Sous l'ère moderne, où le commerce n'a plus de frontières physiques, les catastrophes causant des dommages à grande échelle se sont multipliées. Lorsque de telles catastrophes surviennent, l'utilisation de procédures de recours collectif pour juger de l'ensemble des litiges semble être la solution indiquée. Pourtant, l'expérience nous démontre qu'il n'en est rien... Bien que le même événement cause des dommages à de multiples victimes, le recours judiciaire demeure l'exercice d'un droit personnel. Le tribunal devra alors rechercher l'application de règles matérielles spécifiques à ce recours. En l'absence de règles spécifiques, les règles de conflit de lois du tribunal saisi seront applicables. Or, ces règles ne sont pas uniformes et entraînent une multiplicité de lois applicables. Conséquemment, il est opportun de s'interroger sur la pertinence de réunir des recours individuels en un seul recours. L'expérience américaine nous démontre les difficultés engendrées par la multiplicité de lois applicables à un recours collectif, bien que des solutions législatives et jurisprudentielles aient été proposées. Ces solutions demeurent imparfaites, tant au niveau de leur constitutionalité que du respect des droits fondamentaux. Devant la complexité occasionnée par le regroupement de ces litiges, les tribunaux sont réfractaires à leur autorisation, dès lors qu'une multiplicité de lois applicables est démontrée. Ainsi, la solution juridique la mieux adaptée aux recours collectifs internationaux demeure celle de saisir les tribunaux, de plusieurs recours collectifs distincts; chacun regroupé en fonction de la loi applicable à chaque recours, déterminée, au préalable, par l'analyse des règles de conflit de lois du tribunal saisi. / When a catastrophic accident occurs, involving numerous victims from different states, we could, instinctively believe that justice will be best served by consolidation of all claims arising from such event. However, this idea isn’t completely right... Ironically, while catastrophic accident is a collective event, the claims pertaining such event remain a personal right of action. Therefore, the court might select the law applicable to each claim in accordance with its own conflict of laws rules. Because those rules aren't similar among states, the law applicable to each claim will differ accordingly. Consequently, the judge hearing a multi-state class action pertaining the same event will need to deal with multiplicity of laws. Historically, American federal courts have almost systematically rejected consolidation of claims when a multiplicity of laws appears. Many solutions to overwhelm problems arising from multiplicity of laws have been raised by the legal communauty. To date, none of those solutions seems perfect. It brings the conclusion that the idea of an international class action to resolve in a single lawsuit, all claims resulting from a single accident, might be an utopia. However, efficiency of class actions proceedings remains. As soon as litigants determine, prior to the certification request, the law applicable to each individual claims, in accordance with the relevant conflict-of-laws rules, in order to join all claims that are similarly conducted by the same law. In doing so, problems related to diversity of laws in class actions are avoided.

Les émulsions de Pickering ont suscité un intérêt croissant dans de nombreux domaines de la recherche en raison de leur grande stabilité et versatilité. Une attention particulière a été accordée à la fabrication des systèmes complexes et originaux qui peuvent être obtenus avec différentes particules. Cette étude a consisté dans l'analyse des différents paramètres physico-chimiques des particules, des milieux liquides et des systèmes dispersées, et leur relation avec le comportement mécanique et la structure des émulsions afin de prédire et de moduler les caractéristiques de ces dernières. Nous avons étudié plus particulièrement pour la première fois, le diagramme de phase concernant les inversions de phase du type catastrophique et transitionnelle des émulsions de Pickering. Nous avons utilisé des particules de silice avec des structures et hydrophobicités différentes. En particulier, nous avons montré que le mélange de particules de différente hydrophobicités peut moduler finement l'inversion de phase aussi bien que les propriétés rhéologiques et structurales des émulsions. La fabrication de membranes à partir des émulsions de Pickering précédentes a été proposée comme un exemple de l'utilisation de ces systèmes modèles pour la conception de matériaux complexes. / Pickering emulsions have gained interest in many fields of research due their properties like higher stability and versatility. Special attention has been given to the processing of complex and original systems which can be obtained by using different particles. This study consists in the analysis of the different physicochemical parameters of particles, liquid media as well dispersion systems, and their relationship with emulsions structural and mechanical behavior in order to predict and modulate the emulsions characteristics. We studied extensively for the first time the phase diagram of catastrophic and transitional phase inversion of Pickering emulsions. We used silica particles with different structure and hydrophobicity. In particular we showed that mixing particles with different hydrophobicity can finely modulate the phase inversion as well the rheological and structural properties of the emulsions. The manufacturing of emulsified membranes based on previous Pickering emulsions was proposed as an example of the use of these systems as templates for the design of complex materials.

Dans la plupart des situations de guerre utilisant des enfants, tout au moins dans les configurations actuelles, c’est une partie du groupe-communauté lui-même qui porte la transgression. Il apparaît souvent dans ces guerres une dimension inavouable liée à la dynamique interne du groupe. Dans ces configurations il s’agit d’une organisation collective du massacre par une partie du groupe, sur la base de codes communs et partagés saisissable au travers de la théorie des catastrophes (Thom, 1961) comme des opérateurs de bifurcation. Comme si d’une part, la figure de l’intrus (Duez., 2003) qui peut se comprendre comme produit du travail de figuration, de mise en forme de l’objet perdait sa principale fonction structurante, et d’autre part, le sujet perdait toute possibilité de vivre intimement le groupe comme son espace psychique, comme la scène destinée à transférer immédiatement ses différents conflits et enjeux psychiques, re-convoquant les scénalités archaïques et invitant à définir de nouvelles formes à investir. Ceci vient complexifier tout travail de symbolisation. Dans ces contextes de transgressions, où l’enfant est au prise avec l’autre-semblable devenu nocif et porteur de la mort, l’émergence de la figure de l’enfant-soldat obéit à une dynamique dont la force (celle utilisée par le chef de guerre) n’est pas le seul paramètre. Cette figure est logée dans le groupe-combattant qui est une émanation méta-défensives du groupe-communauté. / In most current configurations of war front using children, we use to find a part of the community-group bearing itself the transgression. It often appears in these wars, a shameful dimension related to the internal dynamics of the group killing itself. By the fact, in these configurations, we have a collective massacre committed by part of Group, on the basis of common and shared social rules and anthropological codes. The catastrophe theory (Thom, 1961) is an opportunity to explain those phenomena using attractive and repulsive forces in individual an groupal catastrophic bifurcation. It seems, on one hand, the “alien intrusion” (Duez B. 2003) which can be considered as the result of the psychic figuration work out and shaping of the psychic object, loses its main structuring function. On the other hand, the individual lost any opportunity to intimately experience the group as a psychic area on which he can transfer his immediate psychic conflicts and issues. Then, archaic scenes take the rule and invite the individual to find new internal form to deal with for symbolisation process. In these transgressions contexts, where children have to deal with harmful adult bearer of death, the rising of Child Soldier phenomena obeys a dynamic in witch strength (used by lord of war) is the common but not the unique parameter to consider. Child Soldier emerge in the war group which is a meta-defensive organization of the community reacting to the state of war.

Today, there is a growing number of buildings in a neighborhood and business parks that are utilizing renewable energy generation, to reduce their electric bill and carbon footprint. The most current way of implementing a renewable energy generation is to use solar panels or a windmill to generate power; then use a charge controller connected to a battery bank to store power. Once stored, the user can then access a clean source of power from these batteries instead of the main power grid. This type of power structure is utilizing a single module system in respect of one building. As the industry of renewable power generation continues to increase, we start to see a new way of implementing the infrastructure of the power system. Instead of having just individual buildings generating power, storing power, using power, and selling power there is a fifth step that can be added, sharing power. The idea of multiple buildings connected to each other to share power has been named a microgrid by the power community. With this ability to share power in a microgrid system, a catastrophic event which cause shutdowns of power production can be better managed. This paper then discusses the data from simulations and a built physical model of a resilient microgrid utilizing these principles.

While deep learning has proved to be useful in a large variety of tasks, a limitation remains of needing all classes and samples to be present at the training stage in supervised problems. This is a major issue in the field of biomedical imaging since keeping samples in the training sets consistently is often a liability. Furthermore, this issue prevents the simple updating of older models with only the new data when it is introduced, and prevents collaboration between companies. In this work, we examine an array of Continual Learning approaches to try to improve upon the baseline of the naive finetuning approach when retraining on new tasks, and achieve accuracy levels similar to the ones seen when all the data is available at the same time. Continual learning approaches with which we attempt to mitigate the problem are EWC, UCB, EWC Online, SI, MAS, CN-DPM. We explore some complex scenarios with varied classes being included in the tasks, as well as close to ideal scenarios where the sample size is balanced among the tasks. Overall, we focus on X-ray images, since they encompass a large variety of diseases, with new diseases requiring retraining. In the preferred setting, where classes are relatively balanced, we get an accuracy of 63.30 versus a baseline of 53.92 and the target score of 66.83. For the continued training on the same classes, we get an accuracy of 35.52 versus a baseline of 27.73. We also examine whether learning rate adjustments at task level improve accuracy, with some improvements for EWC Online. The preliminary results indicate that CL approaches such as EWC Online and SI could be integrated into radiography data learning pipelines to reduce catastrophic forgetting in situations where some level of sequential training ability justifies the significant computational overhead. / Även om djupinlärning har visat sig vara användbart i en mängd olika uppgifter, kvarstår en begränsning av att behöva alla klasser och prover som finns på utbildningsstadiet i övervakade problem. Detta är en viktig fråga inom området biomedicinsk avbildning eftersom det ofta är en belastning att hålla prover i träningsuppsättningarna. Dessutom förhindrar det här problemet enkel uppdatering av äldre modeller med endast nya data när de introduceras och förhindrar samarbete mellan företag. I det här arbetet undersöker vi en rad kontinuerliga inlärningsmetoder för att försöka förbättra baslinjen för den naiva finjusteringsmetoden vid omskolning på nya uppgifter och närma sig noggrannhetsnivåer som de som ses när alla data är tillgängliga samtidigt. Kontinuerliga inlärningsmetoder som vi försöker mildra problemet med inkluderar bland annat EWC, UCB, EWC Online, SI. Vi utforskar några komplexa scenarier med olika klasser som ingår i uppgifterna, samt nära idealiska scenarier där exempelstorleken balanseras mellan uppgifterna. Sammantaget fokuserar vi på röntgenbilder, eftersom de omfattar ett stort antal sjukdomar, med nya sjukdomar som kräver omskolning. I den föredragna inställningen får vi en noggrannhet på 63,30 jämfört med en baslinje på 53,92 och målpoängen på 66,83. Medan vi för den utökade träningen på samma klasser får en noggrannhet på 35,52 jämfört med en baslinje på 27,73. Vi undersöker också om justeringar av inlärningsfrekvensen på uppgiftsnivå förbättrar noggrannheten, med vissa förbättringar för EWC Online. De preliminära resultaten tyder på att CL-metoder som EWC Online och SI kan integreras i rörledningar för röntgendatainlärning för att minska katastrofal glömska i situationer där en viss nivå av sekventiell utbildningsförmåga motiverar den betydande beräkningskostnaden.

Medical doctors understaffing is becoming a compelling problem in many healthcare systems. This problem can be alleviated by utilising Computer-Aided Diagnosis (CAD) systems to substitute doctors in different tasks, for instance, histopa-thological image classification. The recent surge of deep learning has allowed CAD systems to perform this task at a very competitive performance. However, a major challenge with this task is the need to periodically update the models with new data and/or new classes or diseases. These periodical updates will result in catastrophic forgetting, as Convolutional Neural Networks typically requires the entire data set beforehand and tend to lose knowledge about old data when trained on new data. Incremental learning methods were proposed to alleviate this problem with deep learning. In this thesis, two incremental learning methods, Learning without Forgetting (LwF) and a generative rehearsal-based method, are investigated. They are evaluated on two criteria: The first, capability of incrementally adding new classes to a pre-trained model, and the second is the ability to update the current model with an new unbalanced data set. Experiments shows that LwF does not retain knowledge properly for the two cases. Further experiments are needed to draw any definite conclusions, for instance using another training approach for the classes and try different combinations of losses. On the other hand, the generative rehearsal-based method tends to work for one class, showing a good potential to work if better quality images were generated. Additional experiments are also required in order to investigating new architectures and approaches for a more stable training.

Dans l’apprentissage fédéré, un modèle global est appris en agrégeant les mises à jour du modèle calculées à partir d’un ensemble de nœuds clients, un défi clé dans ce domaine est l’hétérogénéité des données entre les clients qui dégrade les performances du modèle. Les algorithmes d’apprentissage fédéré standard effectuent plusieurs étapes de gradient avant de synchroniser le modèle, ce qui peut amener les clients à minimiser exagérément leur propre objectif local et à s’écarter de la solution globale. Nous démontrons que dans un tel contexte, les modèles de clients individuels subissent un oubli catastrophique par rapport aux données d’autres clients et nous proposons une approche simple mais efficace qui modifie l’objectif d’entropie croisée sur une base par client en repondérant le softmax de les logits avant de calculer la perte. Cette approche protège les classes en dehors de l’ensemble d’étiquettes d’un client d’un changement de représentation brutal. Grâce à une évaluation empirique approfondie, nous démontrons que notre approche peut atténuer ce problème, en apportant une amélioration continue aux algorithmes d’apprentissage fédéré standard. Cette approche est particulièrement avantageux dans les contextes d’apprentissage fédéré difficiles les plus étroitement alignés sur les scénarios du monde réel où l’hétérogénéité des données est élevée et la participation des clients à chaque cycle est faible. Nous étudions également les effets de l’utilisation de la normalisation par lots et de la normalisation de groupe avec notre méthode et constatons que la normalisation par lots, qui était auparavant considérée comme préjudiciable à l’apprentissage fédéré, fonctionne exceptionnellement bien avec notre softmax repondéré, remettant en question certaines hypothèses antérieures sur la normalisation dans un système fédéré / In Federated Learning, a global model is learned by aggregating model updates computed from a set of client nodes, a key challenge in this domain is data heterogeneity across clients which degrades model performance. Standard federated learning algorithms perform multiple gradient steps before synchronizing the model which can lead to clients overly minimizing their own local objective and diverging from the global solution. We demonstrate that in such a setting, individual client models experience a catastrophic forgetting with respect to data from other clients and we propose a simple yet efficient approach that modifies the cross-entropy objective on a per-client basis by re-weighting the softmax of the logits prior to computing the loss. This approach shields classes outside a client’s label set from abrupt representation change. Through extensive empirical evaluation, we demonstrate our approach can alleviate this problem, providing consistent improvement to standard federated learning algorithms. It is particularly beneficial under the challenging federated learning settings most closely aligned with real world scenarios where data heterogeneity is high and client participation in each round is low. We also investigate the effects of using batch normalization and group normalization with our method and find that batch normalization which has previously been considered detrimental to federated learning performs particularly well with our re-weighted softmax, calling into question some prior assumptions about normalization in a federated setting

Solvent-free extrusion emulsification (SFEE) is a recently developed process for producing submicron particles with high viscosity polymers inside a twin-screw extruder without the use of hazardous solvents. Its dependency on a catastrophic phase inversion makes the process knowingly sensitive to a variety of formulation and operational variables, causing a narrow window of production. The purpose of this thesis was to investigate and improve process stability as well as widening operational window. Transient effects of the start-up procedure was investigated by considering the process stability and particle size distribution. The transient sensitivity corresponded to the residency of material in the dispersion zone. When a sub-optimal water/surfactant fraction was allowed to produce an undesired polymer-water (thick lamella) morphology, this morphology continued to persist until the critical first half of the dispersion zone was purged of existing mass. Lot to lot variability of polyester resin was used to investigate the sensitivities of the SFEE process more deeply to better understand the mechanism involved. In this case, acid number was shown to have a significant effect on the initial amount of water needed in the dispersion zone for phase inversion, resulting in an emulsification boundary dependent on the resin acid number. In fact, a significant correlation was found between the acidic end groups of the resin and the maximum amount of water content that could be used in the dispersion zone. The effect of feed rate, screw speed, dispersion length, and surfactant concentration were studied for their individual influence on widening the emulsification boundary. The most significant improvement was observed by applying a longer dispersion length or lower feed rate because both significantly increase the residence time. The effect of residence time on the emulsification boundary was attributed to the total strain imposed on the polymer/water mixture which was related to interfacial growth in the dispersion zone. / Dissertation / Doctor of Philosophy (PhD)

No description available.

This study is concerned with the nature of suffering as experienced by women struggling with problems related to the body. Since the body is viewed to be integral in the formation of a woman's identity and self-concept, any illness, ailment or deficiency associated with it may lead the woman to experience pain and suffering. To explore meanings of personal suffering related to the body three contexts have been chosen. These are familial breast cancer, eating disorders and infertility. Common themes of suffering that were co-constructed in the interviews between six women participants and myself form the basis of this study. These themes emerged through the process of social constructionism and dialogue. Through the process of language, personal realities and meanings were discussed and shared to elicit a greater understanding of the nature of suffering. A qualitative approach, using the case study method, was also adopted to provide rich descriptions of the different experiences with suffering. The case study presentations illustrate the linguistic domain between the participants and myself. It is hoped that the information presented in this study will contribute to a therapeutic understanding of personal suffering as experienced by women. / Psychology / D. Litt et Phil.