This essay proves the Universal Approximation Theorem for discriminatory activation functions, in particular continuous sigmoidal functions, over compact spaces. In other words, a neural network with a discriminatory activation function can approximate any continuous function over a compact space. The theorem guarantees the effectivity of neural networks. / Denna uppsats bevisar Universala Approximations Satsen för diskriminerande funktioner, särskillt kontinuerliga sigmoidala funktioner, över kompakta rum. Med andra ord, ett neuralt nätverk med en discriminerande aktiveringsfunktion kan approximera alla kontinguerliga functioner över kompakta rum. Satsen garanterar effektivitet av neurala nätverk.

Dans cette thèse, nous présentons des modèles appartenant au domaine de l’intelligence artificielle afin de prédire l’indice magnétique global am à partir des paramètres du vent solaire. Ceci est fait dans l’optique de fournir des modèles opérationnels basés sur les données enregistrées par le satellite ACE situé au point de Lagrange L1. L’indice am ne possède pas à l’heure actuelle de modèles de prédiction. Pour prédire cet indice, nous avons fait appel à des modèles non-linéaires que sont les réseaux de neurones, permettant de modéliser le comportement complexe et non-linéaire de la magnétosphère terrestre. Nous avons dans un premier temps travaillé sur le développement et l’optimisation des modèles de réseaux classiques comme le perceptron multi-couche. Ces modèles ont fait leurs preuves en météorologie spatiale pour prédire aussi bien des indices magnétiques spécifiques à des systèmes de courant comme l’indice Dst, caractéristique du courant annulaire, que des indices globaux comme l’indice Kp. Nous avons en particulier étudié un réseau temporel appelé Time Delay Neural Network (TDNN) et évalué sa capacité à prédire l’indice magnétique am à une heure, uniquement à partir des paramètres du vent solaire. Nous avons analysé la sensibilité des performances des réseaux de neurones en considérant d’une part les données fournies par la base OMNI au niveau de l’onde de choc, et d’autre part des données obtenues par le satellite ACE en L1. Après avoir étudié la capacité de ces réseaux à prédire am, nous avons développé un réseau de neurones encore jamais utilisé en météorologie spatiale, le réseau Long Short Term Mermory ou LSTM. Ce réseau possède une mémoire à court et à long terme, et comme le TDNN, fournit des prédictions de l’indice am uniquement à partir des paramètres du vent solaire. Nous l’avons optimisé afin de modéliser au mieux le comportement de la magnétosphère et avons ainsi obtenu de meilleures performances de prédiction de l'indice am par rapport à celles obtenues avec le TDNN. Nous avons souhaité continuer le développement et l’optimisation du LSTM en travaillant sur l’utilisation de fonctions de couplage en entrée de ce réseau de neurones, et sur le développement de réseaux multisorties pour prédire les indices magnétiques am sectoriels ou aσ, spécifiques à chaque secteur Temps Magnétique Local. Enfin, nous avons développé une nouvelle technique combinant réseau LSTM et processus gaussiens, afin de fournir une prédiction probabiliste jusqu’à six heures des indices magnétiques Dst et am. Cette méthode a été dans un premier temps développée pour l’indice magnétique Dst afin de pouvoir comparer les performances du modèle hybride à des modèles de référence, puis appliquée à l’indice magnétique am. / In this thesis, we present models which belongs to the field of artificial intelligence to predict the geomagnetic index am based on solar wind parameters. This is done in terms to provide operational models based on data recorded by the ACE satellite located at the Lagrangian point L1. Currently, there is no model providing predictions of the geomagnetic index am. To predict this index, we have relied on nonlinear models called neural networks, allowing to model the complex and nonlinear dynamic of the Earth’s magnetosphere. First, we have worked on the development and the optimisation of basics neural networks like the multilayer perceptron. These models have proven in space weather to predict geomagnetic index specific to current systems like the Dst index, characteristic of the ring current, as well as the global geomagnetic index Kp. In particular, we have studied a temporal network, called the Time Delay Neural Network (TDNN) and we assessed its ability to predict the geomagnetic index am within one hour, base only on solar wind parameters. We have analysed the sensitivity of neural network performance when considering on one hand data from the OMNI database at the bow shock, and on the other hand data from the ACE satellite at the L1 point. After studying the ability of neural networks to predict the geomagnetic index am, we have developped a neural network which has never been used before in Space Weather, the Long Short Term Memory or LSTM. Like the TDNN, this network provides am prediction based only on solar wind parameters. We have optimised this network to model at best the magnetosphere behaviour and obtained better performance than the one obtained with the TDNN. We continued the development and the optimisation of the LSTM network by using coupling functions as neural network features, and by developing multioutput networks to predict the sectorial am also called aσ, specific to each Magnetical Local Time sector. Finally, we developped a brand new technique combining the LSTM network and gaussian process, to provide probabilistic predictions up to six hours ahead of geomagnetic index Dst and am. This method has been first developped to predict Dst to be able to compare the performance of this model with reference models, and then applied to the geomagnetic index am.

No description available.

<p>Developing a reliable in-situ non-destructive testing method to determine the strength of in-place concrete is critical because a fast-paced construction schedule exposes concrete pavement and/or structures undergoing substantial loading conditions, even at their early ages. Conventional destructive testing methods, such as compressive and flexural tests, are very time-consuming, which may cause construction delays or cost overruns. Moreover, the curing conditions of the tested cylindrical samples and the in-place concrete pavement/structures are quite different, which may result in different strength values. An NDT method that could directly correlate the mechanical properties of cementitious materials with the sensing results, regardless of the curing conditions, mix design, and size effect is needed for the in-situ application.</p><p>The piezoelectric sensor-based electromechanical impedance (EMI) technique has shown promise in addressing this challenge as it has been used to both monitor properties and detect damages on the concrete structure. Due to the direct and inverse effects of piezoelectric, this material can act as a sensor, actuator, and transducer. This research serves as a comprehensive study to investigate the feasibility and efficiency of using piezoelectric sensor-based EMI to evaluate the strength of newly poured concrete. To understand the fundamentals of this method and enhance the durability of the sensor for in-situ monitoring, this work started with sensor fabrication. It has studied two types of polymer coating on the effect of the durability of the sensor to make it practical to be used in the field.</p><p>The mortar and concrete samples with various mix designs were prepared to ascertain whether the results of the proposed sensing technique were affected by the different mixtures. The EMI measurement and compressive strength testing methods (ASTM C39, ASTM C109) were conducted in the laboratory. The experimental results of mortar samples with different water-to-cement ratios (W/C) and two types of cement (I and III) showed that the correlation coefficient (R²) is higher than 0.93 for all mixes. In the concrete experiments, the correlation coefficient between the EMI sensing index and compressive strength of all mixes is higher than 0.90. The empirical estimation function was established through a concrete slab experiment. Moreover, several trial implementations on highway construction projects (I-70, I-74, and I-465) were conducted to monitor the real-time strength development of concrete. The data processing method and the reliable index of EMI sensing were developed to establish the regression model to correlate the sensing results with the compressive strength of concrete. It has been found that the EMI sensing method and its related statistical index can effectively reflect the compressive strength gain of in-place concrete at different ages.</p><p>To further investigate the in-situ compressive strength of concrete for large-scale structures, we conducted a series of large concrete slabs with the dimension of 8 feet × 12 feet × 8 inches in depth was conducted at outdoor experiments field to simulate real-world conditions. Different types of compressive strength samples, including cast-in-place (CIP) cylinder (4” × 6”) – (ASTM C873), field molded cylinder (4” × 8”) – (ASTM C39), and core drilled sample (4” × 8”) – (ASTM C42) were prepared to compare the compressive strength of concrete. The environmental conditions, such as ambient temperatures and relative humidity, were also recorded. The in-situ EMI monitoring of concrete strength was also conducted. The testing ages in this study were started from 6 hours after the concrete cast was put in place to investigate the early age results and continued up to 365 days (one year) later for long-term monitoring. The results indicate that the strength of the CIP sample is higher than the 4” x 8” molded cylinder , and that core drilled concrete is weaker than the two aforementioned. The EMI results obtained from the slab are close to those obtained from CIP due to similar curing conditions. The EMI results collected from 4 × 8-inch cylinder samples are lower than slab and CIP, which aligns with the mechanical testing results and indicates that EMI could capture the strength gain of concrete over time.</p><p>The consequent database collected from the large slab tests was used to build a prediction model for concrete strength. The Artificial Neuron Network (ANN) was investigated and experimented with to optimize the prediction of performances. Then, a sensitivity analysis was conducted to discuss and understand the critical parameters to predict the mechanical properties of concrete using the ML model. A framework using Generative Adversarial Network (GAN) based on algorithms was then proposed to overcome real data usage restrictions. Two types of GAN algorithms were selected for the data synthesis in the research: Tabular Generative Adversarial Networks (TGAN) and Conditional Tabular Generative Adversarial Networks (CTGAN). The testing results suggested that the CTGAN-NN model shows improved testing performances and higher computational efficiency than the TGAN model. In conclusion, the AI-guided concrete strength sensing and prediction approaches developed in this dissertation will be a steppingstone towards accomplishing the reliable and intelligent assessment of in-situ concrete structures.</p><br>

<div>Artificial Intelligence (AI) plays an increasingly pivotal role in drug discovery. In particular, artificial neural networks such as deep neural networks drive this area of research. The research presented in this thesis is considered a synergistic combination of physicochemical models of protein-ligand interactions such as molecular dynamics simulation, novel machine learning concepts and the use of big data for solving fundamental problems in Structure-Based Drug Design (SBDD). This area of research involves the use of three-dimensional (3D) structural data of biomolecules to assist lead discovery and optimization in a time- and cost-efficient manner. </div><div>The main focus of the thesis research is the development of models, algorithms and methods to facilitate binding-mode elucidation, affinity prediction for congeneric series of molecules and flexible docking. </div><div><br></div><div>For pose-prediction, we developed a Convolutional Neural Network model incorporating hydration information, named DeepWatsite, which displays accurate binding-mode prediction and the capability to highlight different roles of water molecules in protein-ligand binding. In order to train the neural network model, we created a comprehensive database for hydration information of thousands of protein systems. This was made possible through the development of an efficient GPU-accelerated version of Watsite, a program for generating hydration profiles of protein systems through molecular dynamics simulations.\newline</div><div>\indent For accurate affinity prediction for congeneric series of compounds, we developed a new methodological platform for mixed-solvent simulation based on the lambda-dynamics concept. Additionally, we developed a deep-learning model that combines molecular dynamics simulations and a distance-aware graph attention algorithm. Validation studies using this method revealed that its accuracy is competitive to resource-intensive free energy perturbation (FEP) calculations. To train the model, we generated a synthetic database of congeneric series of compounds extracted from the highest-quality medicinal chemistry articles. Molecular-dynamics simulations were used to simulate all the generated systems as method for data augmentation.\newline</div><div>\indent For flexible docking, we developed a machine-learning assisted docking strategy that relies on protein-ligand distance matrix predictions. This technique is built upon Weisfeiler-Lehman neural network concept with an attention mechanism. Comprehensive validation on docking and cross-docking datasets demonstrated the potential of this method to become a docking concept with higher accuracy and efficiency than existing state-of-the-art flexible docking techniques. </div><div><br></div><div>In summary, the thesis proved the general applicability of deep-learning to various tasks in SBDD. Furthermore, it demonstrates that treating biomolecules as dynamic entities can improve the quality of computational methods in structure-based drug design.</div>

Presented work tries to reflect the structure of human world, which is able to create an overman as an artificial intelligence through its self-destructive tendency. It investigates the possibilities of our imagination and if we are able to think about artificial intelligence as a sui generis continuation of human species. Hand in hand it tries to create a tension between particular ethico-political decisions and subsequent binding structure, which they implies.

Denna rapport presenterar en förstudie om digitalisering av Mörrumsåns avrinningsområde. Studien undersökte möjligheterna att använda digitala verktyg och tekniker för att förbättra vattenreglering och engagemang från intressenter, särskilt kommunerna i området. Förstudien visade att det fanns generellt gott intresse för att införa digitala lösningar för att förbättra samordningen av vattenreglering. Totalt identifierades 36 potentiella mätpunkter i avrinningsområdet, varav vissa betraktades som högre prioritet. En viktig del av förstudien var ett visualiseringsförslag, där geografiska informationssystem (GIS) är ett möjligt exempel till visualiseringen. Utmaningar identifierades kring sensorval och deras hållbarhet samt noggrannhet i olika miljöer.  Framtidsvisionen inkluderade användning av artificiell intelligens (AI) och maskininlärning (ML) för att förutsäga torka och översvämningar. Slutsatserna visade att digitalisering av vattenhantering kan förbättra samordningen och effektiviteten samt minska risken för torka och översvämningar, men kräver mer datainsamling, standardisering och samarbete mellan intressenter för framgångsrik implementering. Genom att använda digitala verktyg kan samordningen, effektiviteten och hållbarheten förbättras i Mörrumsåns avrinningsområde med målet att bevara ekosystem och vattenresurser för framtida generationer.

The most performant Machine Learning (ML) classifiers have been labeled black-boxes due to the complexity of their decision process. eXplainable Artificial Intelligence (XAI) methods aim to alleviate this issue by crafting an interpretable explanation for a models prediction. A drawback of most XAI methods is that they are heuristic with some drawbacks such as non determinism and locality. Formal Explanations (FE) have been proposed as a way to explain the decisions of classifiers by extracting a set of features that guarantee the prediction. In this thesis we explore these guarantees for different use cases: speeding up the inference speed of tree-based Machine Learning classifiers, curriculum learning using said classifiers and also reducing training data. We find that under the right circumstances we can achieve up to 6x speedup by partially compiling the model to a set of rules that are extracted using formal explainability methods. / De mest effektiva maskininlärningsklassificerarna har betecknats som svarta lådor på grund av komplexiteten i deras beslutsprocess. Metoder för förklarbar artificiell intelligens (XAI) syftar till att lindra detta problem genom att skapa en tolkbar förklaring för modellens prediktioner. En nackdel med de flesta XAI-metoder är att de är heuristiska och har vissa nackdelar såsom icke-determinism och lokalitet. Formella förklaringar (FE) har föreslagits som ett sätt att förklara klassificerarnas beslut genom att extrahera en uppsättning funktioner som garanterar prediktionen. I denna avhandling utforskar vi dessa garantier för olika användningsfall: att öka inferenshastigheten för maskininlärningsklassificerare baserade på träd, kurser med hjälp av dessa klassificerare och även minska träningsdata. Vi finner att under rätt omständigheter kan vi uppnå upp till 6 gånger snabbare prestanda genom att delvis kompilera modellen till en uppsättning regler som extraheras med hjälp av formella förklaringsmetoder.

<p>This thesis aims to design and develop the spatial adaptation approach through spatial transformers to improve the accuracy of human keypoint recognition models. We have studied different model types and design choices to gain an accuracy increase over models without spatial transformers and analyzed how spatial transformers increase the accuracy of predictions. A neural network called Widenet has been leveraged as a specialized network for providing the parameters for the spatial transformer. Further, we have evaluated methods to reduce the model parameters, as well as the strategy to enhance the learning supervision for further improving the performance of the model. Our experiments and results have shown that the proposed deep learning framework can effectively detect the human key points, compared with the baseline methods. Also, we have reduced the model size without significantly impacting the performance, and the enhanced supervision has improved the performance. This study is expected to greatly advance the deep learning of human key points and movement dynamics. </p>

<p dir="ltr">Measuring human kinematics and kinetics is critical for ergonomists to evaluate ergonomic risks related to physical workloads, which are essential for ensuring workplace health and safety. Human kinematics describes human body postures and movements in 6 degrees of freedom (DOF). In contrast, kinetics describes the external forces acting on the human body, such as the weight of loads being handled. Measuring them in the workplace has remained costly as they require expensive equipment, such as motion capture systems, or are only possible to measure manually, such as measuring the weight through a force gauge. Due to the limitations of existing measurement methods, most ergonomics assessments are conducted in laboratory settings, mainly to evaluate and improve the design of workspaces, production tools, and tasks. Continuous monitoring of workers' ergonomic risks during daily operations has been challenging, yet it is critical for ergonomists to make timely decisions to prevent workplace injuries.</p><p dir="ltr">Motivated by this gap, this dissertation proposed three studies that introduce novel low-cost, minimally intrusive, and automated methods to measure human kinematics and kinetics for ergonomics assessments. Specifically, study 1 proposed ErgoNet, a deep learning and computer vision network that takes a monocular image as input and predicts the absolute 3D human body joint positions and rotations in the camera coordinate system. It achieved a Mean Per Joint Position Error of 10.69 cm and a Mean Per Joint Rotation Error of 13.67 degrees. This study demonstrated the ability to measure 6 DOF joint kinematics for continuous and dynamic ergonomics assessments for biomechanical modeling using just a single camera. </p><p dir="ltr">Studies 2 and 3 showed the potential of using pressure-sensing gloves (i.e., tactile gloves) to predict ergonomics risks in lifting tasks, especially the weight of loads. Study 2 investigated the impacts of different lifting risk factors on the tactile gloves' pressure measurements, demonstrating that the measured pressure significantly correlates with the weight of loads through linear regression analyses. In addition, the lifting height, direction, and hand type were found to significantly impact the measured pressure. However, the results also illustrated that a linear regression model might not be the best solution for using the tactile gloves' data to predict the weight of loads, as the weight of loads could only explain 58 \% of the variance of the measured pressured, according to the R-squared value. Therefore, study 3 proposed using deep learning model techniques, specifically the Convolution Neural Networks, to predict the weight of loads in lifting tasks based on the raw tactile gloves' measurements. The best model in study 3 achieved a mean absolute error of 1.58 kg, representing the most accurate solution for predicting the weight of loads in lifting tasks. </p><p dir="ltr">Overall, the proposed studies introduced novel solutions to measure human kinematics and kinetics. These can significantly reduce the costs needed to conduct ergonomics assessments and assist ergonomists in continuously monitoring or evaluating workers' ergonomics risks in daily operations.</p>