The present study considers the numerical simulation of the different flow characteristics involved in the conjugate heat transfer analysis of an internally cooled gas turbine blade. Conjugate simulations require full coupling of convective heat transfer in fluid regions to the heat diffusion in solid regions. Therefore, accurate prediction of heat transfer quantities on both external and internal surfaces has the uppermost importance and highly connected with the performance of the employed turbulence models. The complex flow on both surfaces of the internally cooled turbine blades is caused from the boundary layer laminar-to-turbulence transition, shock wave interaction with boundary layer, high streamline curvature and sequential flow separation. In order to discover the performances of different turbulence models on these flow types, analyses have been conducted on five different experimental studies each concerned with different flow and heat transfer characteristics. Each experimental study has been examined with four different turbulence models available in the commercial software (ANSYS FLUENT13.0) to decide most suitable RANS-based turbulence model. The Realizable k-&epsilon / model, Shear Stress Transport k-&omega / model, Reynolds Stress Model and V2-f model, which became increasingly popular during the last few years, have been used at the numerical simulations. According to conducted analyses, despite a few unreasonable predictions, in the majority of the numerical simulations, V2-f model outperforms other first-order turbulence models (Realizable k-&epsilon / and Shear Stress Transport k-&omega / ) in terms of accuracy and Reynolds Stress Model in terms of convergence.

Notre projet porte sur l'étude fonctionnelle du facteur de transcription GATA2 qui contrôle la mise en place de certaines populations neuronales. Mon travail de thèse est basé sur l'étude de sa fonction dans la moelle épinière.<br />Dans la moelle épinière ventrale embryonnaire, les motoneurones et 4 classes d'interneurones V0, V1, V2 et V3 sont générés dans des territoires distincts. Les V2 sont spécifiés dans un territoire adjacent à celui des motoneurones et sont subdivisés en V2a et V2b. Notre but est de déterminer le rôle de Gata2 durant la spécification des V2 et son l'influence sur la prolifération des progéniteurs neuraux.<br />Par une approche de perte et gain de fonction, nous avons montré que :<br />1. Gata2 inhibe la prolifération des progéniteurs neuraux avec un effet cellulaire-non autonome, en induisant un inhibiteur du cycle et en réprimant la voie Notch. Ce contrôle peut être découplé de la différenciation neuronale et ne nécessite pas d'activité proneurale.<br />2. Gata2 joue favorise l'émergence des V2 et réprime la différenciation des motoneurones, par la modulation des voies Shh et TGF-Β. Gata2 participe à la dichotomie des V2 car il favorise la différenciation des V2b et inhibe celle des V2a. Nous avons montré que les V2a sont glutamatergiques et les V2b sont GABAergiques.<br />Cette étude a mis en évidence la complexité des mécanismes moléculaires à l'origine de la diversité neuronale.

No description available.

This diploma work deals with the program Cabri II Plus and its opportunities as for the application into the lessons of Physics at elementary and high schools, grammar schools and also into particular seminars at colleges with a study program focused on the applied science (statitcs). This work is aimed mainly for tutors as a mean of both efficiency and students engagement enhancement. For these particular purposes the set of sample assignments has been created. The contribution of the work will be examined right within the lessons of Physics at L. Kuba Elementary School in České Budějovice and in particular seminars at the Department of Applied Physics and Technics at the Faculty of Education at the University of South Bohemia.

Submitted by Automa??o e Estat?stica (sst@bczm.ufrn.br) on 2016-03-31T22:16:14Z No. of bitstreams: 1 RafaelAguiarMoura_DISSERT.pdf: 538475 bytes, checksum: 9fede85e6e7a0265d5f33c6b1ff1c7e1 (MD5) / Approved for entry into archive by Arlan Eloi Leite Silva (eloihistoriador@yahoo.com.br) on 2016-04-01T22:45:34Z (GMT) No. of bitstreams: 1 RafaelAguiarMoura_DISSERT.pdf: 538475 bytes, checksum: 9fede85e6e7a0265d5f33c6b1ff1c7e1 (MD5) / Made available in DSpace on 2016-04-01T22:45:34Z (GMT). No. of bitstreams: 1 RafaelAguiarMoura_DISSERT.pdf: 538475 bytes, checksum: 9fede85e6e7a0265d5f33c6b1ff1c7e1 (MD5) Previous issue date: 2015-03-30 / Coordena??o de Aperfei?oamento de Pessoal de N?vel Superior - CAPES / Em conson?ncia com o modelo de competi??o de gram?ticas (KROCH, 1989; 2001), segundo o qual a mudan?a nos dom?nios sint?ticos constitui um processo que se desenvolve via competi??o entre diferentes gram?ticas, descrevemos e analisamos as constru??es V2/V3 superficiais em ora??es matrizes/ra?zes de cartas pessoais brasileiras dos s?culos 19 e 20. O corpus, composto por 154 por cartas pessoais do Rio de Janeiro e do Rio Grande do Norte, est? dividido em tr?s metades de s?culo: (i) segunda metade do s?culo 19; (ii) primeira metade do s?culo 20; e (iii) segunda metade do s?culo 20. O nosso foco foi a observa??o da natureza dos constituintes pr?-verbais em constru??es V2 (verbo em segunda posi??o na senten?a) e V3 (verbo em terceira posi??o na senten?a) superficiais, com uma aten??o especial sobre o posicionamento do sujeito. Embasados nos diversos estudos diacr?nicos acerca dos padr?es de ordena??o do Portugu?s (Ambar (1992); Ribeiro (1995, 2001); Paix?o de Sousa (2004); Paiva (2011); Coelho e Martins (2009, 2012)), nosso estudo procurou constatar quais s?o os padr?es emp?ricos da ordena??o que envolvem as constru??es V2/V3 superficiais e como esses padr?es se estruturam sintaticamente dentro de uma perspectiva te?rica formal (CHOMSKY, 1981; 1986), mais especificamente, em conformidade com os estudos de Antonelli (2011) e de Costa e Galves (2002). Os resultados da pesquisa mostram que os dados da segunda metade do s?culo 19 ? diferentemente dos dados da primeira e da segunda metade do s?culo 20 ? apresentam um maior equil?brio em rela??o ? natureza sint?tica do constituinte pr?-verbal (cont?guo ou n?o), de forma que, nesse per?odo, a ocorr?ncia de ordens com o sujeito em posi??o pr?-verbal chega a, no m?ximo, 52% (231/444 dados); enquanto que, nos 48% (213/444 dados) restantes, os constituintes pr?-verbais s?o representados por um constituinte n?o sujeito, quase sempre um adjunto adverbial. Diante dos resultados, advogamos que as cartas pessoais brasileiras do s?culo 19 apresentam padr?es de ordena??o associados a um sistema V2 e a um sistema SV, configurando, portanto, um poss?vel processo de competi??o entre diferentes gram?ticas que instanciam ou um sistema V2 ou um sistema SV. Ou seja, as cartas brasileiras do s?culo 19 instanciam uma competi??o entre a gram?tica do Portugu?s Cl?ssico (um sistema V2) e as gram?ticas do Portugu?s Brasileiro e do Portugu?s Europeu (um sistema SV). Logo, esse per?odo est? sujeito ? realiza??o de duas marca??es param?tricas distintas: (i) verbo movido at? o n?cleo Fin (gram?tica do Portugu?s Cl?ssico) e (ii) verbo movido at? o n?cleo T (gram?tica do Portugu?s Brasileiro/Portugu?s Europeu). Por outro lado, nas cartas pessoais do s?culo 20 (primeira e segunda metades), h? um not?rio aumento dos padr?es de ordena??o associados ao sistema SV, que se mostra mais est?vel. / In line with the model of grammar competition (Kroch, 1989; 2001), according to which the change in the syntactic domains is a process that develops via competition between different grammars, we describe and analyze the superficial constructions V2 / V3 in matrices / roots sentences of brazilian personal letters of the 19th and 20th centuries. The corpus, composed by 154 personal letters of Rio de Janeiro and Rio Grande do Norte, is divided into three century halves: (i) latter half of the 19th century; (ii) first half of the 20th century; and (iii) latter half of the 20th century. Our focus was the observation of the nature of preverbal constituents in superficial constructions V2 (verb in second position in the sentence) and V3 (verb in third position in the sentence), with a special attention on the position of the subject. Based on the various diachronical studies about the Portuguese ordination standards (Ambar (1992); Ribeiro (1995, 2001); Paix?o de Sousa (2004); Paiva (2011), Coelho and Martins (2009, 2012)), our study sought to realize what are empirical ordination standards that involve superficial constructions V2 / V3 and how these patterns structure syntactically within a formal theoretical perspective (Chomsky, 1981; 1986), more specifically, in accordance with studies of Antonelli (2011), and Costa & Galves (2002). The survey results show that the data from the second half of the 19th century ? unlike the first and second half of the 20th century data ? have a greater balance in relation to the syntactic nature of preverbal constituent (contiguous or not), so that, in this period, the occurrence of orders with the subject in a preverbal position arrives at, at most, 52% (231/444 data); while in the 48% (213/444 data) remaining, the preverbal constituents are represented by a non-subject constituent, almost always an adverbial adjunct. Seen the results, we advocate that the brazilian personal letters of the 19th century have ordination patterns associated with a V2 system and an SV system, configuring, therefore, a possible competition process between different grammars that instantiate or a V2 system or an SV system. In other words, the brazilian letters of the 19th century instantiate a competition between the grammar of Classic Portuguese (a V2 system) and the grammars of Brazilian Portuguese and European Portuguese (an SV system). Therefore, that period is subject to the completion of two distinct parametric markings: (i) verb moved to the Fin core (grammar of Classic Portuguese) and (ii) verb moved to the T core (grammar of Brazilian Portuguese /European Portuguese). On the other hand, in the personal letters of the 20th century (first and second halves), there is a clear increase in ordenation patterns associated with the SV system, which shows more stable.

Modern state-of-the-art object detection models are based on convolutional neural networks and can be divided into single-shot detectors and two-stage detectors. Two-stage detectors exhibit impressive detection performance but their complex pipelines make them slow. Single-shot detectors are not as accurate as two-stage detectors, but are faster and can be used for real-time object detection. Despite the fact that single-shot detectors are faster, a large number of calculations are still required to produce a prediction that not many embedded devices are capable of doing in a reasonable time. Therefore, it is natural to ask if single-shot detectors could become faster even. Pruning is a technique to reduce the size of neural networks. The main idea behind network pruning is that some model parameters are redundant and do not contribute to the final output. By removing those redundant parameters, fewer computations are needed to produce predictions, which may lead to a faster inference and since the parameters are redundant, the model accuracy should not be affected. This thesis investigates two approaches for pruning the SSD-MobileNet- V2 single-shot detector. The first approach prunes the single-shot detector by a large portion and retrains the remaining parameters only once. In the other approach, a smaller portion is pruned, but pruning and retraining are done in an iterative fashion, where pruning and retraining constitute one iteration. Beyond comparing two pruning approaches, the thesis also studies the tradeoff between model accuracy and inference speed that pruning induces. The results from the experiments suggest that the iterative pruning approach preserves the accuracy of the original model better than the other approach where pruning and finetuning are performed once. For all four pruning levels that the two approaches are compared iterative pruning yields more accurate results. In addition, an inference evaluation indicates that iterative pruning is a good compression method for SSD-MobileNet-V2, finding models that both are faster and more accurate than the original model. The thesis findings could be used to guide future pruning research on SSD-MobileNet- V2, but also on other single-shot detectors such as RetinaNet and the YOLO models. / Moderna modeller för objektsdetektering bygger på konvolutionella neurala nätverk och kan delas in i ensteg- och tvåstegsdetektorer. Tvåstegsdetektorer uppvisar imponerande detektionsprestanda, men deras komplexa pipelines gör dem långsamma. Enstegsdetektorer uppvisar oftast inte lika bra detektionsprestanda som tvåstegsdetektorer, men de är snabbare och kan användas för objektdetektering i realtid. Trots att enstegsdetektorer är snabbare krävs det fortfarande ett stort antal beräkningar för att få fram en prediktering, vilket inte många inbyggda enheter kan göra på rimlig tid. Därför är det naturligt att fråga sig om enstegsdetektorer kan bli ännu snabbare. Nätverksbeskärning är en teknik för att minska storleken på neurala nätverk. Huvudtanken bakom nätverksbeskärning är att vissa modellparametrar är överflödiga och inte bidrar till det slutliga resultatet. Genom att ta bort dessa överflödiga parametrar krävs färre beräkningar för att producera en prediktering, vilket kan leda till att nätverket blir snabbare och eftersom parametrarna är överflödiga bör modellens detektionsprestanda inte påverkas. I den här masteruppsatsen undersöks två metoder för att beskära enstegsdetektorn SSD-MobileNet-V2. Det första tillvägagångssättet går ut på att en stor del av detektorn vikter beskärs och att de återstående parametrarna endast finjusteras en gång. I det andra tillvägagångssättet beskärs en mindre del, men beskärning och finjustering sker på ett iterativt sätt, där beskärning och finjustering utgör en iteration. Förutom att jämföra två metoder för beskärning studeras i masteruppsatsen också den kompromiss mellan modellens detektionsprestanda och inferenshastighet som beskärningen medför. Resultaten från experimenten tyder på att den iterativa beskärningsmetoden bevarar den ursprungliga modellens detektionsprestanda bättre än den andra metoden där beskärning och finjustering utförs en gång. För alla fyra beskärningsnivåer som de två metoderna jämförs ger iterativ beskärning mer exakta resultat. Dessutom visar en hastighetsutvärdering att iterativ beskärning är en bra komprimeringsmetod för SSD-MobileNet-V2, eftersom modeller som både snabbare och mer exakta än den ursprungliga modellen går att hitta. Masteruppsatsens resultat kan användas för att vägleda framtida forskning om beskärning av SSD-MobileNet-V2, men även av andra enstegsdetektorer, t.ex. RetinaNet och YOLO-modellerna.

Le récepteur V2 (V2R) de la vasopressine est un récepteur couplé aux protéines G (RCPG), jouant un rôle fondamental dans le maintien de l’homéostasie hydrosodique. À l’instar de nombreux RCPGs, il est capable d’interagir avec plusieurs types de protéines G hétérotrimériques et possède des voies de signalisation peu explorées aux mécanismes mal compris. Ces voies non canoniques font l’objet des travaux exposés dans ce mémoire. Il s’agit d’explorer les caractéristiques et mécanismes de la signalisation de V2R via G12, et de la voie d’activation d’ERK 1/2 par transactivation du récepteur de l’insulin-like growth factor 1, IGF1R. Par des études de transfert d’énergie de résonance de bioluminescence (BRET), nous exposons la capacité de V2R à interagir avec la sous-unité Gα12 ainsi que la modulation de la conformation de l’hétérotrimère G12 par l’agoniste de V2R, l’arginine-vasopressine. Ces travaux dévoilent également la modulation de l’interaction entre Gα12 et son effecteur classique RhoA, suggérant un engagement de RhoA, ainsi que la potentialisation via Gα12 de la production d’AMP cyclique. À l’aide de diverses méthodes d’inhibition sélective, nos résultats précisent les mécanismes de la transactivation. Ils supportent notamment le rôle initiateur de l’activation de Src par V2R et l’absence d’implication des ligands connus d’IGF1R dans la transactivation. La métalloprotéase MMP 3 apparaît par ailleurs comme un bon candidat pour réguler la transactivation. Ce projet met en lumière des modes de signalisation peu explorés de V2R, dont l’implication physiologique et physiopathologique pourrait s’avérer significative, au-delà d’un apport fondamental dans la compréhension de la signalisation des RCPGs. / Vasopressin V2 receptor is a G protein coupled receptor (GPCR) responsible for the homeostatic regulation of water and sodium recapture from the urine to the bloodstream. Akin to numerous GPCRs, this receptor can interact with more than one heterotrimeric G protein subtype, and is still associated with some poorly explored signaling pathways with indefinite mechanisms. These non-canonical pathways are the focus of this project. This work aims at unveiling the characteristics and mechanisms underlying G12 mediated signaling by V2R and ERK 1/2 activation through the transactivation of the tyrosine kinase Insulin-like growth factor 1 receptor (IGF1R). Using bioluminescence resonance energy transfer (BRET) experiments, we reveal V2R’s ability to interact with the Gα12 subunit, as well as the modulation of G12 heterotrimer’s conformation in response to V2R agonist arginine vasopressin (AVP). AVP-induced modulation of Gα12’s interaction with its classical effector RhoA upon stimulation with AVP suggests the engagement of RhoA, and our data also reveals that Gα12 potentiates AVP-induced cAMP production. Using diverse selective inhibition strategies, our results further define the mechanism of transactivation. Our data support a starter position of AVP-induced Src activation and discard IGF1R known agonists as the potential autocrine/paracrine factor responsible for IGF1R activation. Furthermore, our results suggest that the metalloproteinase MMP 3 is a good candidate for IGF1R transactivation. This project sheds light on lesser known signaling pathways involving V2R, which could reveal important on a physiological and pathophysiological scale, besides bringing a better understanding of the principles of GPCR signaling.

This master thesis project is carried out by one student at the Royal Institute of Technology in collaboration with Cybercom Group. The aim was to evaluate and compare system design strategies for fruit recognition in harvesting robots and the performance of supervised machine learning classification methods when applied to this specific task. The thesis covers the basics of these systems; to which parameters, constraints, requirements, and design decisions have been investigated. The framework is used as a foundation for the implementation of both sensing system, and processing and classification algorithms. A plastic tomato plant with fruit of varying maturity was used as a basis for training and testing, and a Kinect v2 for Windows including sensors for high resolution color-, depth, and IR data was used for image acquisition. The obtained data were processed and features of objects of interest extracted using MATLAB and a SDK for Kinect provided by Microsoft. Multiple views of the plant were acquired by having the plant rotate on a platform controlled by a stepper motor and an Ardunio Uno. The algorithms tested were binary classifiers, including Support Vector Machine, Decision Tree, and k-Nearest Neighbor. The models were trained and validated using a five fold cross validation in MATLABs Classification Learner application. Peformance metrics such as precision, recall, and the F1-score, used for accuracy comparison, were calculated. The statistical models k-NN and SVM achieved the best scores. The method considered most promising for fruit recognition purposes was the SVM. / Det här masterexamensarbetet har utförts av en student från Kungliga Tekniska Högskolan i samarbete med Cybercom Group. Målet var att utvärdera och jämföra designstrategier för igenkänning av frukt i en skörderobot och prestandan av klassificerande maskininlärningsalgoritmer när de appliceras på det specifika problemet. Arbetet omfattar grunderna av dessa system; till vilket parametrar, begränsningar, krav och designbeslut har undersökts. Ramverket användes sedan som grund för implementationen av sensorsystemet, processerings- och klassifikationsalgoritmerna. En tomatplanta i pplast med frukter av varierande mognasgrad användes som bas för träning och validering av systemet, och en Kinect för Windows v2 utrustad med sensorer för högupplöst färg, djup, och infraröd data anvöndes för att erhålla bilder. Datan processerades i MATLAB med hjälp av mjukvaruutvecklingskit för Kinect tillhandahållandet av Windows, i syfte att extrahera egenskaper ifrån objekt på bilderna. Multipla vyer erhölls genom att låta tomatplantan rotera på en plattform, driven av en stegmotor Arduino Uno. De binära klassifikationsalgoritmer som testades var Support Vector MAchine, Decision Tree och k-Nearest Neighbor. Modellerna tränades och valideras med hjälp av en five fold cross validation i MATLABs Classification Learner applikation. Prestationsindikatorer som precision, återkallelse och F1- poäng beräknades för de olika modellerna. Resultatet visade bland annat att statiska modeller som k-NN och SVM presterade bättre för det givna problemet, och att den sistnömnda är mest lovande för framtida applikationer.

Dans le cortex visuel primaire du chat (aires 17 et 18), les neurones répondant aux orientations présentes dans l’environnement (comme le contour des objets) sont organisés en colonnes perpendiculaires à la surface du cortex. Il a précédemment été montré qu'un changement drastique des orientations présentes dans l’environnement change la réponse des neurones. Par exemple, un neurone répondant à des orientations horizontales pourra répondre, après apprentissage d'un nouvel environnement, à des orientations obliques. Nous avons voulu, dans cette thèse, suivre les changements de propriétés de populations entières de neurones suite à ce type d'apprentissage. A cet effet, nous avons utilisé la technique d'imagerie optique des signaux intrinsèques, qui permet de mesurer l'activité d'une surface de cortex en utilisant le signal BOLD (blood-oxygen-level dependent). Cette thèse s'articule sur trois axes : l'effet de l'apprentissage au niveau local, l'effet de l’apprentissage à l'échelle de l'aire cérébrale, et la modélisation de l’apprentissage. Dans la première partie, nous avons comparé les changements d’orientations des neurones en fonction du gradient d’orientation local. Ce gradient est fort quand deux neurones voisins ont des orientations très différentes, et faible quand leurs orientations sont semblables. Les résultats montrent que plus les neurones sont entourés de neurones aux orientations différentes, plus l'apprentissage change leur réponse à l’orientation. Ceci suggère que les connexions locales ont une influence déterminante sur l'ampleur de l’apprentissage. Dans la deuxième partie, nous avons comparé le changement d’orientation des neurones des aires 17 et 18 avant et après apprentissage. Les résultats ne sont pas notablement différents entre les aires 17 et 18. On peut toutefois noter que les changements d’orientations dans l’aire 18 ont des amplitudes plus variables que dans l’aire 17. Ceci peut provenir du fait que l’aire 18 reçoit des afférences plus variées que l’aire 17, notamment une afférence directe des cellules Y du CGLd (Corps Genouillé Latéral dorsal). Dans la troisième partie, nous avons modélisé l'apprentissage expérimentalement observé à l'aide de réseaux de neurones utilisant un apprentissage Hebbien (cartes auto-organisatrices). Nous avons montré que le « feedback » des aires supérieures vers le cortex visuel primaire était souhaitable pour la conservation de la sélectivité à l'orientation des neurones. De manière générale, cette thèse montre l'importance des connexions locales dans la plasticité neuronale. Notamment, elles garantissent un apprentissage homéostatique, c'est-à- dire conservant la représentativité des orientations au niveau du cortex. De manière complémentaire, elle montre également l’importance des aires supérieures dans le maintient à long terme des orientations apprises par les neurones lors de l'apprentissage. / In the cat primary visual cortex (areas 17 and 18), neurons responding to orientations in the environment (such as the outline of objects) are organized in columns perpendicular to the cortical surface. It was previously shown that a drastic change in orientations in the environment changes the response of neurons. For example, a neuron responding to a horizontal orientation will respond, after learning a new environment, to an oblique orientation. In this thesis, we seek to follow the changes of properties of large populations of neurons due to this type of learning. To this end, we used the intrinsic signals optical imaging technique, which measures the activity of a cortical surface using the BOLD (blood-oxygen-level dependent) signal. This thesis follows three axes: the effect of learning at the local level, the effect of learning at the visual area scale, and the modeling of learning. In the first part, we compared the changes in orientation of neurons according to the local gradient of orientation. This gradient is strong when two neighboring neurons have very different orientations, and weak when their orientations are similar. The obtained relation between the gradient and the magnitude of change in orientation shows that when neurons are increasingly surrounded by neurons with different orientations, they change their response to orientation to a greater extent. This suggests that local connections have a decisive influence on the extent of learning. In the second part, we followed the change in the orientation of neurons in the areas 17 and 18, before and after learning. The results are not significantly different between area 17 and area 18. However, it is noteworthy that orientation changes in area 18 are more variable in amplitude than in area 17. This may be because area 18 receives more diverse inputs than area 17, including a direct input from dLGN (dorsal Lateral Geniculate Nucleus) Y cells. In the third part, we modeled the experimentally observed learning with neural networks using a Hebbian learning rule (networks are self-organizing maps). We have shown that feedback from higher areas to the primary visual cortex was desirable for the neurons orientation selectivity conservation. Overall, this thesis shows the importance of local connections in neuronal plasticity. In particular, they guarantee a homeostatic learning, i.e. maintaining the representativeness of orientations in the cortex. In a complementary manner, it also shows the importance of the superior areas in the conservation of learned orientations.

Thèse numérisée par la Direction des bibliothèques de l'Université de Montréal.