Plant Defensins type 1 (PDF1) sont principalement décrites pour leur rôle dans l'immunité innée en réponse à des attaques pathogènes via l'activation de la voie de signalisation de l'éthylène (Et) et de l'acide jasmonique (JA). Les défensines PDF1 du genre Arabidopsis sont également impliquées dans la tolérance cellulaire au zinc chez la levure. In planta, de nombreux résultats mettent en évidence une corrélation entre la forte accumulation des transcrits AhPDF1 et leur contribution dans la tolérance à un excès de zinc. Dans cette étude, l'analyse du transcriptome (qRT-PCR) révèle que les paralogues PDF1s, aussi bien chez A. thaliana que chez A. halleri sont très peu voire pas du tout sensibles au zinc. Toutefois, il y a une spécialisation des PDF1s en réponse à l'activation de la voie de l'acide jasmonique dans le genre Arabidopsis. De plus, la contribution fonctionnelle des membres de la famille PDF1s dans la tolérance au zinc a été caractérisée chez A. thaliana à l'aide d'une approche génétique combinant des mutants KO après insertion d'un ADN-T et la technologie de miRNA artificiel. L'étude de ces mutants souligne par ailleurs la diversité fonctionnelle au sein de la famille des défensines AtPDF1s qui ne confèrent pas toutes la tolérance au zinc. En effet, une diversité de déterminants moléculaires des PDF1s a été mise en évidence lors de cette étude. La forte accumulation des PDF1s n'est pas l'unique paramètre requis pour la tolérance au zinc. Il faut également considérer la spécificité de tissu où s'expriment ces PDF1s. A ces considérations s'ajoutent aussi des régulations post-transcriptionnelles et post-traductionnelles. L'étude de ces modifications est envisagée afin de comprendre la contribution des différentes défensines PDF1s dans la tolérance au zinc. / Plant Defensin type 1 (PDF1s) are mainly recognized for their response to pathogen attack via ethylene (Et)/jasmonate (JA) signaling activation pathway. However, PDF1s originating from Arabidopsis genus also showed their capacity to induce cellular zinc tolerance up on expression in yeast. In planta, a group of observation highlighted the correlation of AhPDF1 high transcript accumulation for their contribution to zinc tolerance. Here, transcriptomic analysis (qRT-PCR) revealed that in both A. thaliana and A. halleri species, PDF1 paralogues were barely or not at all responsive to zinc. Nevertheless, there is a species specialization of PDF1s in response to activation of JA-signaling in Arabidopsis genus. In addition, in A. thaliana, the functional contribution of PDF1 members in zinc tolerance was investigated through genetic approach. Examining combination of T-DNA insertion knockout mutant and artificial miRNA, these studies were first direct demonstration of the functional involvement of AtPDF1s in zinc tolerance. These also highlighted the functional diversity among AtPDF1s because not all of them could play a role in zinc tolerance. Indeed, a diversity of PDF1 molecular determinants for zinc tolerance in plants was underlined. Remarkably, PDF1 high transcript is not the only important parameter for zinc tolerance and PDF1 tissue specificity could be an important factor to consider. Moreover, post-transcriptional and post-translational regulation might occur. Studies on these modifications are now the further questions in order to understand the contribution of the different PDF1s to zinc tolerance.

Cette Thèse se divise en deux parties. Dans la partie mathématique, nous étudions différentes edp d'ordre supérieur à 3 issues des neurosciences avec un point de vue probabiliste. Nous démontrons une formule de FK pour une grande classe de solutions de KdV (pas seulement les n-solitons), à l'aide des déterminants de Fredholm et des transformées de Laplace d'intégrales de Skorohod itérées. Concernant les edp d'ordre supérieur à 3, les processus itérés qui consistent en la composition de deux processus indépendants, l'un correspondant à la position et l'autre au temps, sont liés à leurs solutions. En effet, nous montrons une formule de FK pour des solutions d'edp d'ordre supérieur à 3 basée sur des fonctionnelles de processus itérés, même dans le cas non Markovien, étendant ainsi les résultats existants. Nous proposons aussi un schéma numérique pour la simulation de trajectoires de diffusions itérées basé sur le schéma d'Euler, qui converge p.s., uniformément en temps, avec un taux de convergence d'ordre $1/4$. Une estimation de l'erreur est proposée. Dans la partie biologique, nous avons collecté plusieurs articles en neuroscience et d'autres domaines de biologie, où les edp précédentes sont utilisées. En particulier, on s'intéresse à la simulation et à la propagation du potentiel d'action lorsque la capacité de la membrane cellulaire n'est pas supposée constante. Ces articles ont en commun le fait qu'ils remettent en question le fameux modèle d'Hodgkin-Huxley datant des années cinquante. En effet, même si ce modèle a été très efficace pour la compréhension du signal neuronal, il ne prend pas en compte tous les phénomènes résultants de la propagation du potentiel d'action. / This Thesis consists of two parts. In the mathematical part we study Korteweg--de Vries (KdV) equation and high-order pdes with a probabilistic point of view in order to obtain Feynman-Kac (FK) type formulas. This study was motivated by recent biological models. We prove a FK representation for a larger class of solutions of KdV equation (not only n-solitons), using Fredholm determinants and Laplace transforms of iterated Skorohod integrals. Regarding higher order pdes, iterated processes that consist in the composition of two independent processes, one corresponding to position and the other one to time, are naturally related to their solutions. Indeed, we prove FK formulas for solutions of high order pdes based on functionals of iterated processes even in the non Markovian case, thus extending the existing results. We also propose a scheme for the simulation of iterated diffusions trajectories based on Euler scheme, that converges a.s., uniformly in time, with a rate of convergence of order $1/4$. An estimation of the error is proposed. In the biological part, we have collected several papers in neuroscience and other fields of biology where the previous types of pdes are involved. In particular, we are interested in the simulation of the propagation of the action potential when the capacitance of the cell membrane is not assumed to be constant. These papers have in common the fact that they question the famous Hodgkin Huxley model dating back to the fifties. Indeed this model even if it has been very efficient for the understanding of neuronal signaling does not take into account all the phenomena that occur during the propagation of the action potential.

This master thesis is focused on an analysis of the impact of corporate board diversity on firm performance using evidence from the Czech Republic. Return on assets, EBIT margin, EBITDA margin, and net profit margin were used as the performance measures. Three types of board diversity were examined in this thesis - gender, nationality, and age. Gender and nationality diversity were both represented by the percentage of women or foreigners on board, and by a dummy variable indicating the presence of women or a foreigner on board. The age diversity was measured by the average age of all board members. Some additional control variables were included in the models such as firm age, firm size, liquidity, etc. Our research is using data from 114 Czech enterprises between the years 2010 and 2018. The fixed-effects or the random-effects method was used to determine the relationship between board diversity and firm performance. The results show that there is a significant and positive relationship between the presence of women on board and firm performance. The average age of board members has a significant negative effect on all dependent variables. Nationality diversity has a mixed effect on firm performance. JEL Classification D22, G32, G34, J15, J16 Keywords Board diversity, firm performance,...

The purpose of this study was to explore and describe the determinant factors affecting adherence to antiretroviral therapy among HIV infected patients in Addis Ababa, Ethiopia. A cross-sectional study design was used and data were collected by interviewing 290 study participants from two health facilities using structured questionnaire. The research finding revealed 80.0% of the study participants had optimal combined adherence to dose, schedule and dietary instructions in the past three days. And, the non adherence rate was 20.0%. In multivariate analysis only WHO clinical stage, change of ARV medication, knowledge about HIV disease and ART, and use of reminders were found to be independently associated with adherence to antiretroviral therapy. The most common reasons for missing HIV medications in the past one month were forgetfulness (35.1%), being busy with other things (17.5%), and running out of pills (10.5%). Adherence improving interventions should be emphasized to address multi-faceted problems. This study recommends setting of convenient appointment schedule, disclosure of one's HIV status, maintaining confidentiality of patient-related information, enhancing patient-provider relationship, use of reminders including SMS text messages, and engagement of PLHIV in adherence improving interventions through peer support, and providing regular health education to the PLHIV to improve adherence of patients to ART / Health Studies / M.A. (Public Health)

Object detection has been a fundamental part of many emerging technologies, such as autonomous vehicles, robotics, and security. As deep learning is the main reason behind the leap of performance in object detection, it has mostly been associated with a post-processing step of non-maximum suppression (NMS) to reduce the number of resulting bounding boxes output from the network to, ideally, one box per object. As non-maximum suppression blindly suppress the overlap with a pre-defined threshold, it introduces the problem of suppressing false negatives in crowded scenes by choosing a high threshold, or vice versa. This problem is critical, especially in the autonomous vehicle industry, as this concerns the safety of passengers. The problem of the machine understanding whether these bounding boxes belong to the same object or two near-by objects is still not directly solvable. Although a lot of previous research tried to invent a new box-reduction method, every method has its own drawbacks while solving the problem. That is why, until now, many researchers are still using non‐maximum suppression. In this research, a literature review was carried out to determine the best NMS alternatives. Then, an approach for box reduction based on determinantal point process (DPP) was implemented. Furthermore, an evaluation pipeline was introduced for experimental analysis for the differences between NMS and DPP. Although NMS shows a better performance in terms of precision and recall, DPP chooses better fitting bounding boxes. / Objektdetektering har varit en grundläggande del av många nya tekniker, t.ex. autonoma fordon, robotik och säkerhet. Eftersom djupinlärning är den främsta orsaken till den stora prestandaskillnaden vid objektsdetektering har den oftast varit förknippad med ett efterbehandlingssteg med icke-maximal undertryckning (NMS) för att minska antalet resulterande avgränsande rutor som produceras av nätverket till, idealt sett, en ruta per objekt. Eftersom icke-maximal undertryckning blint undertrycker överlappningen med ett fördefinierat tröskelvärde, uppstår problemet med att undertrycka falskt negativa resultat i överfulla scener genom att välja ett högt tröskelvärde, eller tvärtom. Detta problem är kritiskt, särskilt inom industrin för autonoma fordon, eftersom det gäller passagerarnas säkerhet. Problemet med att maskinen ska förstå om dessa avgränsande rutor tillhör samma objekt eller två närliggande objekt är fortfarande inte direkt lösbart. Även om man i tidigare forskning har försökt hitta en ny metod för att reducera boxar, har varje metod sina egna nackdelar när den löser problemet. Det är därför som många forskare fram till nu fortfarande använder sig av icke-maximalt undertryckande. I denna forskning gjordes en litteraturstudie för att fastställa de bästa NMS-alternativen. Därefter implementerades en metod för boxförminskning baserad på determinant punktprocess (DPP). Dessutom infördes en utvärderingsledning för experimentell analys av skillnaderna mellan NMS och DPP. Även om NMS visar en bättre prestanda när det gäller precision och återkallande, väljer DPP bättre passande avgränsande lådor.

Changes linked to forest management institutions in diverse communities and cultural settings continue to attract research interest. However, comparative insights on their manifestations are lacking in Former British and Former French Cameroon. We trace the evolution of forest management institutions in the Kilum-Ijim and Santchou landscapes of Cameroon and analyse their compliance determinants, using key informant interviews (n = 12), focus group discussions (n = 6) and household surveys (n = 150). The results revealed a fairly stable culturally embedded institutional landscape in Kilum-Ijim, and a significant multiplication of structural institutions for Santchou. The logistic regression showed that perception and compliance with forest management institutions substantially depend on educational attainment, migration, membership in organisations, length of stay in the area and proximity of respondents to protected areas. The conclusion drawn is that landscapes which came under British colonial influence have fairly stable culturally embedded institutions, when compared to those that came under French influence. These site-specific traits shed light on the complexities linked to embedded institutions and their evolution. It further edifies theoretical perspectives on critical institutionalism. Studies on the source and content of the ‘last vestiges’ of these institutions are required.

本研究採用Lee 和Poon 所提出的隱藏常態變數模型來估計混合連續與間斷型變數之參數估計，並估計其馬式距離。此外，並利用穩健估計來估計混合型資料參數及其馬式距離，可在有離群值時解決最大蓋似估計的不穩定。 / Poon and Lee (1987) applied normal latent variable model to deal with the parameters estimation for the data mixed with continuous and discrete variables and Bedrick et al. (2000) used this idea to evaluate the Mahalanobis distance. In this thesis, we extend a similar idea to robustly estimate Multivariate Data Mixed with Continuous and Discrete Variables with the same model. Furthermore, we evaluate the Mahalanobis distance which can determine similarity of variables. The proposed method can overcome the unreliability of MLE while there exist outliers in the data.

Motivated by experiments on titanium alloys of Petrinic et al. (2006), which show the formation of cracks through the growth and coalescence of voids in ductile fracture, we consider the problem of formulating a variational model in nonlinear elasticity compatible both with cavitation and the appearance of discontinuities across two-dimensional surfaces. As in the model for cavitation of Müller and Spector (1995) we address this problem, which is connected to the sequential weak continuity of the determinant of the deformation gradient in spaces of functions having low regularity, by means of adding an appropriate surface energy term to the elastic energy. Based upon considerations of invertibility, we derive an expression for the surface energy that admits a physical and a geometrical interpretation, and that allows for the formulation of a model with better analytical properties. We obtain, in particular, important regularity results for the inverses of deformations, as well as the weak continuity of the determinants and the existence of minimizers. We show, further, that the creation of surface can be modeled by carefully analyzing the jump set of the inverses, and we point out some connections between the analysis of cavitation and fracture, the theory of SBV functions, and the theory of Cartesian currents of Giaquinta, Modica, and Soucek. In addition to the above, we extend previous work of Sivaloganathan, Spector and Tilakraj (2006) on the approximation of minimizers for the problem of cavitation with a constraint in the number of flaw points, and present some numerical results for this problem.

Contexte. L’adolescence est une période complexe marquée par une quête identitaire importante. Les comportements alimentaires adoptés à cette période tendent à perdurer dans le temps et influencent la croissance et la santé à long terme. De multiples facteurs d’ordre individuel, social, environnemental et sociétal peuvent influencer leurs choix alimentaires. Objectifs. Cette étude vise à explorer ce qui est le plus spontanément exprimé comme motivations alimentaires par les adolescents québécois âgés de 12 à 14 ans. Inspiré de l’étude des motivations alimentaires de Renner et coll. (2012), les déterminants des choix alimentaires de ces jeunes seront identifiés pour ensuite dégager les principaux facteurs d’influence agissant sur leurs choix alimentaires (Story et coll. 2002). Méthodes. Ce projet est une analyse de données qualitatives tirées de la recherche « Ados 12-14 : Les dimensions socioculturelles des pratiques alimentaires et d’activité physique des adolescents » (2014). Il repose sur l’analyse d’entrevues individuelles menées auprès de 50 filles et garçons âgés entre 12 et 14 ans. Sur un corpus d’entrevues retranscrites intégralement, une approche déductive fut utilisée avec l’échelle de mesure de motivations alimentaires de Renner et coll. (2012) comme grille d’analyse. Résultats. Parmi les nombreux facteurs d’influence, les facteurs individuels et ceux liés aux environnements sociaux se sont particulièrement démarqués. Au final, l’influence parentale, les préférences alimentaires, le désir de créer une impression sociale favorable et la santé sont déterminants, parfois même au détriment du plaisir. Conclusion. Cette étude révèle une variété de facteurs pouvant influencer les comportements alimentaires des adolescents québécois. Les constats dégagés de l’analyse des données effectuée sont pertinents pour les acteurs en nutrition en santé publique lesquels y trouveront des facteurs à prendre en compte dans la mise en place d’interventions de promotion des saines habitudes alimentaires auprès des adolescents québécois. / Context. Adolescence is a complex period marked by an important identity quest. Eating behaviours adopted during this period of life tend to influence growth and health in the long term. Individual, social, environmental and societal factors can influence teenagers’ eating habits. Objectives. This study is aimed at exploring what is most spontaneously expressed as feeding motivations by teenagers between 12 and 14 years old. Inspired by Renner et al.’s study (2012) about eating motivations, determinants of the adolescent food choices will be identified and then the main influence factors guiding their eating habits will be explored (Story et al., 2002). Methods. This project is an analysis of the qualitative data from the research project “Teenagers 12-14: Sociocultural dimensions of eating habits and physical activity of teenagers’’ (Ados 12-14: Les dimensions socioculturelles des pratiques alimentaires et d’activité physique des adolescents (2014)). It is based on the analysis of individual interviews with 50 girls and boys aged between 12 and 14 years old. Interview transcripts were analyzed using a deductive approach and the motivation measuring scale of Renner et al. (2012) as an analysis grid. Results. Amongst numerous factors, individual and socioenvironmental factors stood out. Globally, parents’ influence, eating preferences, the desire to make a good social impression, and health were decisive factors, sometimes even stronger than pleasure. Conclusion. Many factors impacting eating behaviors in the teenage population of Québec were identified in this study. The results generated by this data analysis will be useful for public health nutritionists for the design of intervention plans for the promotion of healthy eating habits within the Québec teenage population.

Dans ce travail, nous analysons et comparons les comportements de localisation de trois différentes formes d'investissements durables (les greenfields seuls, les investissements directs étrangers totaux (IDE) et les investissements productifs intérieurs). Parmi un groupe de déterminants, dont une large partie est inspirée de la littérature empirique existante, nous étudions tout particulièrement le rôle que joue le taux de change réel sur l'attractivité des territoires vis-à-vis de chacun de ces investissements. Un tel examen est opéré à partir d'une économétrie de données de panel. Il consiste à tester un modèle de référence sur les greenfields, les IDE totaux ainsi que sur les investissements productifs – avec des spécifications adaptées à chaque cas – et à interpréter l'ensemble des résultats et en particulier les écarts obtenus. Pour appréhender les variables dépendantes, nous nous appuyons sur les séries statistiques de la CNUCED et de l'OCDE, ainsi que sur celles du FDI Markets (Financial Times), et de l'European Investment Monitor (Ernst & Young) qui ont jusqu'à présent été peu exploitées, voire ignorées, par les études portant sur le sujet. Plusieurs échantillons (global, européen et OCDE) et plusieurs périodes (2004-2010, 1998-2010, 1995-2010) sont retenus en fonction de la disponibilité des données. Nos résultats indiquent que les greenfields, les IDE totaux (et donc, par déduction, les fusions-acquisitions) et les investissements productifs répondent à des critères de localisation bien différents correspondant à leurs caractéristiques propres. L'impact qu'exerce notamment le taux de change réel sur chacun de ces capitaux est des plus intéressants. De faibles prix relatifs apparaissent ainsi comme attirant de façon sensible les investissements productifs, qu'ils soient d'origine domestique ou étrangère (greenfields). En revanche, il semble que ces prix relatifs n'aient aucune influence sur les recompositions transnationales de capital, déduites à partir des IDE totaux. / In this work, we analyze and compare locational behavior of three forms of durable investments (greenfields, foreign direct investments, and internal productive investments). Among a group of determinants, mainly inspired by the existing empirical literature, we study more particularly the role played by the real exchange rate on countries' attractiveness towards each of those investments. Panel data econometrics is used for this study. We test a basis model on greenfields, total FDIs, and on productive investments – with specifications adapted to each case – and then analyze the results, particularly when they show differences. In order to proxy the dependent variables, we use the UNCTAD and the OECD's data series, as well as the FDI Markets (Financial Times) and the European Investment Monitor (Ernst & Young) which have somehow been ignored by the studies on the subject until now. Several samples (global, European and OECD) and several periods (2004-2010, 1998-2010 and 1995-2010) are analyzed depending on data availability. Our results prove that greenfields, total FDIs (and mergers and acquisitions by deduction), and productive investments respond to different location criteria according to their characteristics. In particular, the impact of the real exchange rate on each of those investments is very interesting. While low relative prices attract significantly productive investments, financed by domestic or foreign multinational companies (greenfields), they don't seem to have any impact on mergers and acquisitions (deduced from total FDIs).