights, the thesis presents a data-driven short-termsolar forecasting framework using all-sky imagers (ASIs) and deep learning. Specifically, cloud movement is tracked using optical flow models, and future sky states are generated to serve as inputs for Convolutional Neural Networks (CNNs) and Long Short-TermMemory (LSTM) networks. This hybrid approach enables accurate Global Horizontal Irradiance (GHI) predictions up to 20 minutes ahead. The developed models demonstrate significant improvements over baseline persistence methods, achieving ramp skill scores of up to 39% under sunny conditions. To address the limitations of existing methods in complex weather scenarios, the thesis further develops an innovative hybrid AI framework that combines superpixel-based cloud detection, Support Vector Machines (SVMs), CNNs, and Kalman filtering. This approach integrates high-resolution sky images, advanced computer vision techniques, and adaptive weather classification to deliver reliable GHI forecasts for horizons up to one hour. Tested on extensive datasets from the Netherlands, the method showed marked improvements in forecast accuracy, particularly under challenging conditions such as overcast or rainy skies, where conventional models often fail. Finally, the thesis translates these forecasting advancements into practical applications for congestion management, power imbalance mitigation, and building energy management. By benchmarking statistical, AI-based, and sky-imager-driven PV forecasting techniques, the study demonstrates that the integration of real-time sky image data significantly enhances short-term PV power forecasts. This, in turn, supports grid operators in implementing proactive congestion control, reduces reliance on costly balancing reserves, and enables intelligent energy management strategies within buildings, including load shifting, battery storage optimization, and increased PV self-consumption. In summary, in this thesis, advanced short-termsolar forecasting techniques have been developed to address key operational challenges arising fromthe growing integration of solar photovoltaics (PV) into modern energy systems. By combining all-sky imaging, artificial intelligence, and hybrid machine learning frameworks, this work demonstrates significant improvements in the accuracy and reliability of solar nowcasting. The proposed methodologies provide practical solutions for congestion management, power imbalance reduction, and optimized building energy management. Overall, this thesis contributes to enabling a more reliable and efficient integration of solar energy, supporting the broader goals of grid stability, energy flexibility, and the ongoing energy transition.
As cyber incidents increase in complexity, diversity and frequency, cybersecurity practitioners find it more challenging to extract meaningful threat intelligence and insights from cyber incident reports. This problem is worsened by the limited number of these reports; as cyber incident victims may withhold and/or downplaytheir reports due to reputational and privacy concerns. Therefore, this study aims to determine whether cyber incident reports, which have been stripped of personal data (pseudonymised), can be classified according to the Spanish INCIBE Cyber Incident Taxonomy. Seven transformers (SecBERT, BERT, DistilBERT, BERTweet, SecRoBERTa, ALBERT, RoBERTa) and four traditional machine learning classifiers (Random Forest, Multinomial Naïve Bayes, XGBoost, Support Vector Machine, using two encoders - Bag of Words and Term Frequency - Inverse Document Frequency), were trained to classify cyber incidents which were pseudonymised using Data Masking, Data Tokenisation and Data Substitution. The best-performing model attained an F1-score of 78.5% on the non-pseudonymised CECILIA-10C-900 dataset and 83% on the dataset, which was pseudonymised using Data Masking, D-CECILIA-10C-900-MAS. Therefore, this research demonstrates that it is possible for a single AI model to balance enhanced privacy with strong threat intelligence analysis capabilities.
As cyber incidents increase in complexity, diversity and frequency, cybersecurity practitioners find it more challenging to extract meaningful threat intelligence and insights from cyber incident reports. This problem is worsened by the limited number of these reports; as cyber incident victims may withhold and/or downplaytheir reports due to reputational and privacy concerns. Therefore, this study aims to determine whether cyber incident reports, which have been stripped of personal data (pseudonymised), can be classified according to the Spanish INCIBE Cyber Incident Taxonomy. Seven transformers (SecBERT, BERT, DistilBERT, BERTweet, SecRoBERTa, ALBERT, RoBERTa) and four traditional machine learning classifiers (Random Forest, Multinomial Naïve Bayes, XGBoost, Support Vector Machine, using two encoders - Bag of Words and Term Frequency - Inverse Document Frequency), were trained to classify cyber incidents which were pseudonymised using Data Masking, Data Tokenisation and Data Substitution. The best-performing model attained an F1-score of 78.5% on the non-pseudonymised CECILIA-10C-900 dataset and 83% on the dataset, which was pseudonymised using Data Masking, D-CECILIA-10C-900-MAS. Therefore, this research demonstrates that it is possible for a single AI model to balance enhanced privacy with strong threat intelligence analysis capabilities.
Corrected version of Chapter 13 of the monograph "A Rubik-tér: Négydimenziós bináris rács mint fizikai világ" (DOI: 10.5281/zenodo.22162465), together with the reproducible measurement code. The gauge-sector measurements in v52 are not tenable: they were produced with a globally synchronised fence flipping only in the x direction, a 3–5 cell initial condition, and a non-stationary measurement window. This affects the claimed "D = 4 optimum" and the 2.18% relative standard deviation. Main results of the corrected measurements: the matched active fence reduces the Q8 gauge group to the centraliser of i, ⟨i⟩ ≅ Z4, verified by the link histogram in every dimension; the effect of the fence is confined to a single shell with a sharp cutoff, and the bulk is a free Z4 field; the surface term depends on D but monotonically, so D = 4 is not distinguished; the stationary value depends on the relation between the initial sheet element and the fence element. Every figure reported here is reproducible with the attached kerites_modell_v53.py; the commands are listed at the end of the document. Artificial intelligence (Claude, Anthropic) assisted in verifying the measurements, identifying the errors, and editing the text.
This extension presents the formal mathematical translation of the Empathic Logic Model (ELM). Because foundational psychological mechanisms and narrative elements function as qualitative components, specific text blocks containing these elements are designated with the tag UNFORMALIZABLE AS WRITTEN. This tag applies exclusively to the specific qualitative narrative sentences it precedes, not to the section as a whole. The formalized mathematical equations, discrete logic, and computational translations are embedded directly below and between these qualitative text blocks. For the comprehensive qualitative framework, readers can reference the primary manuscript via Zenodo: https://doi.org/10.5281/zenodo.18614652 ### Note on this Python Library Scope and Evaluation: Python Library Scope and Operational Interpretation. ELM is an applied computational architecture for dynamic interactive systems, evaluated through operational execution, transition robustness, fault tolerance, and recovery, including adversarial stress testing under Full-Stack Turing Deadlock, PBFT, and ABFT-inspired conditions, rather than abstract mathematical proofs. Its architecture is not intrinsically limited to human–human interaction and may be instantiated in human–machine interaction, machine-learning pipelines, AI safety, robotics, autonomous systems, and other interactive agents, subject to implementation and empirical validation. Artificial intelligence systems were utilized in the formalization and Python translation presented herein. ### Future Translation Note: Its translation into engineering architectures remains open to domain-specific implementation, with the resulting configurations determined by the requirements and operational constraints of each application field and informed by subsequent empirical development. ### Library and Execution Note: The resulting Python implementation may be used as an ELM computational library, with application-specific execution blocks constructed by implementers according to the requirements and operational constraints of their respective systems. ### Explanation of this Library: ELM governs the interpretation process by determining what the system should do next with received input and how that input should be handled. The input may originate from a sensory parser or other data parser in machine systems, or from biological sensory organs in human beings. By operating directly after the input layer of a system, ELM provides a structured mechanism for managing interpretation toward understanding rather than premature judgment, contextual data rather than static guessing, and the detection and resolution of uncertainty and predictive errors. This includes both preventing and resolving predictive errors—ELM helps identify the triggers of predictive errors and resolve them to prevent the errors from occurring in the first place—as well as helping to resolve predictive errors when others commit them. ELM examines whether the received input contains sufficient contextual information, insufficient contextual information, or distorted contextual information. When contextual information is unavailable or distorted, ELM routes the system toward obtaining or refining the required contextual information through the appropriate contextual-extraction and interaction processes, while maintaining the safeguards specified by the ELM architecture to avoid introducing further predictive errors into the interaction. When the input already contains sufficient contextual information, or when missing contextual information has been obtained through the appropriate extraction and refinement processes established in ELM, ELM subjects the resulting contextual information to the verification and stabilization processes defined by its architecture. When the relevant criteria are satisfied, the resulting information can then be made available to the system for its application-specific purpose. ELM does not impose artificial agreement and does not seek to eliminate disagreement. It prioritizes proportional coherence over certainty, clarity over control, and understanding over judgment. Accordingly, ELM provides an end-to-end computational routing architecture for the interpretation process within the operational scope defined by its formal specification, including contextual assessment, contextual acquisition and refinement, verification, stabilization, state transitions, interruption handling, and recovery. This allows systems such as human beings, corporations, institutions, robots, artificial intelligence systems, machine-learning pipelines, autonomous systems, and other interactive agents to operate on contextually processed and verified information rather than relying solely on static guessing or unverified input, subject to their respective implementation requirements and empirical validation. Mathematical Formalization of ELM into Computational Architecture and Finite State Machine (discrete mathematics and calculus), available at: https://doi.org/10.5281/zenodo.22149070
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.
