Our study investigates how training with sparse simulated data versus sparse real data affects image reconstruction. We compared on several criteria such as number of events, speed and high dynamic range, HDR. The results indicate that the difference between simulated data and real data is not large. Training with real data performed often better, but only by 2%. The findings confirm what earlier studies have shown; training with simulated data generalises well, even when training on sparse datasets as this study shows. / Vår studie undersöker hur träning med gles simulerad data och gles verklig data från en eventkamera, påverkar bildrekonstruktion. Vi tränade två modeller, en med simulerad data och en med verklig för att sedan jämföra dessa på ett flertal kriterier som antal event, hastighet och high dynamic range, HDR. Resultaten visar att skillnaden mellan att träna med simulerad data och verklig data inte är stor. Modellen tränad med verklig data presterade bättre i de flesta fall, men den genomsnittliga skillnaden mellan resultaten är bara 2%. Resultaten bekräftar vad tidigare studier har visat; träning med simulerad data generaliserar bra, och som denna studie visar även vid träning på glesa datamängder.

The purpose of the study was to investigate the effects of real data based and calculator supported statistics activities on 7th grade students&amp / #8217 / statistics performance and attitudes towards statistics when the statistics performance of the students prior to the instruction and the previous mathematics grades were controlled. A quasi-experimental design was used to investigate the research problem. The research was conducted by 84 seventh grade students. There were three groups in the study. Two of them were experimental groups and one of them was control group. The first group received instruction by the traditional method (TM), the second group received instruction by real data based and calculator supported statistical activities (RDBCSSA), and the third group was instructed by real data based statistical activities (RDBSA), the number of the subjects was 28, 27 and 29 respectively. Data were collected through three different measuring instruments: 1.Statistical Performance Test 1 (SPT1) / 2.Statistical Performance Test 2 (SPT2) / 3.Statistics Attitude Scale (SAS). The SPT1 was used as a pre-test. The SPT2 was administered as a post-test. SPT1 and SPT2 were used to determine the statistics performance of the students before and after the instruction. SAS was used to determine the attitudes of the students toward statistics. The data of this study were analyzed by analysis of variance (ANOVA) and analysis of covariance (ANCOVA). The results revealed that there was no significant mean difference among the groups with respect to statistics performance. Also there was no significant mean difference among the groups with respect to attitudes towards statistics. The mean scores of the Statistics Attitude Scale items were calculated and the results revealed that the students had positive attitudes to the statements in Statistics Attitude Subscale 1 (Confidence in Learning Statistics) and they were neutral to the statements of the Statistics Attitude Subscale 2. Also the students wrote their opinions about the teaching methods and their impressions were analyzed by making a frequency table. Most of the TM students mentioned that the examples should be more attractive such that the activity sheets could contain real data based examples. However some of the students mentioned that traditional method was good and the subject was understood very well. Most of the RDBSA students mentioned that the teaching method was enjoyable. Most of the RDBCSSA students expressed that the calculators made the lessons enjoyable and the study easy.

Difficulties in learning and understanding statistics in college education have led to a reform movement in statistics education in the early 1990s. Although much work has been done, there is more work that needs to be done in statistics education. The progress depends on how well the educators bring interesting real-life data into the classroom. The goal was to understand how course design based on First Principles of Instruction could facilitate tertiary-level students' conceptual understanding when learning introductory statistics in a technology-enhanced learning environment. An embedded single descriptive case design was employed to investigate how integrating technology and real data into a tertiary level statistics course would affect students' statistical literacy, reasoning, and thinking. Data including online assignment postings, online discussions, online peer evaluations, a comprehensive assessment, and open-ended interviews were analyzed to understand how the implementation of First Principles of Instruction affected a student's conceptual understanding in a tertiary level introductory statistics course. In addition, the teaching and learning quality (TALQ) survey was administered to evaluate the teaching and learning quality of the designed instruction from the student's perspective. Results from both quantitative and qualitative data analyses indicate that the course designed following Merrill's First Principles of Instruction contributes to a positive overall effectiveness of promoting students' conceptual understanding in terms of literacy, reasoning, and thinking statistically. However, students' statistical literacy, specifically, the understanding of statistical terminology did not develop to a satisfactory level as expected.

L'apprentissage automatique doit faire face à différentes difficultés lorsqu'il est confronté aux particularités des données réelles. En effet, ces données sont généralement complexes, volumineuses, de nature hétérogène, de sources variées, souvent acquises automatiquement. Parmi les difficultés les plus connues, on citera les problèmes liés à la sensibilité des algorithmes aux données bruitées et le traitement des données lorsque la variable de classe est déséquilibrée. Le dépassement de ces problèmes constitue un véritable enjeu pour améliorer l'efficacité du processus d'apprentissage face à des données réelles. Nous avons choisi dans cette thèse de réfléchir à des procédures adaptatives du type boosting qui soient efficaces en présence de bruit ou en présence de données déséquilibrées.Nous nous sommes intéressés, d’abord, au contrôle du bruit lorsque l'on utilise le boosting. En effet, les procédures de boosting ont beaucoup contribué à améliorer l'efficacité des procédures de prédiction en data mining, sauf en présence de données bruitées. Dans ce cas, un double problème se pose : le sur-apprentissage des exemples bruités et la détérioration de la vitesse de convergence du boosting. Face à ce double problème, nous proposons AdaBoost-Hybride, une adaptation de l’algorithme Adaboost fondée sur le lissage des résultats des hypothèses antérieures du boosting, qui a donné des résultats expérimentaux très satisfaisants.Ensuite, nous nous sommes intéressés à un autre problème ardu, celui de la prédiction lorsque la distribution de la classe est déséquilibrée. C'est ainsi que nous proposons une méthode adaptative du type boosting fondée sur la classification associative qui a l’intérêt de permettre la focalisation sur des petits groupes de cas, ce qui est bien adapté aux données déséquilibrées. Cette méthode repose sur 3 contributions : FCP-Growth-P, un algorithme supervisé de génération des itemsets de classe fréquents dérivé de FP-Growth dans lequel est introduit une condition d'élagage fondée sur les contre-exemples pour la spécification des règles, W-CARP une méthode de classification associative qui a pour but de donner des résultats au moins équivalents à ceux des approches existantes pour un temps d'exécution beaucoup plus réduit, enfin CARBoost, une méthode de classification associative adaptative qui utilise W-CARP comme classifieur faible. Dans un chapitre applicatif spécifique consacré à la détection d’intrusion, nous avons confronté les résultats de AdaBoost-Hybride et de CARBoost à ceux des méthodes de référence (données KDD Cup 99). / Machine learning often overlooks various difficulties when confronted real data. Indeed, these data are generally complex, voluminous, and heterogeneous, due to the variety of sources. Among these problems, the most well known concern the sensitivity of the algorithms to noise and unbalanced data. Overcoming these problems is a real challenge to improve the effectiveness of the learning process against real data. In this thesis, we have chosen to improve adaptive procedures (boosting) that are less effective in the presence of noise or with unbalanced data.First, we are interested in robustifying Boosting against noise. Most boosting procedures have contributed greatly to improve the predictive power of classifiers in data mining, but they are prone to noisy data. In this case, two problems arise, (1) the over-fitting due to the noisy examples and (2) the decrease of convergence rate of boosting. Against these two problems, we propose AdaBoost-Hybrid, an adaptation of the Adaboost algorithm that takes into account mistakes made in all the previous iteration. Experimental results are very promising.Then, we are interested in another difficult problem, the prediction when the class is unbalanced. Thus, we propose an adaptive method based on boosted associative classification. The interest of using associations rules is allowing the focus on small groups of cases, which is well suited for unbalanced data. This method relies on 3 contributions: (1) FCP-Growth-P, a supervised algorithm for extracting class frequent itemsets, derived from FP-Growth by introducing the condition of pruning based on counter-examples to specify rules, (2) W-CARP associative classification method which aims to give results at least equivalent to those of existing approaches but in a faster manner, (3) CARBoost, a classification method that uses adaptive associative W-CARP as weak classifier. Finally, in a chapter devoted to the specific application of intrusion’s detection, we compared the results of AdaBoost-Hybrid and CARBoost to those of reference methods (data KDD Cup 99).

This diploma thesis deals with application optimization method in circular economy branch. The introduction is focused on explaining main features of the issue and its benefits for economy and environment. Afterwards are mentioned some obstacles, which are preventing transition from current waste management. Mathematical apparatus, which is used in practical section, is described in the thesis. Core of the thesis is mathematical optimization model, which is implemented in the GAMS software, and generator of input data is made in VBA. The model includes all of significant waste management options with respect to economic and enviromental aspect, including transport. Functionality is then demostrated on a small task. Key thesis result is application of the model on real data concerning Czech Republic. In conclusion an analysis of computation difficulty, given the scale of the task, is accomplished.

Le bus est le moins cher des transports en commun. En contrepartie, il est beaucoup plus difficile à exploiter que le tramway ou le métro qui sont mieux protégés des influences extérieures. Un exemple typique est l’apparition de trains de bus, groupes de véhicules appartenant à la même ligne et arrivant ensemble à un arrêt. Ce phénomène augmente le temps d’attente moyen des usagers aux arrêts et induit un mauvais usage des bus disponibles. Cette thèse développe les outils permettant de garantir la régularité des lignes. Les recherches menées au cours de cette thèse s’articulent suivant deux directions.Un premier constat est que les modèles de lignes de bus existants ne prennent pas en compte les éléments extérieurs que sont les feux de circulation et le trafic environnant. L’absence d’une modélisation mixte intégrant aussi bien les dynamiques internes des lignes que les influences extérieures contraint fortement la diversité des stratégies de contrôle qui ont été proposées jusqu’ici. En effet, les régulations s’appliquent principalement au niveau des arrêts par l’intermédiaire des conducteurs et ne cherchent jamais à réguler le trafic à l’aide des feux de circulation. Un premier axe de recherche développé dans cette thèse est le raffinement des modèles de bus pour prendre en compte le trafic.Plusieurs méthodes d’estimation de temps de parcours sur un boulevard à feu sont proposées. Elles sont basées sur le modèle LWR, compromis fort satisfaisant entre simplicité d’usage et robustesse pour reproduire des situations réelles.Un second constat est que les stratégies de régulation classiques ne sont que rarement basées sur une prévision à court-terme de l'état du système. Elles sont donc souvent actionnées une fois que la situation est trop dégradée, ce qui les rend parfois inaptes à compenser l'instabilité des lignes. Le deuxième axe de recherche consiste à appliquer les modèles raffinés dans un contexte d’exploitation en temps-réel. Le modèle prévoit l'évolution des lignes de bus à court terme, ce qui permet d’actionner préventivement une stratégie de régulation adaptée. En particulier, une méthode de prévision à court terme est développée et testée sur des données réelles. Elle est ensuite combinée à une méthode récente de contrôle des bus. / Bus is cheaper than other transport modes. However, maintaining optimal operations is harder than for streetcars or subways since buses are surrounded by traffic flows. Sometimes, buses of the same route bunch and travel together instead of keeping constant time headways. This phenomenon increases the average waiting time of passengers. As a result, they may tend to shift to other transport modes. This thesis proposes some methods to keep bus routes regular. Two main lines of research are investigated.First, classical models of bus routes do not account for external events like traffic signals and traffic flows. Due to this gap, existing control strategies only apply on buses through their drivers.Traffic flows are not controlled to favor buses compared to cars. Thus, the first area of research consists in refining bus models to account for external events. Several travel time estimation methods on urban arterials are proposed. They are based on the kinematic wave model (LWR). It is known to be a fine trade-off between simplicity and robustness to properly reproduce traffic dynamics.Second, control strategies are often applied once the bus route is too disrupted to be restored to regularity. Predictions of future bus route states could improve the efficiency of regulations. The second area of research consists in using the refined bus models in real time operations. The model forecasts the evolution of buses on their route for short-term. The predictions are evaluated thanks to real data to guarantee their quality. Then it enables regulations to be applied before bunching. In particular, height holding control methods are presented and compared in simulation.

A integração on-line de dados reais com ambientes de realidade virtual imersivos, criando sistemas híbridos, é um campo de pesquisa multidisciplinar, ainda pouco explorado. O alto nível de envolvimento do usuário proporcionado pela realidade virtual imersiva e a confiabilidade e precisão de informações provenientes da aquisição de dados reais, de maneira on-line, podem auxiliar o processo de tomada de decisão, em diversas áreas de aplicação. Este trabalho apresenta uma metodologia de integração on-line de dados reais de manufatura com ambientes de realidade virtual imersivos, mais especificamente, com sistemas CAVE. A metodologia foi desenvolvida voltada para a área de manufatura virtual, com foco em máquinas-ferramenta. Foram criados e interligados dois métodos de aquisição de dados on-line, um deles específico para máquinas-ferramenta, através do protocolo MTConnect e outro, genérico, via imagens de câmeras. O trabalho foi desenvolvido, em geral, com base em bibliotecas e softwares livres, bem como com dispositivos de baixo custo. Por fim, avaliou-se a metodologia através de estudos de caso, baseados em dados reais on-line, provenientes de máquinas-ferramenta reais e de simuladores de dados de máquinas reais. Os resultados deste trabalho podem auxiliar o desenvolvimento de ferramentas que aumentem do nível de confiabilidade das tomadas de decisões baseadas em simulações virtuais da área de manufatura. E, devido à sua implementação modular, ao formato hierárquico e unificado de sua estrutura e à padronização de seus tipos de dados, comunicações e formas de utilização, pode ser adequada a quaisquer áreas de aplicação que utilizem integração de dados reais on-line com ambientes virtuais imersivos. / The on-line integration of real manufacturing data with immersive virtual reality environments, forming hybrid systems, is an underexplored multidisciplinary research field. The high level of users\' involvement provided by immersive virtual reality and the reliability and accuracy of information acquired on-line from real data, can assist the decision making process, in many research areas. This research presents a way to on-line integrate manufacturing real data with immersive virtual reality environments, more specifically, with CAVEs. This work was developed toward the area of virtual manufacturing, with focus on machine tools. Two data acquisition methods were created and placed to work together, one specific for machine tools, through the MTConnect protocol and the other, generic, through camera images acquisition. This work was based on free libraries and software, as well as low cost devices. Finally, this work was evaluated through case studies, based on on-line real data, from real machine tools and real machine data simulators. The results achieved with this research can contribute to the development of new tools to increase the reliability level of manufacturing virtual simulations. Moreover, due to its modular implementation, the hierarquical and unified structure format and the standardization of its communication methods and data types, it can be arranged to almost any research areas which use integration of real on-line data with immersive virtual environments.

A integração on-line de dados reais com ambientes de realidade virtual imersivos, criando sistemas híbridos, é um campo de pesquisa multidisciplinar, ainda pouco explorado. O alto nível de envolvimento do usuário proporcionado pela realidade virtual imersiva e a confiabilidade e precisão de informações provenientes da aquisição de dados reais, de maneira on-line, podem auxiliar o processo de tomada de decisão, em diversas áreas de aplicação. Este trabalho apresenta uma metodologia de integração on-line de dados reais de manufatura com ambientes de realidade virtual imersivos, mais especificamente, com sistemas CAVE. A metodologia foi desenvolvida voltada para a área de manufatura virtual, com foco em máquinas-ferramenta. Foram criados e interligados dois métodos de aquisição de dados on-line, um deles específico para máquinas-ferramenta, através do protocolo MTConnect e outro, genérico, via imagens de câmeras. O trabalho foi desenvolvido, em geral, com base em bibliotecas e softwares livres, bem como com dispositivos de baixo custo. Por fim, avaliou-se a metodologia através de estudos de caso, baseados em dados reais on-line, provenientes de máquinas-ferramenta reais e de simuladores de dados de máquinas reais. Os resultados deste trabalho podem auxiliar o desenvolvimento de ferramentas que aumentem do nível de confiabilidade das tomadas de decisões baseadas em simulações virtuais da área de manufatura. E, devido à sua implementação modular, ao formato hierárquico e unificado de sua estrutura e à padronização de seus tipos de dados, comunicações e formas de utilização, pode ser adequada a quaisquer áreas de aplicação que utilizem integração de dados reais on-line com ambientes virtuais imersivos. / The on-line integration of real manufacturing data with immersive virtual reality environments, forming hybrid systems, is an underexplored multidisciplinary research field. The high level of users\' involvement provided by immersive virtual reality and the reliability and accuracy of information acquired on-line from real data, can assist the decision making process, in many research areas. This research presents a way to on-line integrate manufacturing real data with immersive virtual reality environments, more specifically, with CAVEs. This work was developed toward the area of virtual manufacturing, with focus on machine tools. Two data acquisition methods were created and placed to work together, one specific for machine tools, through the MTConnect protocol and the other, generic, through camera images acquisition. This work was based on free libraries and software, as well as low cost devices. Finally, this work was evaluated through case studies, based on on-line real data, from real machine tools and real machine data simulators. The results achieved with this research can contribute to the development of new tools to increase the reliability level of manufacturing virtual simulations. Moreover, due to its modular implementation, the hierarquical and unified structure format and the standardization of its communication methods and data types, it can be arranged to almost any research areas which use integration of real on-line data with immersive virtual environments.

The aim of this master thesis was to develop a versatile and reliable experimentalplatform of mobile robots, solving tracking problems, for education and research.Evaluation of parallel bearings-only tracking and control algorithms on a multi-corearchitecture has been performed. The platform was implemented as a mobile wirelesssensor network using multiple mobile robots, each using a mounted camera for dataacquisition. Data processing was performed on the mobile robots and on a server,which also played the role of network communication hub. A major focus was toimplement this platform in a flexible manner to allow for education and futureresearch in the fields of signal processing, wireless sensor networks and automaticcontrol. The implemented platform was intended to act as a bridge between the idealworld of simulation and the non-ideal real world of full scale prototypes.The implemented algorithms did estimation of the positions of the robots, estimationof a non-cooperating target's position and regulating the positions of the robots. Thetracking algorithms implemented were the Gaussian particle filter, the globallydistributed particle filter and the locally distributed particle filter. The regulator triedto move the robots to give the highest possible sensor information under givenconstraints. The regulators implemented used model predictive control algorithms.Code for communicating with filters in external processes were implementedtogether with tools for data extraction and statistical analysis.Both implementation details and evaluation of different tracking algorithms arepresented. Some algorithms have been tested as examples of the platformscapabilities, among them scalability and accuracy of some particle filtering techniques.The filters performed with sufficient accuracy and showed a close to linear speedupusing up to 12 processor cores. Performance of parallel particle filtering withconstraints on network bandwidth was also studied, measuring breakpoints on filtercommunication to avoid weight starvation. Quality of the sensor readings, networklatency and hardware performance are discussed. Experiments showed that theplatform was a viable alternative for data acquisition in algorithm development and forbenchmarking to multi-core architecture. The platform was shown to be flexibleenough to be used a framework for future algorithm development and education inautomatic control.

This thesis deals with processing of real data regarding waste collection. It describes select parts of the fields of statistical tests, identification of outliers, correlation analysis and linear regression. This theoretical basis is applied through the programming language Python to process the data into a form suitable for creating linear models. Final models explain between 70 \% and 85 \% variability. Finally, the information obtained through this analysis is used to specify recommendations for the waste management company.