Denna rapporten studerar effekterna i det svenska järnvägsnätet och kvaliten på de estimerade ankomsttider för tågen. Effekterna som studeras i rapporten är resultatet av hur kvaliten på ankomsttideren ändras när en positioneringsapparat introduceras i järnvägstrafiken. Rapporten studerar också hur kvaliten på uppskattningen av ankomsttidernas ändras då man reducerar antalet driftledningscentraler och centraliserar järnvägsorganisationen. Rapporten fortsätter med att rangordna dessa olika metoder och utreda vilken av de två metoderna som har störst inverkan och som är den mest fördelaktiga metoden. Ett försök görs också att sätta ett värde på hur mycket bättre varje metod, och olika kombinationer av metoderna, är i förhållande till varandra. / Denna rapporten studerar effekterna i det svenska järnvägsnätet och kvaliten på de estimerade ankomsttider för tågen. Effekterna som studeras i rapporten är resultatet av hur kvaliten på ankomsttideren ändras när en positioneringsapparat introduceras i järnvägstrafiken. Rapporten studerar också hur kvaliten på uppskattningen av ankomsttidernas ändras då man reducerar antalet driftledningscentraler och centraliserar järnvägsorganisationen. Rapporten fortsätter med att rangordna dessa olika metoder och utreda vilken av de två metoderna som har störst inverkan och som är den mest fördelaktiga metoden. Ett försök görs också att sätta ett värde på hur mycket bättre varje metod, och olika kombinationer av metoderna, är i förhållande till varandra. / Email: jo.persson@gmail.com

Being in the era of Big data, the applicability and importance of data-driven models like artificial neural network (ANN) in the modern statistics have increased substantially. In this dissertation, our main goal is to contribute to the development and the expansion of these ANN models by incorporating Bayesian learning techniques. We have demonstrated the applicability of these Bayesian ANN models in interdisciplinary research including health and cybersecurity. Breast cancer is one of the leading causes of deaths among females. Early and accurate diagnosis is a critical component which decides the survival of the patients. Including the well known ``Gail Model", numerous efforts are being made to quantify the risk of diagnosing malignant breast cancer. However, these models impose some limitations on their use of risk prediction. In this dissertation, we have developed a diagnosis model using ANN to identify the potential breast cancer patients with their demographic factors and the previous mammogram results. While developing the model, we applied the Bayesian regularization techniques (evidence procedure), along with the automatic relevance determination (ARD) prior, to minimize the network over-fitting. The optimal Bayesian network has 81\% overall accuracy in correctly classifying the actual status of breast cancer patients, 59\% sensitivity in accurately detecting the malignancy and 83\% specificity in correctly detecting non-malignancy. The area under the receiver operating characteristic curve (0.7940) shows that this is a moderate classification model. We then present a new Bayesian ANN model for developing a nonlinear Poisson regression model which can be used for count data modeling. Here, we have summarized all the important steps involved in developing the ANN model, including the forward-propagation, backward-propagation and the error gradient calculations of the newly developed network. As a part of this, we have introduced a new activation function into the output layer of the ANN and error minimizing criterion, using count data. Moreover, we have expanded our model to incorporate the Bayesian learning techniques. The performance our model is tested using simulation data. In addition to that, a piecewise constant hazard model is developed by extending the above nonlinear Poisson regression model under the Bayesian setting. This model can be utilized over the other conventional methods for accurate survival time prediction. With this, we were able to significantly improve the prediction accuracies. We captured the uncertainties of our predictions by incorporating the error bars which could not achieve with a linear Poisson model due to the overdispersion in the data. We also have proposed a new hybrid learning technique, and we evaluated the performance of those techniques with a varying number of hidden nodes and data size. Finally, we demonstrate the suitability of Bayesian ANN models for time series forecasting by using an online training algorithm. We have developed a vulnerability forecast model for the Linux operating system by using this approach.

The specification and modulation of cell-type specific dendritic morphologies plays a pivotal role in nervous system development, connectivity, structural plasticity, and function. Regulation of gene expression is controlled by a wide variety of cellular and molecular mechanisms, of which two major types are transcription factors (TFs) and microRNAs (miRNAs). In Drosophila, dendritic complexity of dendritic arborization (da) sensory neurons of the peripheral nervous system are known to be regulated by two transcription factors Cut and Knot, although much remains unknown about the molecular mechanisms and regulatory networks via which they regulate the final arbor shape through spatio-temporal modulation of dendritic development and dynamics. Here we use bioinformatics analysis of transcriptomic data to identify putative genomic targets of these TFs with a particular emphasis on those that effect neuronal cytoskeletal architecture. We use transcriptomic, as well as data from various genomic and protein interaction databases, to build a weighted functional gene regulatory network for Knot, to identify the biological pathways and downstream genes that this TF regulates. To corroborate bioinformatics network predictions, knot putative targets, which classify into neuronal and cytoskeletal functional groups, have been experimentally validated by in vivo genetic perturbations to elucidate their role in Knot-mediated Class IV (CIV) dendritogenesis. MicroRNAs (miRNAs) have emerged as key post-transcriptional regulators of gene expression, however identification of biologically-relevant target genes for this epigenetic regulatory mechanism remains a significant challenge. To address this knowledge gap, we have developed a novel R based tool, IntramiR-ExploreR, that facilitates integrated discovery of miRNA targets by incorporating target databases and novel target prediction algorithm to arrive at high confidence intragenic miRNA target predictions. We have explored the efficacy of this tool using D.melanogaster as a model organism for bioinformatics analyses and functional validation, and identified targets for 83 intragenic miRNAs. Predicted targets were validated, using in vivo genetic perturbation. Moreover, we are constructing interaction maps of intragenic miRNAs focusing on neural tissues to uncover regulatory codes via which these molecules regulate gene expression to direct cellular development.

Diploma thesis is focused on regions' tax revenue predictions. Emphasis is placed on determining whether these predictions could be deliberately distorted by their creators. The thesis evaluates prediction of all regions with the exception region Prague City which is subject to different rules of tax revenue assignment. Besides quantifying deviations in real tax revenue collection and prediction of individual regions, evaluation and exploration of susceptibility to systematic distortion thesis also deals with the comparison regions 'errors in predictions and errors made by Ministry of Finance of the Czech Republic. Theses of deliberate misrepresentation tax revenues volume is not confirmed as result of statistical analysis performed. The idea that regions compile their predictions according to the Ministry of Finance forecasts, which are known before creating their own predictions, is also refused.

In silico prioritization of undiscovered associations can help find causal genes of newly discovered diseases. Some existing methods are based on known associations, and side information of diseases and genes. We exploit the possibility of using a neural network model, Neural inductive matrix completion (NIMC), in disease-gene prediction. Comparing to the state-of-the-art inductive matrix completion method, using neural networks allows us to learn latent features from non-linear functions of input features. Previous methods use disease features only from mining text. Comparing to text mining, disease ontology is a more informative way of discovering correlation of dis- eases, from which we can calculate the similarities between diseases and help increase the performance of predicting disease-gene associations. We compare the proposed method with other state-of-the-art methods for pre- dicting associated genes for diseases from the Online Mendelian Inheritance in Man (OMIM) database. Results show that both new features and the proposed NIMC model can improve the chance of recovering an unknown associated gene in the top 100 predicted genes. Best results are obtained by using both the new features and the new model. Results also show the proposed method does better in predicting associated genes for newly discovered diseases.

Diploma thesis focuses on information and cyber threats in 2019. It comprises theoretical basis for better understanding of the issue. Afterward the thesis describes the analysis of the current situation which combined several analyses primarily aimed on Czech companies. In the last part draft measures is created which contain predictions and preventive actions and recommendations for companies.

La variabilité phénotypique existante au sein d’une population est d’une importance cruciale ; elle permet l’adaptation à de nouvelles conditions par la sélection naturelle de traits bénéfiques. La variabilité phénotypique est le résultat du polymorphisme génétique de chaque individu, couplé à l’influence de divers facteurs environnementaux. Ces travaux ont pour objectif d’élucider quels sont les facteurs génétiques responsables de la variabilité phénotypique de chaque individu afin de comprendre comment celle-ci évolue de génération en génération et peut s’accentuer au-delà des prédispositions parentales. Finalement, les résultats obtenus seront utilisés pour prédire un phénotype à partir d’un génotype inconnu. Nous avons utilisé des techniques de phénomique et de génomique de haut débit pour décomposer avec une précision inédite la variabilité phénotypique d’une large population de souches diploïdes de Saccharomyces cerevisiae. Le génotype exact de plus de 7000 souches uniques a ainsi été obtenu via le croisement et le séquençage de souches haploïdes distinctes. Nous avons mesuré la capacité de croissance de ces souches et identifié les composants génétiques influant sur ce trait. De plus, nous avons identifié des « loci de caractères quantitatifs » additifs et non-additifs, et étudié la fréquence du phénomène d’hétérosis et ses mécanismes. Enfin, en utilisant les données phénotypiques et génotypiques de la même population de levures, nous avons pu prédire les traits de chaque individu avec une presque parfaite exactitude. Ces travaux ont ainsi permis d’identifier avec précision les facteurs génétiques modulant la variation phénotypique d’une population diploïde, et de prédire un trait à partir du génotype et de l’ensemble des données phénotypiques. En plus de ce projet, nous travaillons aussi sur l’identification des bases génétiques à l’origine de la non-viabilité des gamètes, ainsi que sur la compréhension des caractères complexes chez des souches hybrides intra-espèce. De par l’étude de 9000 gamètes séquencés issus de six hybrides différents, nous avons pour objectif de caractériser leur profil de recombinaison et d’observer quelle est l’influence du fond génétique sur ce dernier. De plus, nous avons caractérisé la capacité de croissance de ces gamètes dans neuf conditions environnementales différentes et nous prévoyons de disséquer l’architecture génétique de ces traits dans différents fonds génétiques. / The phenotypic variation between individuals in a population is of crucial importance. It allows populations to evolve to novel conditions by the natural selection of beneficial traits. Variation in traits can be caused by genetic or environmental factors. This work endeavors to study the genetic factors that underlie phenotypic variation in order to understand how variation can be created from one generation to the next; to know what genetic mechanisms are most prominent; to learn how variation can extend beyond the parents; and finally, to use this in order to predict phenotypes of unknown genetic constellations. We used large scale phenomics and genomics to give an unprecedented decomposition of the phenotypic variation in a large population of diploid Saccharomyces cerevisiae strains. Constructing phased outbred lines by large scale crosses of sequenced haploid strains allowed us to infer the genetic makeup of more than 7,000 colonies. We measured the growth of these strains and decomposed the phenotypic variation into its genetic components. In addition, we mapped additive and nonadditive quantitative trait loci, we investigated the occurrence of heterosis and its genetic basis, and using the same populations we used phenotypic and genetic data to predict traits with near perfect accuracy. By using the phased outbred line approach, we succeeded in giving a conclusive account of what genetic factors define phenotypic variation in a diploid population, and in accurately predicting phenotypes from genetic and phenotypic data. Beyond the phased outbred line project, I am currently investigating the genetic basis of gamete inviability and complex traits in intraspecies yeast hybrids. Using 9,000 sequenced gametes from six different hybrids we aim to characterize their recombination landscape and how the genetic background influences it. Furthermore, we have phenotyped these gametes in nine conditions and will dissect the genetic architecture of these traits across multiple genomic backgrounds.

Comprendre le comportement des utilisateurs est fondamentale pour créer des systèmes de communication efficaces. Dévoiler les interactions complexes entre les utilisateurs dans le monde réel ou en ligne, déchiffrer leurs activité sur Internet, ou comprendre la mobilité humaine - toutes les formes des activités - peuvent avoir un impact direct sur la performance d'un réseau de communication. Mais l'observation du comportement de l'utilisateur n'est pas suffisant. Pour transformer l'information en connaissance utile, il faut cependant aller au-delà de l'observation et l' explication du passé et de créer des modèles permettant de prédire le comportement. Dans cette thèse, nous nous concentrons sur le cas des utilisateurs qui consomment du contenu dans leurs trajets quotidiens, en particulier lorsque la connectivité est faible ou intermittente. Nous considérons que les utilisateurs peuvent communiquer entre eux en utilisant l'infrastructure mais aussi directement en utilisant les communications opportunistes. Nous proposons de nouvelles perspectives sur la façon d'utiliser des information sur le comportement des utilisateurs dans la conception de solutions plus efficaces pour les communications mobiles opportunistes. En particulier, nous mettons en avant que le comportement des utilisateurs, à la fois en termes de consommation de contenu et les contacts entre les utilisateurs mobiles, peut être utilisé pour élaborer des stratégies dynamiques de réplication de données. / Understanding user behavior is fundamental in the design of efficient communication systems. Unveiling the complex online and real-life interactions among users, deciphering online activity, or understanding user mobility patterns all forms of user activity have a direct impact on the performance of the network. But observing user behavior is not sufficient. To transform information in valuable knowledge, one needs however to make a step forward and go beyond observing and explaining the past to building models that will predict future behavior. In this thesis, we focus on the case of users consuming content on the move, especially when connectivity is poor or intermittent. We consider both traditional infrastructure-based communications and opportunistic device-to-device transfers between neighboring users. We offer new perspectives of how to use additional information about user behavior in the design of more efficient solutions for mobile opportunistic communications. In particular, we put forward the case that the collective user behavior, both in terms of content consumption and contacts between mobile users, can be used to build dynamic data replication strategies.

The rehabilitation of dams often requires spillway capacity upgrades. Replacing a less hydraulically efficient linear weir with a labyrinth weir can be an effective way to increase discharge efficiency (discharge at a given upstream head) for a fixed-width channel. Labyrinth weirs are linear weirs folded in plan view to increase total spillway crest length (which in turn increases discharge efficiency within a channel). Labyrinth weirs potentially have limitless geometric configurations. This study was performed to analyze the effects of varying certain geometric parameters on discharge efficiency and design method predictions. Due to limited cross-sectional flow area near the upstream apex, labyrinth weirs experience nappe collision and local submergence that potentially reduce discharge efficiency. The increase of upstream apex width may be a feasible method to decrease the negative effects of nappe interference, which in turn may increase discharge efficiency. This was analyzed in this study by testing a series of eight laboratory scaled labyrinth weirs (with sidewall angles of 12°), with various upstream apex widths. Upstream apex width tests were performed in a fixed and varied channel width setting. The design method developed by Crookston and Tullis is based on laboratory scaled physical models. This method is very useful in the estimation of performance for geometrically similar prototype labyrinth weirs. However, due to difficulty in obtaining data on completed prototype weirs, design method predictions are rarely verified. To help validate Froude scaling and design method predictions of prototype weirs, a series of physical model tests (with sidewall angles of 15°) were performed with varying scale sizes (0.5 to 3.0 compared to the size of weir used in the design method). To expand the applicability of the design method to common geometric variations, tests were performed on weirs of varying weir height and cycle width (with sidewall angles of 15°). These variations were applied independently and analyzed to determine their effects on discharge efficiency and design method predictions. A correction factor is then presented to be used in conjunction with Crookston and Tullis’s design method for these geometric variations. All conclusions are presented in this thesis.

No description available.