Search

El Big Data y la inteligencia artificial se están articulando para luchar contra la desinformación, que comprende todos los aspectos posibles de lo falso, contenido inexacto o engañoso, proyectado, presentado y promovido interesadamente para ocasionar algún daño. Para hacer frente a este fenómeno, organizaciones, entidades, gobiernos y medios de comunicación están poniendo en marcha diferentes iniciativas. Muchas de ellas hacen uso del Big Data y de la inteligencia artificial (IA) que, con el desarrollo de los algoritmos, diseñan y ponen en marcha bots y plataformas cuya pretensión es acabar con la información intencionalmente errada. El presente ensayo ofrece una visión panorámica y un recorrido por algunos de los proyectos que la Unión Europea ha puesto en marcha basados en el Big Data, la ciencia de datos y la inteligencia artificial para desarmar la desinformación y los bulos, como los proyectos FactCheckEU o FANDANGO. Para ello, se ha aplicado un estudio cualitativo documental bibliográfico que ha permitido proponer una serie de medidas que pueden limitar la difusión de información falsa y concluir que la Unión Europea tiene actualmente el gran reto de combatir la desinformación y está dedicando una parte importante de sus recursos a identificar el problema y proponer soluciones., Big Data and artificial intelligence are being harnessed to fight disinformation, which encompasses all possible aspects of false, inaccurate or misleading content, projected, presented and promoted to cause harm. To tackle this phenomenon, organisations, entities, governments and the media are launching different initiatives. Many of them make use of Big Data and artificial intelligence (AI) which, with the development of algorithms, design and implement bots and platforms whose aim is to put an end to intentionally misleading information. This essay offers an overview and a tour of some of the projects that the European Union has launched based on Big Data, data science and artificial intelligence to disarm disinformation and hoaxes, such as the FactCheckEU or FANDANGO projects. To this end, a qualitative documentary bibliographic study has been applied to propose a series of measures that can limit the dissemination of false information and to conclude that the European Union currently faces the great challenge of combating disinformation and is dedicating a significant part of its resources to identifying the problem and proposing solutions.

OBJECTIVE: To develop a generative adversarial network (GAN) to segment major blood vessels from retinal flat-mount images from oxygen-induced retinopathy (OIR) and demonstrate the utility of these GAN-generated vessel segmentations in quantifying vascular tortuosity. DESIGN: Development and validation of GAN. SUBJECTS: Three datasets containing 1084, 50, and 20 flat-mount mice retina images with various stains used and ages at sacrifice acquired from previously published manuscripts. METHODS: Four graders manually segmented major blood vessels from flat-mount images of retinas from OIR mice. Pix2Pix, a high-resolution GAN, was trained on 984 pairs of raw flat-mount images and manual vessel segmentations and then tested on 100 and 50 image pairs from a held-out and external test set, respectively. GAN-generated and manual vessel segmentations were then used as an input into a previously published algorithm (iROP-Assist) to generate a vascular cumulative tortuosity index (CTI) for 20 image pairs containing mouse eyes treated with aflibercept versus control. MAIN OUTCOME MEASURES: Mean dice coefficients were used to compare segmentation accuracy between the GAN-generated and manually annotated segmentation maps. For the image pairs treated with aflibercept versus control, mean CTIs were also calculated for both GAN-generated and manual vessel maps. Statistical significance was evaluated using Wilcoxon signed-rank tests (P ≤ 0.05 threshold for significance). RESULTS: The dice coefficient for the GAN-generated versus manual vessel segmentations was 0.75 ± 0.27 and 0.77 ± 0.17 for the held-out test set and external test set, respectively. The mean CTI generated from the GAN-generated and manual vessel segmentations was 1.12 ± 0.07 versus 1.03 ± 0.02 (P = 0.003) and 1.06 ± 0.04 versus 1.01 ± 0.01 (P < 0.001), respectively, for eyes treated with aflibercept versus control, demonstrating that vascular tortuosity was rescued by aflibercept when quantified b

© 2024 Elsevier. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0, In this paper we present SurvLIMEpy, an open-source Python package that implements the SurvLIME algorithm. This method allows to compute local feature importance for machine learning algorithms designed for modelling Survival Analysis data. The presented implementation uses a matrix-wise formulation, which allows to speed up the execution time. Additionally, SurvLIMEpy assists the user with visualisation tools to better understand the result of the algorithm. The package supports a wide variety of survival models, from the Cox Proportional Hazards Model to deep learning models such as DeepHit or DeepSurv. Two types of experiments are presented in this paper. First, by means of simulated data, we study the ability of the algorithm to capture the importance of the features. Second, we use three open source survival datasets together with a set of survival algorithms in order to demonstrate how SurvLIMEpy behaves when applied to different models., Peer Reviewed, Postprint (author's final draft)

A novel fault diagnosis procedure is proposed in this paper to estimate faults using a linear parameter varying (LPV) model whose scheduling parameters depend on the fault. A wrong determination of the operating conditions could lead the system to an undesired performance or even to an unstable situation, when classical fault diagnosis approaches are applied. This paper addresses this issue by formulating fault diagnosis as a dynamic optimization problem, solved by using a novel hybrid technique that combines a Luenberger-based observer with artificial intelligent (AI) optimization-based algorithms. The observer supervises the health of the system, while AI-based algorithms are able to reconstruct the faulty signal in real-time when the observer determines that the system is under a fault. The efficiency of the proposed fault diagnosis scheme, the three AI-based algorithms based on artificial bee colony and particle swarm optimization, and the gradient-based algorithm developed in this paper, are assessed using a numerical example., This work has been co-financed by the Spanish State Research Agency (AEI) and the European Regional Development Fund (ERFD) through the project SaCoAV (ref. MINECO PID2020-114244RB-100), by the European Regional Development Fund of the European Union in the framework of the ERDF Operational Program of Catalonia 2014-2020 (ref. 001-P-001643 Looming Factory) and by the DGR of Generalitat de Catalunya (SAC group ref. 2017/SGR/482)., Peer Reviewed, Postprint (author's final draft)

El Big Data y la inteligencia artificial se están articulando para luchar contra la desinformación, que comprende todos los aspectos posibles de lo falso, contenido inexacto o engañoso, proyectado, presentado y promovido interesadamente para ocasionar algún daño. Para hacer frente a este fenómeno, organizaciones, entidades, gobiernos y medios de comunicación están poniendo en marcha diferentes iniciativas. Muchas de ellas hacen uso del Big Data y de la inteligencia artificial (IA) que, con el desarrollo de los algoritmos, diseñan y ponen en marcha bots y plataformas cuya pretensión es acabar con la información intencionalmente errada. El presente ensayo ofrece una visión panorámica y un recorrido por algunos de los proyectos que la Unión Europea ha puesto en marcha basados en el Big Data, la ciencia de datos y la inteligencia artificial para desarmar la desinformación y los bulos, como los proyectos FactCheckEU o FANDANGO. Para ello, se ha aplicado un estudio cualitativo documental bibliográfico que ha permitido proponer una serie de medidas que pueden limitar la difusión de información falsa y concluir que la Unión Europea tiene actualmente el gran reto de combatir la desinformación y está dedicando una parte importante de sus recursos a identificar el problema y proponer soluciones., Big Data and artificial intelligence are being harnessed to fight disinformation, which encompasses all possible aspects of false, inaccurate or misleading content, projected, presented and promoted to cause harm. To tackle this phenomenon, organisations, entities, governments and the media are launching different initiatives. Many of them make use of Big Data and artificial intelligence (AI) which, with the development of algorithms, design and implement bots and platforms whose aim is to put an end to intentionally misleading information. This essay offers an overview and a tour of some of the projects that the European Union has launched based on Big Data, data science and artificial intelligence to disarm disinformation and hoaxes, such as the FactCheckEU or FANDANGO projects. To this end, a qualitative documentary bibliographic study has been applied to propose a series of measures that can limit the dissemination of false information and to conclude that the European Union currently faces the great challenge of combating disinformation and is dedicating a significant part of its resources to identifying the problem and proposing solutions.

Although the field of Artificial Intelligence in Education (AIEd) has a substantial history as a research domain, never before has the rapid evolution of AI applications in education sparked such prominent public discourse. Given the growing AIEd literature base in higher education, it is important to ensure that the field has a solid research and conceptual grounding as AI adoption increases. This review of reviews is the first comprehensive meta review to explore the scope and nature of AIEd in higher education (AIHEd) research, by synthesising secondary research (e.g., systematic reviews), indexed in the Web of Science, Scopus, ERIC, EBSCOHost, IEEE Xplore, ScienceDirect and ACM Digital Library, or captured through snowballing in OpenAlex, ResearchGate and Google Scholar. Reviews were included if they synthesised applications of AI solely in formal higher or continuing education, were published in English between 2018 and July 2023, were journal articles or full conference papers, and if they had a method section. 66 publications were included for data extraction and synthesis in EPPI Reviewer, which were predominantly systematic reviews (66.7%), published by authors from North America (27.3%), conducted in teams (89.4%) in mostly domestic-only collaborations (71.2%). Findings show that these reviews mostly focused on AIHEd generally (47.0%) or Profiling and Prediction (28.8%) as thematic foci, however key findings indicated a predominance of the use of Adaptive Systems and Personalisation in higher education. Research gaps identified suggest a need for greater ethical, methodological, and contextual considerations within future research, alongside interdisciplinary approaches to AIHEd application. Suggestions are provided to guide future primary and secondary research.

Although the field of Artificial Intelligence in Education (AIEd) has a substantial history as a research domain, never before has the rapid evolution of AI applications in education sparked such prominent public discourse. Given the growing AIEd literature base in higher education, it is important to ensure that the field has a solid research and conceptual grounding as AI adoption increases. This review of reviews is the first comprehensive meta review to explore the scope and nature of AIEd in higher education (AIHEd) research, by synthesising secondary research (e.g., systematic reviews), indexed in the Web of Science, Scopus, ERIC, EBSCOHost, IEEE Xplore, ScienceDirect and ACM Digital Library, or captured through snowballing in OpenAlex, ResearchGate and Google Scholar. Reviews were included if they synthesised applications of AI solely in formal higher or continuing education, were published in English between 2018 and July 2023, were journal articles or full conference papers, and if they had a method section. 66 publications were included for data extraction and synthesis in EPPI Reviewer, which were predominantly systematic reviews (66.7%), published by authors from North America (27.3%), conducted in teams (89.4%) in mostly domestic-only collaborations (71.2%). Findings show that these reviews mostly focused on AIHEd generally (47.0%) or Profiling and Prediction (28.8%) as thematic foci, however key findings indicated a predominance of the use of Adaptive Systems and Personalisation in higher education. Research gaps identified suggest a need for greater ethical, methodological, and contextual considerations within future research, alongside interdisciplinary approaches to AIHEd application. Suggestions are provided to guide future primary and secondary research.

Digital twins (DTs) promise innovation for the marine renewable energy sector using modern technological advances and the existing maritime knowledge frameworks. The DT is a digital equivalent of a real object that reflects and predicts its behaviours and states in a virtual space over its lifetime. DTs collect data from multiple sources in pilots and leverage newly introduced low-cost sensor systems. They synchronize, homogenize, and transmit the data to a central hub and integrate it with predictive and learning models to optimize plant performance and operations. This research presents critical aspects of DT implementation challenges in marine energy digitalization DT approaches that use and combine data systems. Firstly, the DT and the existing framework for marine knowledge provided by systems are presented, and the DT's main development steps are discussed. Secondly, the DT implementing main stages, measurement systems, data harmonization and preprocessing, modelling, comprehensive data analysis, and learning and optimization tools, are identified. Finally, the ILIAD (Integrated Digital Framework for Comprehensive Maritime Data and Information Services) project has been reviewed as a best EU funding practice to understand better how marine energy digitalization DT's approaches are being used, designed, developed, and launched.

AI-as-a-Service (AIaaS) combines Artificial Intelligence (AI) and cloud computing to make AI accessible to enterprises without implementing complex solutions or technologies on-premise. Many small and medium-sized enterprises (SME) that lack competencies in the AI and technology sector consider AIaaS as a promising option to implement AI solutions. However, the differences between AIaaS and AI on-premise have not attracted much research. The intention of this paper is to contribute to this area by analysing the literature in the field and investigating a concrete example in more detail. Exploring AIaaS is crucial to better understand the opportunities and limitations of AI services. The contributions of the paper are (a) an analysis of the literature on AIaaS to identify factors affecting AI implementation and how AIaaS solutions differ from on-premise solutions when introducing AI in a company, (b) a case study of an SME that compares AIaaS and AI on-premise in practice, and (c) the application potential of a morphological box to compare AIaaS and AI on-premise.

The Association for the Study of Reproductive Biology (ASEBIR) Interest Group in Embryology (in Spanish 'Grupo de Interés de Embriología') reviewed key morphokinetic parameters to assess the contribution of time-lapse technology (TLT) to the ASEBIR grading system. Embryo grading based on morphological characteristics is the most widely used method in human assisted reproduction laboratories. The introduction and implementation of TLT has provided a large amount of information that can be used as a complementary tool for morphological embryo evaluation and selection. As part of IVF treatments, embryologists grade embryos to decide which embryos to transfer or freeze. At the present, the embryo grading system developed by ASEBIR does not consider dynamic events observed through TLT. Laboratories that are using TLT consider those parameters as complementary data for embryo selection. The aim of this review was to evaluate review time-specific morphological changes during embryo development that are not included in the ASEBIR scoring system, and to consider them as candidates to add to the scoring system.

255 p., The need for rehabilitation has increased over the years. The individualisation of rehabilitation plays an important role in improving the lives of people in need. Within this individualisation of rehabilitation, the use of physical activity data can be very useful to know the condition of the person. But the knowledge of constant physical activity is almost impossible for therapists.The use of technology, with the purpose of helping to know the physical activity performed by people, is becoming more and more common. Through the sensorization of the person, very interesting data can be obtained to help in rehabilitation. In addition, thanks to this sensorization, it is possible to know other types of events, such as a fall, whose rapid action can avoid increasing the need for rehabilitation.Typically this type of sensorization is done using wearable sensors, but several researchers have proposed the use of sensors integrated into assistive devices for walking for those who need them, as this solution is less invasive. While many of the devices can measure physical activity, they have the disadvantage that the sensing elements are fixed to the assistive devices for walking. This may prevent the user from being comfortable, as it is not the assistive device for walking that he/she normally uses. For this reason, the design of an interchangeable sensor tip between different assistive devices for walking is very important.This paper presents an innovative prototype of a sensorized tip that can be interchanged between different assistive devices for walking. It is capable of classifying different physical activities as well as detecting falls. The developed tip is able to adapt to different assistive devices for walking on the market. In addition, this sensorized tip consists of a series of sensors that provide the information needed to classify physical activities and detect falls. These sensors are: a force sensor to measure axial forces; a 3-axis accelerometer, a 3-ax, Errehabilitazio beharra areagotu egin da urteak igaro ahala. Indibidualizazioak oso eginkizun garrantzitsua du errehabilitazio beharra duten pertsonen bizitza hobetzeko. Errehabilitazioaren indibidualizazio horren barruan, jarduera fisikoari buruzko datuak erabiltzea oso baliagarria izan daiteke pertsonaren egoera ezagutzeko. Baina jarduera fisiko etengabea ezagutzea ia ezinezkoa da terapeutentzat.Teknologiaren erabilera, pertsonek egiten duten jarduera fisikoa ezagutzeko, gero eta ohikoagoa da. Pertsonaren sentsorizazioaren bidez, oso datu interesgarriak lor daitezke errehabilitazioan laguntzeko. Gainera, sentsorizazio horri esker, beste gertaera mota batzuk ezagutzea lor daiteke, hala nola erorketa bat gertatzen denean, eta horien jarduera azkarrak errehabilitazio beharra areagotzea ekidin dezake.Oro har, sentsorizazio mota hori sentsore eramangarrien bidez egiten da, baina hainbat ikertzailek proposatu dute sentsore integratuak erabiltzea laguntza teknikoko gailuetan, behar duten pertsonentzat, eta soluzio hori ez da hain inbaditzailea. Gailu askok jarduera fisikoa neur badezakete ere, elementu sentsoreak ibiltzeko laguntza gailuetan finko egotearen eragozpena dute. Horrek erabiltzailea eroso egotea eragotz dezake, ez baita normalean erabiltzen duen ibiltzeko laguntza gailua. Hori dela eta, ibiltzeko laguntza gailu ezberdinen artean alda daitekeen punta sentsorizatu baten diseinua oso garrantzitsua da.Lan honek ibiltzeko laguntza gailu ezberdinen artean trukatzeko gai den punta sentsorizatuaren prototipo berritzailea aurkezten du. Jarduera fisikoak sailkatzeko eta erorikoak detektatzeko gai da. Garatutako punta merkatuko ibiltzeko laguntza gailuetara egokitzeko gai da. Gainera, sentsorizatutako punta honek sentsoreak ditu, jarduera fisikoen sailkapena egiteko eta erorikoak detektatzeko informazioa ematen dutenak. Sentsore hauek dira: indar axialak neurtzeko indar-sentsore bat; 3 ardatzeko azelerometro bat, 3 ardatzeko giroskopio bat eta 3 ardatzeko magnetometro b

226 p., Actualmente la industria está inmersa en su cuarta revolución, caracterizada por la digitalización de las plantas y los procesos industriales. A este fenómeno se le conoce como Industry 4.0. De manera similar, el campo del petróleo y gas está adoptando dichas medidas de captación de datos y extracción de información a partir de los mismos, dando lugar a la estrategia Oil and Gas 4.0. En este sentido y especialmente en las refinerías, se plantea el uso de sensores virtuales para inferir, la calidad de los subproductos derivados del petróleo. Dichos sensores virtuales permiten complementar la función de los analizadores online integrados en planta, así como de los test de laboratorio proporcionando con antelación y de manera continua estimaciones de la calidad de los subproductos. Este conocimiento temprano de dicha calidad conlleva la ventaja de poder aplicar acciones correctivas de re-preprocesamiento del subproducto que no cumpla las especificaciones con la suficiente antelación, ahorrando costes a la refinería.Así, el objetivo principal de esta tesis es diseñar, desarrollar y validar una metodología para la automatización del diseño y desarrollo de sensores virtuales basada en Machine Learning y muy especialmente en Feature Engineering. Para ello, se ha investigado en el impacto en los algoritmos de Machine Learning de diferentes métodos de preprocesamiento de las características o features, especialmente en los métodos de ponderación (Feature Weighting) y de normalización (Feature Normalisation), demostrando su importancia a la hora de mejorar el rendimiento de los modelos. Como resultado de dicha investigación se ha propuesto una metodología sistemática para la selección del método de preprocesamiento adecuado dado un problema.Además, en base a las conclusiones obtenidas en los estudios realizados, se ha propuesto una solución AutoML que considera diferentes métodos para el preprocesamiento de las variables, así como diversos algoritmos de Machine Learning y bus

Disruptive innovation is widely recognised as a bold enterprise. Risks and uncertainties drive incumbent firms to seek alternative solutions to find disruptive ideas. Machine learning emerges as a powerful tool to reduce uncertainties while processing vast amounts and types of information. However, incumbents encounter immense difficulty in codifying tacit knowledge into effective algorithms and often end up with incremental or tactical outcomes despite bold aspirations. Using the literature on problem finding, we explore the development process of machine learning for ideation. Our action research conducted on a healthcare firm provides theoretical and managerial contributions. First, this study suggests that ideation for disruptive innovation benefits from machine learning by facilitating a heuristic search in which a group of actors evaluate plausible hypotheses rather than seek logically accurate conclusions. Previous studies on ideation stress directional search. Second, we propose an ideation process centred on problem formulation to identify disruptive innovation based on its inherent characteristics (e.g., radical functionality and discontinuous technical standard). Third, we discuss the challenges of adopting algorithm-based systems in the ideation — a process well known for being fuzzy.

255 p., The need for rehabilitation has increased over the years. The individualisation of rehabilitation plays an important role in improving the lives of people in need. Within this individualisation of rehabilitation, the use of physical activity data can be very useful to know the condition of the person. But the knowledge of constant physical activity is almost impossible for therapists.The use of technology, with the purpose of helping to know the physical activity performed by people, is becoming more and more common. Through the sensorization of the person, very interesting data can be obtained to help in rehabilitation. In addition, thanks to this sensorization, it is possible to know other types of events, such as a fall, whose rapid action can avoid increasing the need for rehabilitation.Typically this type of sensorization is done using wearable sensors, but several researchers have proposed the use of sensors integrated into assistive devices for walking for those who need them, as this solution is less invasive. While many of the devices can measure physical activity, they have the disadvantage that the sensing elements are fixed to the assistive devices for walking. This may prevent the user from being comfortable, as it is not the assistive device for walking that he/she normally uses. For this reason, the design of an interchangeable sensor tip between different assistive devices for walking is very important.This paper presents an innovative prototype of a sensorized tip that can be interchanged between different assistive devices for walking. It is capable of classifying different physical activities as well as detecting falls. The developed tip is able to adapt to different assistive devices for walking on the market. In addition, this sensorized tip consists of a series of sensors that provide the information needed to classify physical activities and detect falls. These sensors are: a force sensor to measure axial forces; a 3-axis accelerometer, a 3-ax, Errehabilitazio beharra areagotu egin da urteak igaro ahala. Indibidualizazioak oso eginkizun garrantzitsua du errehabilitazio beharra duten pertsonen bizitza hobetzeko. Errehabilitazioaren indibidualizazio horren barruan, jarduera fisikoari buruzko datuak erabiltzea oso baliagarria izan daiteke pertsonaren egoera ezagutzeko. Baina jarduera fisiko etengabea ezagutzea ia ezinezkoa da terapeutentzat.Teknologiaren erabilera, pertsonek egiten duten jarduera fisikoa ezagutzeko, gero eta ohikoagoa da. Pertsonaren sentsorizazioaren bidez, oso datu interesgarriak lor daitezke errehabilitazioan laguntzeko. Gainera, sentsorizazio horri esker, beste gertaera mota batzuk ezagutzea lor daiteke, hala nola erorketa bat gertatzen denean, eta horien jarduera azkarrak errehabilitazio beharra areagotzea ekidin dezake.Oro har, sentsorizazio mota hori sentsore eramangarrien bidez egiten da, baina hainbat ikertzailek proposatu dute sentsore integratuak erabiltzea laguntza teknikoko gailuetan, behar duten pertsonentzat, eta soluzio hori ez da hain inbaditzailea. Gailu askok jarduera fisikoa neur badezakete ere, elementu sentsoreak ibiltzeko laguntza gailuetan finko egotearen eragozpena dute. Horrek erabiltzailea eroso egotea eragotz dezake, ez baita normalean erabiltzen duen ibiltzeko laguntza gailua. Hori dela eta, ibiltzeko laguntza gailu ezberdinen artean alda daitekeen punta sentsorizatu baten diseinua oso garrantzitsua da.Lan honek ibiltzeko laguntza gailu ezberdinen artean trukatzeko gai den punta sentsorizatuaren prototipo berritzailea aurkezten du. Jarduera fisikoak sailkatzeko eta erorikoak detektatzeko gai da. Garatutako punta merkatuko ibiltzeko laguntza gailuetara egokitzeko gai da. Gainera, sentsorizatutako punta honek sentsoreak ditu, jarduera fisikoen sailkapena egiteko eta erorikoak detektatzeko informazioa ematen dutenak. Sentsore hauek dira: indar axialak neurtzeko indar-sentsore bat; 3 ardatzeko azelerometro bat, 3 ardatzeko giroskopio bat eta 3 ardatzeko magnetometro b

Los capítulos 5 y 6 están sujetos a confidencialidad por la autora. 158 p., La motivación de esta tesis es ayudar a las empresas manufactureras a adoptar tecnologías de la Industria 4.0 mediante el desarrollo de una solución rentable y confiable para mejorar su rendimiento. El enfoque se centra en el mantenimiento predictivo, que tiene como objetivo reducir los tiempos de inactividad de las máquinas y predecir fallos que provocan averías, aumentando así la productividad y las ganancias. La tesis propone una metodología integral, innovadora y novedosa para la predicción de fallos en el contexto de la Industria 4.0. La metodología consta de dos tareas principales: 1) etiquetado automático de bases de datos parcialmente etiquetadas y 2) obtención del mejor conjunto de parámetros que son predictores de un fallo.La primera tarea se logra mediante un sistema autónomo y fácil de usar llamado PLASH que evalúa y optimiza las diferentes soluciones de agrupamiento realizadas por una nueva métrica semi-supervisada llamada PSOM. La metodología proporciona al usuario tanto los resultados de una manera comprensible como una métrica de confiabilidad. Los resultados experimentales demuestran que PLAHS puede adaptarse, evaluar y proporcionar la mejor solución de etiquetado en diferentes tipos de datos y escenarios, incluido un estudio de caso real de una prensa de estampado en frío para la fabricación de pernos.La segunda tarea se logra mediante una metodología inspirada en la hiper-heurística llamada HIMAFP, que identifica de manera autónoma los parámetros significativos de las series temporales que caracterizan el comportamiento de las variables de proceso relevantes que permiten predecir fallos. La metodología encuentra el conjunto óptimo de parámetros que son predictores de un tipo específico de fallo, proporcionando al usuario final información para detectar la anomalía y anticipar el fallo. Los resultados experimentales demuestran que HIMAFP es efectivo al proporcionar a los expertos en el campo valiosos conocimientos sobre el comportamiento de las var

Artificial intelligence (AI) is influencing various sectors of society, including the educational field. The use of AI can have great potential in education, however, it is necessary to know both its performance and its limitations. The main objective of this study is to analyze the prompts made by teachers in initial training in relation to the topic of specific educational support needs, classifying them according to Bloom's Taxonomy. For this, 63 students from the first year of the Primary Education Degree in the subject Information and Communication Technology applied to Education participated. The results show that the highest frequency of prompts made by students correspond to the highest levels of Bloom's taxonomy (apply and create), which suggests that students are capable of using the knowledge acquired in the subject to create new learning situations with their future students. This confirms that the implementation of this methodology is beneficial for the development of cognitive and pedagogical skills of future teachers.

This research proposes a Random Forest RF model to replace the experimental tests required by the ISO 9459–5:2007 for predicting the annual energy supplied and the solar fraction covered by a thermosiphon solar water heating system (TSWHS) for the same locations and daily load volumes that this standard. 38 TSWHS have been tested according to the procedures outlined in the standard ISO 9459-5 and two more have been selected from the Solar Keymark database to get the training and testing data set. From these, data from 36 of the TSWHS were used for RF model training, while data from the remaining four TSWHS were used for its testing. To assess the performance of the RF model, three statistical indicators were calculated: mean absolute percentage error (MAPE), mean absolute error (MAE) and the determination coefficient (R-square). Results show MAPE between 2.94% and 5.86% for the annual energy supplied and the solar fraction and R-Square between 0.995 and 0.998 for the annual energy supplied and between 0.973 and 0.976 for the solar fraction for all locations and daily load volume. Consequently, the RF model could be used successfully to replace the experimental tests required by the Standard.

Desde que hizo su aparición en noviembre de 2022, ChatGPT ha ocupado la atención tanto de medios en línea como tradicionales. Un “rinoceronte gris” que venía dando signos de su presencia en las últimas décadas, pero que no ha sido ni atendido ni abordado adecuadamente antes de convertirse en una realidad. Esta laguna ha propiciado diferentes reacciones ante un modelo que, si bien puede ser útil en determinados contextos y haciendo un uso responsable del mismo, no se haya exento de riesgos. En el presente estudio, examinaremos el estado de la cuestión del fenómeno de ChatGPT en diferentes entornos –Italia, España y la Unión Europea– con la finalidad de apreciar la apremiante necesi-dad de contar con unas líneas básicas regulatorias para que una IA de este tipo pueda implementarse en el mercado sin atentar contra los derechos fundamen-tales de los usuarios y minimice los potenciales peligros inherentes a su uso., Since its emergence in November 2022, ChatGPT has captured the attention of both online and traditional media. A “gray rhino” that had been showing signs of its presence in recent decades, but was neither adequately addressed nor approached before becoming a reality. This gap has led to different reac-tions towards a model that, although it can be useful in certain contexts with responsible use, is not without risks. In this study, we will examine the status of the issue of the ChatGPT phenomenon in different environments - Italy, Spain, and the European Union - with the aim of appreciating the urgent need for basic regulatory guidelines to allow for the implementation of such AI in the market without infringing upon the fundamental rights of users and minimizing the potential risks inherent in its use.

Introducción: La redacción automatizada es uno de los avances tecnológicos que más interrogantes despierta, pero también expectativas, dentro de la aplicación de la inteligencia artificial en los medios de comunicación. En la actualidad, RTVE (Radio Televisión Española) trabaja en un proyecto piloto junto a la empresa Narrativa Inteligencia Artificial para cubrir las elecciones de 2023 en los municipios que forman parte de la llamada España vacía. Un proyecto que contempla la escritura automática de noticias a partir de una herramienta tecnológica que interprete y transforme los resultados electorales y los convierta en noticias sin la intervención humana. Metodología: El presente trabajo parte de la revisión sistemática de bibliografía existente en torno a la materia que trata y constituye un estudio de caso en el que se han realizado entrevistas abiertas a cuatro de los responsables de las empresas mencionadas. Discusión y resultados: Una iniciativa pionera de última generación en la que RTVE, utilizando la tecnología como palanca, pretende incrementar la difusión y el alcance de las noticias en regiones con escasa cobertura periodística. Conclusiones: La inteligencia artificial aplicada al periodismo revierte en competitividad, flexibilidad, fluidez y rapidez. Este cambio de paradigma para la profesión debe tener sus límites en un código ético que regule esta innovadora forma de automatizar la redacción. Sin embargo, su uso adecuado permitirá, como en este caso, integrar y vertebrar territorios, con formatos de noticias adaptados a cada destinatario., Introduction: Automated news writing is one of the technological advances that raises most questions, but also expectations, within the application of artificial intelligence in the media. Currently, RTVE (Radio Televisión Española) is working on a pilot project together with the company Narrativa Inteligencia Artificial to cover the municipal elections of 2023 in the areas that are part of the so-called empty Spain. A project that contemplates the automatic writing of news by a technological tool that interprets and transforms the election results and turns them into news without human intervention. Methodology: This work is based on the systematic review of existing literature on the subject and is a case study in which open interviews have been conducted with four of the managers of the companies mentioned. Discussion and results: A pioneering, state-of-the-art initiative in which RTVE, using technology as a lever, aims to increase the dissemination and reach of news in regions with scarce journalistic coverage. Conclusions: The artificial intelligence applied to journalism leads to competitiveness, flexibility, fluidity and fast. This paradigm shift for the profession must have its limits in a code of ethics that rules this innovative way of automating the writing of news. Nevertheless, its proper use will allow, as in this case, to integrate and vertebrate territories, with news formats adapted to each target audience.

Detecting and isolating faults in collector fields of solar thermal power plants is a crucial and challenging task. The system variables in the collector area are highly coupled, which can lead to a high misclassification rate. For this reason, it becomes necessary to combine knowledge of systems engineering with machine learning techniques that unravel the complex dynamics that govern the systems using historical data. Furthermore, the performance of a solar thermal plant is highly dependent on solar irradiance which changes during the day and is subject to perturbations caused by clouds and other atmospheric conditions. Detecting the fault requires using techniques that cope with the disturbances in solar irradiance. In this work, real irradiance profiles with many types of clouds are used. First, a model-based fault detector is applied, obtaining an accuracy of over 89% for all test irradiances. Then, different machine learning techniques are compared: static neural networks with and without decoupling strategy, dynamic neural networks, dynamic neural networks in cascade, classification trees, random forests, radial basis function networks, and self-organizing maps. The combination of neural networks was the only method that obtained a total accuracy of over 73% and F1-scores over 50% for all the test irradiance profiles.

Letters of recommendation play a significant role in higher education and career progression, particularly for women and underrepresented groups in medicine and science. Already, there is evidence to suggest that written letters of recommendation contain language that expresses implicit biases, or unconscious biases, and that these biases occur for all recommenders regardless of the recommender's sex. Given that all individuals have implicit biases that may influence language use, there may be opportunities to apply contemporary technologies, such as large language models or other forms of generative artificial intelligence (AI), to augment and potentially reduce implicit biases in the written language of letters of recommendation. In this editorial, we provide a brief overview of existing literature on the manifestations of implicit bias in letters of recommendation, with a focus on academia and medical education. We then highlight potential opportunities and drawbacks of applying this emerging technology in augmenting the focused, professional task of writing letters of recommendation. We also offer best practices for integrating their use into the routine writing of letters of recommendation and conclude with our outlook for the future of generative AI applications in supporting this task.

Human emotion perception is a part of affective computing, a branch of computing that studies and develops systems and devices that can recognize, interpret, process, and simulate human affects. Research in human emotion perception, however, has been mostly restricted to psychology-based literature which explores the theoretical aspects of emotion perception, but does not touch upon its practical applications. For instance, human emotion perception plays a pivotal role in an extensive array of sophisticated intelligent systems, encompassing domains such as behavior prediction, social robotics, medicine, surveillance, and entertainment. In order to deploy emotion perception in these applications, extensive research in psychology has demonstrated that humans not only perceive emotions and behavior through diverse human modalities but also glean insights from situational and contextual cues. This dissertation not only enhances the capabilities of existing human emotion perception systems but also forges novel connections between emotion perception and multimedia analysis, social media analysis, and multimedia forensics. Specifically, this work introduces two innovative algorithms that revolutionize the construction of human emotion perception models. These algorithms are then applied to detect falsified multimedia, understand human behavior and psychology on social media networks, and extract the intricate array of emotions evoked by movies. In the first part of this dissertation, we delve into two unique approaches to advance emotion perception models. The first approach capitalizes on the power of multiple modalities to perceive human emotion. The second approach leverages the contextual information, such as the background scene, diverse modalities of the human subject, and intricate socio-dynamic inter-agent interactions. These elements converge to predict perceived emotions with better accuracy, culminating in the development of context-aware human emotion percepti

While deep learning has led to remarkable advancements across various domains, the widespread adoption of neural network models has brought forth significant challenges such as vulnerability to adversarial attacks and model unfairness. These challenges have profound implications for privacy, security, and societal impact, requiring thorough investigation and development of effective mitigation strategies. In this work we address both these challenges. We study adversarial robustness of deep learning models and explore defense mechanisms against poisoning attacks. We also explore the sources of algorithmic bias and evaluate existing bias mitigation strategies in neural networks. Through this work, we aim to contribute to the understanding and enhancement of both adversarial robustness and fairness of deep learning systems.

Mobile robots are widely used and crucial in diverse fields due to their autonomous task performance. They enhance efficiency, and safety, and enable novel applications like precision agriculture, environmental monitoring, disaster management, and inspection. Perception plays a vital role in their autonomous behavior for environmental understanding and interaction. Perception in robots refers to their ability to gather, process, and interpret environmental data, enabling autonomous interactions. It facilitates navigation, object identification, and real-time reactions. By integrating perception, robots achieve onboard autonomy, operating without constant human intervention, even in remote or hazardous areas. This enhances adaptability and scalability. This thesis explores the challenge of developing autonomous systems for smaller robots used in precise tasks like confined space inspections and robot pollination. These robots face limitations in real-time perception due to computing, power, and sensing constraints. To address this, we draw inspiration from small organisms such as insects and hummingbirds, known for their sophisticated perception, navigation, and survival abilities despite their minimalistic sensory and neural systems. This research aims to provide insights into designing compact, efficient, and minimal perception systems for tiny autonomous robots. Embracing this minimalism is paramount in unlocking the full potential of tiny robots and enhancing their perception systems. By streamlining and simplifying their design and functionality, these compact robots can maximize efficiency and overcome limitations imposed by size constraints. In this work, a Minimal Perception framework is proposed that enables onboard autonomy in resource-constrained robots at scales (as small as a credit card) that were not possible before. Minimal perception refers to a simplified, efficient, and selective approach from both hardware and software perspectives to gather and p

The study of adversarial examples has evolved from a niche phenomenon to a well-established branch of machine learning (ML). In the conventional view of an adversarial attack, the adversary takes an input sample, e.g., an image of a dog, and applies a deliberate transformation to this input, e.g., a rotation. This then causes the victim model to abruptly change its prediction, e.g., the rotated image is classified as a cat. Most prior work has adapted this view across different applications and provided powerful attack algorithms as well as defensive strategies to improve robustness. The progress in this domain has been influential for both research and practice and it has produced a perception of better security. Yet, security literature tells us that adversaries often do not follow a specific threat model and adversarial pressure can exist in unprecedented ways. In this dissertation, I will start from the threats studied in security literature to highlight the limitations of the conventional view and extend it to capture realistic adversarial scenarios. First, I will discuss how adversaries can pursue goals other than hurting the predictive performance of the victim. In particular, an adversary can wield adversarial examples to perform denial-of-service against emerging ML systems that rely on input-adaptiveness for efficient predictions. Our attack algorithm, DeepSloth, can transform the inputs to offset the computational benefits of these systems. Moreover, an existing conventional defense is ineffective against DeepSloth and poses a trade-off between efficiency and security. Second, I will show how the conventional view leads to a false sense of security for anomalous input detection methods. These methods build modern statistical tools around deep neural networks and have shown to be successful in detecting conventional adversarial examples. As a general-purpose analogue of blending attacks in security literature, we introduce the Statistical Indistinguishabil

Regular, continuous monitoring of the heart is advantageous to maintaining one’s cardiovascular health as it enables the early detection of potentially life-threatening cardiovascular diseases. Typically, the required devices for continuous monitoring are found in a clinical setting, but recent research developments have advanced remote physiological monitoring capabilities and expanded the options for continuous monitoring from home. This thesis focuses on further extending the monitoring capabilities of consumer electronic devices to motivate the feasibility of reconstructing Electrocardiograms via a smartphone camera. First, the relationship between skin tone and remote physiological sensing is examined as variations in melanin concentrations for people of diverse skin tones can affect remote physiological sensing. In this work, a study is performed to observe the prospect of reducing the performance disparity caused by melanin differences by exploring the sites from which the physiological signal is collected. Second, the physiological signals obtained from the previous part are enhanced to improve the signal-to-noise ratio and utilized to infer ECG as parts of a novel technique that emphasizes interpretability as a guiding principle. The findings in this work have the potential to enable and promote the remote sensing of a physiological signal that is more informative than what is currently possible with remote sensing.

One of the frequent communication challenges in group meetings between native English speakers (NES) and non-native English speakers (NNES) is confusion experienced by the NNES while communicating in English with the NES. To address this, I proposed integrating a facial expression recognition tool with a video conferencing platform to help NES enhance their ability to identify NNES’ confused moments and promote effective communication with NNES. The study was conducted on a video conferencing platform to investigate how precisely the NES and the facial expression recognition tool identify the confused moments of the NNES in comparison with the self-report of the NNES. Furthermore, this study explored the impact of such identification on how the NES adjusted their communication approach when interacting with the NNES in subsequent group meetings. The findings revealed that the self-reports of NNES played a significant role over the facial expression recognition tool, enabling the NES to better perceive the confused moments of the NNES. As a result, the NES gained a better understanding of the NNES and improved their ability to communicate effectively during the meeting. Although using the facial expression recognition tool to detect the NNES’ confusion during the meeting was underestimated by the NES in this study, it provided me a valuable opportunity to investigate the context of how NNES, NES, and the tool identify the confused moments of the NNES. This highlighted the importance of considering the NNES’ self-reports and their facial expressions within the context to improve the current facial expression recognition tool.

The objective of this research is to develop robust-resilient-adaptive Reinforcement Learning (RL) systems that are generic, provide performance guarantees, and can generalize-reason-improve in complex and unknown task environments. To achieve this objective, we focus on exploring the concept of Risk-sensitivity in RL systems and its extensions to Multi-Agent (MA) systems. The development of robust reinforcement learning algorithms is crucial to address challenges such as model misspecification, parameter uncertainty, disturbances, and more. Risk-sensitive methods offer an approach to developing robust RL algorithms by hedging against undesirable outcomes in a probabilistic manner. The robustness properties of risk-sensitive controllers have long been established. We investigate risk-sensitive RL (as a generalization of risk-sensitive stochastic control), by theoretically analyzing the risk-sensitive exponential (exponential of the total reward) criteria and the benefits and improvements the introduction of risk-sensitivity brings to conventional RL. By considering exponential criteria as risk measures, we aim to enhance the reliability of our decision-making process. We explore the exponential criteria to better understand its representation, the implications of its optimization, and the behavioral characteristics exhibited by an agent optimizing this criterion. We demonstrate the advantages of utilizing exponential criteria for the development of RL algorithms. We then shift our focus to developing algorithms that effectively leverage these exponential criteria. To do that, we first focus on developing risk-sensitive RL algorithms within the framework of Markov Decision Processes (MDPs). We then broaden our scope by exploring the application of the Probabilistic Graphical Models (PGM) framework for developing risk-sensitive algorithms. Within this context, we delve into the PGM framework and examine its connection with the MDP framework. We proceed by exploring th

Adapting to and mitigating climate change while addressing food insecurity are top priorities in SubSaharan Africa that require technologies to improve rural livelihoods with minimal environmental costs [1]. Artificial intelligence (AI) offers great promise for climate-smart solutions that improve food security outcomes. While precision agriculture is often the foremost use case for AI in agriculture (e.g. automation of farm equipment or nutrient application), precision agriculture is out of reach for most African farmers due to the required capital and infrastructure. AI solutions using satellite Earth observations (EOs), which we call AI-EO, are more accessible in the near term. EO enables agricultural analyses and insights at global scales, and many datasets are freely available, making EO-based solutions affordable [2]. AI-EO-derived products such as crop type maps and yield estimates are necessary to forecast food production surpluses or deficits, inform trade, and aid decisions. These products can support policies that accelerate the design and adoption of climate-smart agriculture and impact farmer livelihoods by increasing access to actionable early warning, risk financing or insurance [3], farm inputs, markets, and costreducing interventions [2, 4]. Despite their promise, AI-EO solutions for agriculture in Africa are still limited. Most techniques are not generalizable across heterogeneous landscapes. In this paper, we describe the principal sub-fields of research in AI-EO for agriculture in Africa and discuss examples and limitations of existing work. We also propose ten considerations for future work to help increase the impact of AI-EO research in Africa.

AI is the latest issue in higher education that librarians are finding themselves needing to address in the classroom. With the capabilities of AI changing daily, it has been difficult for librarians to keep abreast of the latest developments. Additionally, AI is an overwhelming concept for many librarians to grasp. The Teaching and Learning Services unit at the University of Maryland created instructional resources and a community of support for UMD librarians to teach and learn about AI. These included an LMS module, a structured workshop series, and a collaborative repository to share lesson plans.

Slender structures, widely found from natural environments (e.g., tendrils) to engineering applications (e.g., flexible electronics), frequently experience geometrically nonlinear deformations and substantial topological changes when exposed to simple boundary conditions or modest external stimuli. On one hand, the nonlinear dynamics of slender structures present considerable challenges for the automated manipulation of these structures by robots. On the other hand, the automated interactions between robots and such structures also open up opportunities to enhance our understanding of the mechanics governing slender structures.This dissertation focuses on the synergy between computational mechanics and robotics for the manipulation and study of slender structures. Specifically, it delves into discrete differential geometry (DDG)-based simulations, an emerging field in computational mechanics, to develop a comprehensive sim2Real manipulation framework for generating task-oriented deformable manipulation strategies. Moreover, we conduct automated experiments to gain valuable insights into the behavior of slender structures. Our contributions can be categorized into three main areas:First, we develop a penalty-energy-based method and combine it with Kirchoff rod's theory to simulate rod assemblies with frictional contact responses. Our simulation method is validated, demonstrating its robustness, accuracy, and efficiency across diverse scenarios. These scenarios include modeling flagella bundling, a significant biological phenomenon for bacterial navigation, as well as tying knots. These numerical validations underscore the potential of our approach as a significant step toward the ultimate goal of a computational framework for sim2real manipulation tasks. We then combine our numerical framework with desktop experiments to investigate the mechanics of various types of knots.Second, we combine DDG-based simulations, scaling analysis, and machine learning to develop a si

This article is a theoretical discourse about technological machines and artificial intelligence, highlighting their effective interactive outcomes in nursing. One significant influence is technological efficiency which positively affects nursing care time, enabling nurses to focus more on their patients as the core of nursing. The article examines the impact of technology and artificial intelligence on nursing practice in this era of rapid technological advancements and technological dependence. Strategic opportunities in nursing are advanced, exemplified by robotics technology and artificial intelligence. A survey of recent literature focused on what is known about the influence of technology, healthcare robotics, and artificial intelligence on nursing in the contexts of industrialization, societal milieu, and human living environments. Efficient, precision-driven machines with artificial intelligence support a technology-centered society in which hospitals and healthcare systems become increasingly technology-dependent, impacting healthcare quality and patient care satisfaction. As a result, higher levels of knowledge, intelligence, and recognition of technologies and artificial intelligence are required for nurses to render quality nursing care. Designers of health facilities should be particularly aware of nursings increasing dependence on technological advancements in their practice.

Machine learning models have become a ubiquitous part of society, and it has consequently become of paramount importance to understand how to design safe and reliable models. This dissertation attempts to take steps towards this direction by consider two specific problems in reliable machine learning: adversarial examples, which are small test-time perturbations to the input designed to cause misclassification, and data-copying, which occurs when a generative model simply memorizes its training data (giving poor generalization and dangerous security risks).

In highly dynamic urban environments, software stacks for autonomous driving applications must quickly adapt to fast changing environments. Examples of dynamic scenarios include construction sites, road closures, and lane level updates. Failure to adapt to changes in map definitions can result in catastrophic failures in the system that can lead to accidents or, at best, rule violations in shared public roads. This work focuses on identifying strategies that leverage automatically generated map representations to minimize human-in-the-loop efforts and explores methods for integrating nominal planners in the global planning task. The first part of this dissertation covers multi-class semantic mapping for large scale urban driving applications. As part of this framework, sensor fusion based strategies are applied to provide robust depth and semantic estimates from the scene without making strong assumptions about the road topology. Secondly, rasterized and graphical representations are jointly leveraged to formulate a nominal global planning approach for lane-level navigation. This method utilizes the semantic maps introduced and employs a conditional generative model to explicitly model the multi-modal distribution of trajectories that are feasible when driving in an urban setting. We additionally provide details from real-world testing and the open-source data collected from the UC San Diego campus during 2020-2021. In the last chapter, 2D and 3D centerline prediction methods are introduced to reduce the gap in real-time scene understanding. This contribution outlines an automatic label generation process and additionally leverages an occlusion handling approach to reason about centerline prediction with varying degrees of occlusion. The methods proposed achieve robust performance in diverse driving scenarios with promising directions in autonomous driving architectures.

Neuropsychiatric disorders, specifically mental disorders, generally lack a quantitative and biophysiological basis for diagnostics and treatment due to fundamental limitations in theoretical knowledge of the disorder and brain-mind duality. Due to the complex multiscale nature of the brain, large and multimodal datasets as well as biophysically-based simulations are required to elucidate its functioning. As such, computational methods exist to bridge these scales. Specifically, probabilistic graph models are utilized here to capture inherent uncertainty in the system while being computationally efficient and tractable to allow for scaling and combining biophysically and theoretical-based models with Big Data in order to model and diagnose known mental disorders. First, a theoretically-driven computationally efficient reaction-diffusion model of synaptic transmission using Markov models and eigenmode decomposition is presented to scale molecular simulations to neural networks with applications in pharmacological simulations, artificial neural networks, and neuromorphic engineering. The second part connects the network level to behavior using deep learning, graph models, and manifold learning applied to neuroimaging data in adolescent depression using a combination of theory- and data-driven techniques. In addition to creating scalable models, this work interrogates structural and functional biomarkers and creates a neuroimaging pipeline resulting in automatic disorder detection. Finally, the focus shifts to diagnostics of anxiety and depression using behavioral data in the form of natural language processing, making use of transformer deep learning architectures.

BACKGROUND: The purpose of this study was to develop a deep learning approach to automatically segment the scapular bone on magnetic resonance imaging (MRI) images and to compare the accuracy of these three-dimensional (3D) models with that of 3D computed tomography (CT). METHODS: Fifty-five patients with high-resolution 3D fat-saturated T2 MRI were retrospectively identified. The underlying pathology included rotator cuff tendinopathy and tears, shoulder instability, and impingement. Two experienced musculoskeletal researchers manually segmented the scapular bone. Five cross-validation training and validation splits were generated to independently train two-dimensional (2D) and 3D models using a convolutional neural network approach. Model performance was evaluated using the Dice similarity coefficient (DSC). All models with DSC > 0.70 were ensembled and used for the test set, which consisted of four patients with matching high-resolution MRI and CT scans. Clinically relevant glenoid measurements, including glenoid height, width, and retroversion, were calculated for two of the patients. Paired t-tests and Wilcoxon signed-rank tests were used to compare the DSC of the models. RESULTS: The 2D and 3D models achieved a best DSC of 0.86 and 0.82, respectively, with no significant difference observed. Augmentation of imaging data significantly improved 3D but not 2D model performance. In comparing clinical measurements of 3D MRI and CT, there was a mean difference ranging from 1.29 mm to 3.46 mm and 0.05° to 7.47°. CONCLUSION: We have presented a fully automatic, deep learning-based strategy for extracting scapular shape from a high-resolution MRI scan. Further developments of this technology have the potential to allow for surgeons to obtain all clinically relevant information from MRI scans and reduce the need for multiple imaging studies for patients with shoulder pathology.

BACKGROUND: Efforts to develop neuroimaging-based biomarkers in major depressive disorder (MDD), at the individual level, have been limited to date. As diagnostic criteria are currently symptom-based, MDD is conceptualized as a disorder rather than a disease with a known etiology; further, neural measures are often confounded by medication status and heterogeneous symptom states. METHODS: We describe a consortium to quantify neuroanatomical and neurofunctional heterogeneity via the dimensions of novel multivariate coordinate system (COORDINATE-MDD). Utilizing imaging harmonization and machine learning methods in a large cohort of medication-free, deeply phenotyped MDD participants, patterns of brain alteration are defined in replicable and neurobiologically-based dimensions and offer the potential to predict treatment response at the individual level. International datasets are being shared from multi-ethnic community populations, first episode and recurrent MDD, which are medication-free, in a current depressive episode with prospective longitudinal treatment outcomes and in remission. Neuroimaging data consist of de-identified, individual, structural MRI and resting-state functional MRI with additional positron emission tomography (PET) data at specific sites. State-of-the-art analytic methods include automated image processing for extraction of anatomical and functional imaging variables, statistical harmonization of imaging variables to account for site and scanner variations, and semi-supervised machine learning methods that identify dominant patterns associated with MDD from neural structure and function in healthy participants. RESULTS: We are applying an iterative process by defining the neural dimensions that characterise deeply phenotyped samples and then testing the dimensions in novel samples to assess specificity and reliability. Crucially, we aim to use machine learning methods to identify novel predictors of treatment response based on prospective lon

Advancements in observational methods and data collection techniques have empowered oceanographers to gather extensive data on a wide range of oceanic phenomena. Optical imaging systems have provided unprecedented insight into the microscopic world of marine plankton as well as the structure, health, biodiversity, and ecological dynamics of coral reefs. Advances in low-power autonomous acoustic recording devices have enabled continuous long-term monitoring of marine mammals and ocean noise.These data-driven methods involve the collection, analysis, and interpretation of large datasets to gain insights. Although machine learning offers the potential for automating the analysis of large oceanographic datasets, its utilization in this context is accompanied by challenges and problems due to the high spatiotemporal variability and noise inherent in these datasets. This thesis delves into an extensive exploration of state-of-the-art machine learning techniques, specifically tailored to optimize the extraction of valuable information from dynamic oceanographic datasets. To obtain a comprehensive understanding of the problem, instances of dataset shift and noise are examined in three distinct case studies spanning the vision and acoustic domains. The first case study focuses on the problem of novelty detection and class imbalance in the context of plankton image recognition using Images from the WHOI-Plankton dataset. The second case study explores the problem of object detection when samples are collected from different environments or under varying conditions. Lastly, the third case study aims to develop multi-observational techniques to reduce dataset noise using a dataset of acoustic recordings collected in the Santa Barbara Channel. In each case, the core technical goal is the same: to train a convolutional neural network-based system to learn a robust feature representation that generalizes to unforeseen environmental conditions. To achieve this goal, techniques from

We describe the curation, annotation methodology, and characteristics of the dataset used in an artificial intelligence challenge for detection and localization of COVID-19 on chest radiographs. The chest radiographs were annotated by an international group of radiologists into four mutually exclusive categories, including typical, indeterminate, and atypical appearance for COVID-19, or negative for pneumonia, adapted from previously published guidelines, and bounding boxes were placed on airspace opacities. This dataset and respective annotations are available to researchers for academic and noncommercial use.

In the context of computer vision and artificial intelligence, scene understanding is a complex task that involves a combination of image analysis, pattern recognition, context modeling, and often relies on advanced machine learning techniques. Scene understanding has applications in various fields, including autonomous navigation, robotics, surveillance, augmented reality, and more, where a deeper understanding of visual scenes is crucial for making informed decisions and interacting effectively with the environment. This dissertation aims to contribute to two applications: assistive robotics and precision agriculture.The first part of the dissertation focuses on the emerging research area of infant action recognition, specifically targeting the task of reaching—a significant developmental milestone. Existing action recognition models primarily cater to adults, leaving a gap in pediatric applications. To bridge this gap, BabyNet, a lightweight network, is introduced. It employs annotated bounding boxes to capture spatial and temporal relationships, accurately detecting reaching onset, offset, and complete actions. BabyNet outperforms larger networks and provides more accurate detection compared to smaller networks. Challenges emerge due to the dataset's limited perspectives and the reliance on the detector network's performance, which can hinder recognition in specific scenarios. To overcome these limitations, E-BabyNet is proposed. It employs LSTM and Bidirectional LSTM layers to assess reaching actions and deliver precise onset and offset keyframes, handling transitions between actions effectively. E-BabyNet's evaluation against other models demonstrates its ability to accurately detect reaching actions and minimize false detections.The second part focuses on the automation of leaf sampling known to provide highly accurate and timely information to growers. The first step in such automated procedure is identifying and cutting a leaf from a tree. For this purpose

Machine learning has emerged as a leading force in revolutionizing technology, education, and almost every aspect of our lives. Reinforcement learning is a sub-field of machine learning that deals with the effects of dynamic feedback and systems that interact with the environment. In these settings, classic statistical and algorithmic guarantees often do not hold because of non i.i.d. data, dynamic feedback, and distribution shift.We develop a framework for single trajectory learning of nonlinear dynamical systems using mixing arguments. Our main result studies the landscape of empirical risk minimization for learning nonlinear dynamical systems from a single trajectory, and provides uniform gradient convergence guarantee, which is combined with novel one-point convexity to facilitate the learning of nonlinear dynamical systems. Our proposed framework allows for non-convex loss landscape and our sample complexity and statistical error rates are optimal in terms of the trajectory length, dimensions of the system and input/noise strength.Next, we study the problem of learning bilinear dynamical systems from a single trajectory of the system’s states and inputs. Our main contribution is the application of martingale small-ball arguments to derive learning guarantees for non-mixing bilinear dynamical systems. We further extend our analysis to time varying dynamical systems by studying the problem of learning non-mixing Markov jump systems. Specifically, we learn the dynamics in each mode and the Markov transition matrix, underlying the evolution of the mode switches, from a single trajectory of the system’s states, inputs, and modes. Our sample complexity and statistical error rates are optimal in terms of the trajectory length, the dimensions of the system and the input/noise strength.Lastly, as a preliminary to the problem of finding the best LTI dynamical system that can minimize least-squares loss given a single trajectory of an unknown dynamical system, we study th

This thesis presents a new algorithm to mitigate cloud masking in the analysis of sea surface temperature (SST) data generated by remote sensing technologies such as infrared sensor satellites like the Level-2 Visible-Infrared Imager-Radiometer Suite (VIIRS). Cloud coverage interferes with the analysis of all remote sensing data using wavelengths shorter than ≈ 2 microns, significantly limiting the quantity of usable data and creating a biased geographical distribution towards equatorial and coastal regions. Prior studies have led to use of in-painting algorithms like Navier-Stokes but was typically only used up to 5% masking and had limited success. To address this issue, we propose an unsupervised machine learning algorithm called ENKI which uses a Vision Transformer with Masked Autoencoding to reconstruct pixels that are masked out by clouds. We train four different models of ENKI with training mask ratios (referred to as t) set to 10%, 35%, 50%, and 75% on a generated Ocean General Circulation Model (OGCM) dataset known as LLC4320. To evaluate performance we reconstruct LLC 4320 SST images at a patch masking ratio of 10%, 20%, 30%, 40%, 50% (referred to as p) and examine reconstruction qualitatively and statistically by calculating the root means squared error (RMSE) of reconstructed patches. Through our analysis we discover that edge patches contain a higher error rate and that a bias appears in some models when reconstructing images at p masking ratios away from their training mask ratio t. But we consistently find that at all levels of p masking ratios there is one or multiple models that create reconstructions with a mean RMSE of less than ≈ 0.03K which is lower than the estimated sensor error of VIIRS data which is ≈ 0.078 K for daytime, along scan, and ≈ 0.05 K for nighttime, along-scan. We also conclude the complexity of dynamics within an image and the p masking ratio affect RMSE with higher complexity and p masking seeing higher RMSE values. Critically

The ability to perform nanometer-scale optical imaging and spectroscopy is key to deciphering the low-energy effects in quantum materials, as well as vibrational fingerprints in planetary and extraterrestrial particles, catalytic substances, and aqueous biological samples. These tasks can be accomplished by the scattering-type scanning near-field optical microscopy (s-SNOM) technique that has recently spread to many research fields and enabled notable discoveries. Herein, it is shown that the s-SNOM, together with scanning probe research in general, can benefit in many ways from artificial-intelligence (AI) and machine-learning (ML) algorithms. Augmented with AI- and ML-enhanced data acquisition and analysis, scanning probe optical nanoscopy is poised to become more efficient, accurate, and intelligent.

OBJECTIVE: To train and validate the use of a novel artificial intelligence-based thermal imaging system as a screening tool to rule out malignancy in cutaneous and subcutaneous masses in dogs. ANIMALS: Training study: 147 client-owned dogs with 233 masses. Validation Study: 299 client-owned dogs with 525 masses. Cytology was non-diagnostic in 94 masses, resulting in 431 masses from 248 dogs with diagnostic samples. PROCEDURES: The prospective studies were conducted between June 2020 and July 2022. During the scan, each mass and its adjacent healthy tissue was heated by a high-power Light-Emitting Diode. The tissue temperature was recorded by the device and consequently analyzed using a supervised machine learning algorithm to determine whether the mass required further investigation. The first study was performed to collect data to train the algorithm. The second study validated the algorithm, as the real-time device predictions were compared to the cytology and/or biopsy results. RESULTS: The results for the validation study were that the device correctly classified 45 out of 53 malignant masses and 253 out of 378 benign masses (sensitivity = 85% and specificity = 67%). The negative predictive value of the system (i.e., percent of benign masses identified as benign) was 97%. CLINICAL RELEVANCE: The results demonstrate that this novel system could be used as a decision-support tool at the point of care, enabling clinicians to differentiate between benign lesions and those requiring further diagnostics.

Before applying an AI system to any real-world problem, it is first necessary to explicitly specify an objective for the system to solve. Whether this is a loss function over a dataset, an MDP to plan in, or a simulator in which to train an RL agent, we must eventually write down a specification that reduces our vague real-world problem to an explicit computational problem. In this thesis, I view writing an explicit problem specification as an act of communication. In this way, a problem specification is more effective if it better achieves the ends of the AI designer in communicating the true problem specification.We begin by formalizing the problem faced by the designer, which we call the Specification Design Problem. In this problem, the designer must choose between a set of available problem specifications to be solved and aims to select one whose solution performs well on the designer's true unspecified problem. The designer's goal in the specification design problem is to minimize the specification error -- the utility the designer has lost by simply writing down the wrong problem. We show that specification error decomposes into underspecification error, caused by leaving out helpful information, and misspecification error, caused by specifying incorrect information. Furthermore, we can analyze each of these errors in turn. We can characterize underspecification error through a new measure of the value of information. Alternatively, we can understand misspecification error through a geometric analysis of the crucial trade-offs of a particular domain. Taken together, these tools allow us to provide recommendations to the AI designer, which we demonstrate by determining the conditions under which the AI designer should stop specifying more information.In the second part, we use this perspective to study a class of problem specifications often used in practice to mitigate misspecification errors. Such specifications, including maximizing worst-case utility or mi

Natural language processing (NLP) has long been regarded as the pinnacle of artificial intelligence, aiming to achieve a comprehensive understanding of human languages. In recent years, the field has experienced significant advancements with the transition from rule-based approaches to deep learning methodologies. However, the standard approaches often rely on vast amounts of data for learning, highlighting the necessity for more data-efficient techniques. Additionally, effectively utilizing available resources while addressing the challenges of frequent model updates and safeguarding against malicious attacks that exploit limited resources presents another significant problem in NLP. This dissertation focuses on the development of efficient natural language processing (NLP) models under limited data and the effective utilization of available resources. In the first part, we address the challenge of learning models with limited data. For scenarios where only a few examples are available, we propose a meta-learning approach that leverages task-specific meta information to effectively learn new models. For cases with a moderate amount of data but still insufficient for more demanding tasks, we introduce self-supervised learning techniques to enhance performance by incorporating additional learning tasks from the available data. We also explore the limitations of even state-of-the-art language models, such as GPT-3, in handling out-of-distribution data shifts and propose a tutor-based learning approach that converts out-of-distribution problems into in-distribution ones through step-by-step demonstrations.In the second part, we shift our focus to optimizing resource utilization in NLP. Given the rapidly changing nature of the world, frequent updates of deployed models with new data are crucial. We present innovative approaches for effectively updating models in lifelong learning scenarios. As the adoption of large language models as backbone dialogue systems gains popu

OBJECTIVES: Artificial intelligence (AI) holds great promise for transforming the healthcare industry. However, despite its potential, AI is yet to see widespread deployment in clinical settings in significant part due to the lack of publicly available clinical data and the lack of transparency in the published AI algorithms. There are few clinical data repositories publicly accessible to researchers to train and test AI algorithms, and even fewer that contain specialized data from the perioperative setting. To address this gap, we present and release the Medical Informatics Operating Room Vitals and Events Repository (MOVER). MATERIALS AND METHODS: This first release of MOVER includes adult patients who underwent surgery at the University of California, Irvine Medical Center from 2015 to 2022. Data for patients who underwent surgery were captured from 2 different sources: High-fidelity physiological waveforms from all of the operating rooms were captured in real time and matched with electronic medical record data. RESULTS: MOVER includes data from 58 799 unique patients and 83 468 surgeries. MOVER is available for download at https://doi.org/10.24432/C5VS5G, it can be downloaded by anyone who signs a data usage agreement (DUA), to restrict traffic to legitimate researchers. DISCUSSION: To the best of our knowledge MOVER is the only freely available public data repository that contains electronic health record and high-fidelity physiological waveforms data for patients undergoing surgery. CONCLUSION: MOVER is freely available to all researchers who sign a DUA, and we hope that it will accelerate the integration of AI into healthcare settings, ultimately leading to improved patient outcomes.

The rapid development of the current availability of advanced hardware and software is opening up new opportunities for digital creation every day. This situation provides great freedom for new artistic expressions with advanced audio, graphics, interface, and algorithms, including machine learning. However, while our nature is multimodal, these modalities in the digital domain are genuinely separate, and the computational platform allows innumerable varieties of linkages among them. For this reason, the holistic multimodal experience is highly dependent on the design and connection of different modalities. This dissertation explains the properties of coherent digital multimodal instruments and discusses their creative opportunity from the process of music composition and performance. The chapters introduce the related projects with their design process, demonstrating the role of crossmodal correspondence in scientific simulation and proposing a numerical method to evaluate the crossmodal correspondences using the correlation coefficient between the modalities. This dissertation aims to contribute to reorganizing the design process of multimodal instruments beyond the old and recent customs.