This thesis focuses on the main processes in positive ground flashes and the distant lightning environment for both positive and negative ground flashes. It presents the characteristics of the preliminary breakdown pulses (PBPs), the characteristics of the electric field pulses observed during leader propagation, and the characteristics of the electric fields produced by the first and the subsequent return strokes. It also features the observations of distant positive and negative ground flashes at distances up to 1000 km. The results were based on electric field measurements conducted remotely during summer thunderstorms in Sweden in 2014. We found that the majority of the positive ground flashes were preceded by PBPs. Some were preceded by more than one PBP train, and the parameter values for the subsequent PBP trains were found to be smaller than the values for the first PBP train. Three types of PBPs were also identified. The results suggest that the PBPs in positive ground flashes during summer thunderstorms in Sweden are weak, and the inverted dipole charge cloud configuration is consistent with our observation. A small percentage of these positive ground flashes were observed to be preceded by pronounced leader pulses. The presence of these pulses suggests that the leaders propagate in a stepped manner. We inferred that these pulses were due to the upward-connecting negative leader since their characteristics were similar to those of a negative stepped leader. On the basis of the leader pulses’ time of initiation and the average speed of the leader, we estimated the distance travelled by the leader. One case of positive return stroke preceded by negative leader pulses was also observed, and the occurrence of these pulses was the first in positive ground flashes to be reported. The majority of these positive ground flashes were found to be single-stroke. Comparison between the first and the subsequent return strokes showed that the average durations of the subsequent stroke parameters were smaller than that of the first strokes. The distances reported by the lightning location system suggest that the subsequent strokes probably created new terminations to ground. Two possible reasons were given to explain the reason for the shorter duration of the subsequent return strokes parameters compared to the first strokes. Finally, observations of distant positive and negative ground flashes showed that the electric field waveforms have a typical shape, like a distorted ‘W’ (or distorted ‘M’ for negative ground flashes) followed by small oscillations. These small oscillations were more pronounced in negative ground flashes, especially at a greater distance. The heights of the ionospheric reflections estimated for both ground flashes were found to likely correspond to a D-layer of the ionosphere. Two possible reasons were suggested for the small oscillations observed in the waveforms.

L'analyse automatique de différences anatomiques en neuroimagerie a de nombreuses applications pour la compréhension et l'aide au diagnostic de pathologies neurologiques. Récemment, il y a eu un intérêt croissant pour les méthodes de classification telles que les machines à vecteurs supports pour dépasser les limites des méthodes univariées traditionnelles. Cette thèse a pour thème l'apprentissage automatique pour l'analyse de populations et la classification de patients en neuroimagerie. Nous avons tout d'abord comparé les performances de différentes stratégies de classification, dans le cadre de la maladie d'Alzheimer à partir d'images IRM anatomiques de 509 sujets de la base de données ADNI. Ces différentes stratégies prennent insuffisamment en compte la distribution spatiale des \textit{features}. C'est pourquoi nous proposons un cadre original de régularisation spatiale et anatomique des machines à vecteurs supports pour des données de neuroimagerie volumiques ou surfaciques, dans le formalisme de la régularisation laplacienne. Cette méthode a été appliquée à deux problématiques cliniques: la maladie d'Alzheimer et les accidents vasculaires cérébraux. L'évaluation montre que la méthode permet d'obtenir des résultats cohérents anatomiquement et donc plus facilement interprétables, tout en maintenant des taux de classification élevés. / Brain image analyses have widely relied on univariate voxel-wise methods. In such analyses, brain images are first spatially registered to a common stereotaxic space, and then mass univariate statistical tests are performed in each voxel to detect significant group differences. However, the sensitivity of theses approaches is limited when the differences involve a combination of different brain structures. Recently, there has been a growing interest in support vector machines methods to overcome the limits of these analyses.This thesis focuses on machine learning methods for population analysis and patient classification in neuroimaging. We first evaluated the performances of different classification strategies for the identification of patients with Alzheimer's disease based on T1-weighted MRI of 509 subjects from the ADNI database. However, these methods do not take full advantage of the spatial distribution of the features. As a consequence, the optimal margin hyperplane is often scattered and lacks spatial coherence, making its anatomical interpretation difficult. Therefore, we introduced a framework to spatially regularize support vector machines for brain image analysis based on Laplacian regularization operators. The proposed framework was then applied to the analysis of stroke and of Alzheimer's disease. The results demonstrated that the proposed classifier generates less-noisy and consequently more interpretable feature maps with no loss of classification performance.

Dagens samhälle har fått en explosiv utveckling som förverkligar mycket som för bara några år sedan var inte mer än fantasier.  Dagens utvecklingsförsprång ställer oss, byggnadskonstruktörer, inför rejäla utmaningar. Den globala folktillväxten ökar väsentligt vilket leder till tätbefolkade städer. Detta utvecklar ett stort utrymmesbehov hos många av oss. Allt detta resulterar i att efterfrågan på stora och öppna planlösningar ökar markant. En av dagens tendenser är att beställare och arkitekter har en benägenhet att tänja på gränserna på maximala spännvidder mellan bärande betongväggar, för bjälklagstjockleken 250 mm. Detta är ett tillfredsställande mått för att klara ljudklass B. Dessutom är det opraktiskt att variera bjälklagstjocklekar inom ett projekt, därför vill man ha uniformitet med samma tjocklek över projektet. För att vi ska kunna förverkliga vårt uppdrag har vi varit tvungna att genomgå en lång beräknings- och undersökningsprocess. I våra beräkningar har vi lagt fokus på två upplagsfall. Det första upplagsfallet ”fri-inspänd” och det andra fallet ”inspänd-yttre gavelvägg”. Första fallet har varit det värsta fallet i och med att vi bara har ett stöd som måste bära hela betongbjälklaget, vilket har varit en stor utmaning. Andra fallet var dock betydligt enklare på grund av de två stöden som utgjorde en stor del av ”arbetet” och lyfter upp bjälklaget, hela tyngden vilade inte på armeringen som i föregående fall. Inte bara spännvidden skall klaras utan även angiven sprickvidd på 0,3 mm. Examensarbetet består av förklarande fakta som är strikt relaterad till efterföljande beräkningar. Alla beräkningar har utförts för hand, utan programstöd. / Today's society has received a degenerate development embodying much that just a few years ago was no more than fantasies. This development sets us, structural engineers, facing real challenges. The global population growth increases significantly leading to densely populated cities. This develops a large space need for many of us. All this results in the increasing demand for large and open floor plans significantly. One of the current trends is that the clients and architects have a tendency to push the limits on maximum spans between bearing concrete walls, slabs for thickness 250 mm. This is a satisfactory measure of the concrete content to manage audio class B. Moreover, it is impractical to vary the slab thickness within a project, so he wants to have uniformity with the same thickness over the project. For us to be able to realize these long spans between bearing walls, we have been forced to undergo a long calculation and examination process. To begin with, we have studied the company's requirements and preferences, based on that, we started joists analysis. In our calculations, we have laid emphasis on two cases. The first circulation fall "free - clamped" and the second, "clamped- outer end wall." The first case has been the worst case, in that we only have one support that must bear the entire concrete slab, which has been a major challenge. Second case was considerably easier due to the two supports which made a large part of "work" and lifts the slab, the full weight rested not on the reinforcement as in the previous case. Not just the span must be met, but also given crack width of 0.3 mm. The thesis consists of explanatory facts that are strictly related to the subsequent calculations. All calculations have been performed by hand, without program support.