Training deep convolutional architectures for vision

Les tâches de vision artificielle telles que la reconnaissance d’objets demeurent irrésolues à ce jour. Les algorithmes d’apprentissage tels que les Réseaux de Neurones Artificiels (RNA), représentent une approche prometteuse permettant d’apprendre des caractéristiques utiles pour ces tâches. Ce processus d’optimisation est néanmoins difficile. Les réseaux profonds à base de Machine de Boltzmann Restreintes (RBM) ont récemment été proposés afin de guider l’extraction de représentations intermédiaires, grâce à un algorithme d’apprentissage non-supervisé. Ce mémoire présente, par l’entremise de trois articles, des contributions à ce domaine de recherche. Le premier article traite de la RBM convolutionelle. L’usage de champs réceptifs locaux ainsi que le regroupement d’unités cachées en couches partageant les même paramètres, réduit considérablement le nombre de paramètres à apprendre et engendre des détecteurs de caractéristiques locaux et équivariant aux translations. Ceci mène à des modèles ayant une meilleure vraisemblance, comparativement aux RBMs entraînées sur des segments d’images. Le deuxième article est motivé par des découvertes récentes en neurosciences. Il analyse l’impact d’unités quadratiques sur des tâches de classification visuelles, ainsi que celui d’une nouvelle fonction d’activation. Nous observons que les RNAs à base d’unités quadratiques utilisant la fonction softsign, donnent de meilleures performances de généralisation. Le dernière article quand à lui, offre une vision critique des algorithmes populaires d’entraînement de RBMs. Nous montrons que l’algorithme de Divergence Contrastive (CD) et la CD Persistente ne sont pas robustes : tous deux nécessitent une surface d’énergie relativement plate afin que leur chaîne négative puisse mixer. La PCD à "poids rapides" contourne ce problème en perturbant légèrement le modèle, cependant, ceci génère des échantillons bruités. L’usage de chaînes tempérées dans la phase négative est une façon robuste d’adresser ces problèmes et mène à de meilleurs modèles génératifs.

Privacy in Bitcoin through decentralized mixers

Dans les crypto-monnaies telles Bitcoin, l’anonymité des utilisateurs peut être compromise de plusieurs façons. Dans ce mémoire, nous effectuons une revue de littérature et une classification des différents protocoles existants pour anonymiser les usagers et analysons leur efficacité. S’appuyant sur certains critères désirables dans de tels protocoles, nous proposons un modèle de mixeur synchrone décentralisé. Nous avons ciblé deux approches qui s’inscrivent dans ce modèle, le plan de transaction et le réseau de transactions, le second étant une contribution originale de ce mémoire. Nous expliquons son fonctionnement puis analysons son efficacité dans le contexte actuel d’utilisation de Bitcoin

Short Nylon-6 Fibre/Rubber-Toughened Polystyrene Composites

The thesis describes studies on development of short Nylon-6 fibre composites based on rubber-toughened polystyrene (PS). Toughening was done using natural rubber (NR), styrene-butadiene rubber (SBR) and whole tyre reclaim (WTR). The composites were prepared by melt mixing in an internal mixer at 170 oC. It was found that the optimum blend ratio was 85/15 for PS/NR, 90/10 for PS/SBR and 90/22 for PS/WTR blends. The effect of dynamic vulcanisation on 85/15 PS/NR and 90/10 PS/SBR blends using dicumyl peroxide (DCP) at various concentrations were also studied. The dynamic crosslinking improved the tensile properties, flexural properties, impact strength and dynamic mechanical properties of both the blends. The effect of unmodified and resorcinol formaldehyde latex (RFL)-coated short Nylon-6 fibres on the mechanical properties, morphology and dynamic mechanical properties of 85/15 PS/NR, 90/10 PS/SBR and 90/22 PS/WTR blends were studied. Fibre loading was varied from 0 to 3 wt.%. For 85/15 PS/NR blend, there was a significant enhancement in tensile properties, flexural properties and impact strength with 1 wt.% of both unmodified and RFL-coated fibres. Dynamic mechanical analysis revealed that the storage modulus at room temperature was maximum at 1 wt.% fiber loading for both composites. The surface functionality of the fiber was improved by giving alkali treatment. Maleic anhydride-grafted-polystyrene (MA-g-PS) was prepared and used as a compatibiliser. The effect of MA-g-PS on the composites was investigated with respect to mechanical properties, morphology and dynamic mechanical properties. The compatibiliser loading was varied from 0 to 2 wt.%. The properties were enhanced significantly in the case of treated and untreated fibre composites at a compatibiliser loading of 0.75 wt.%. SEM analysis confirmed better bonding between the fibre and the matrix. Dynamic mechanical studies showed that the storage modulus at room temperature improved for treated fibre composites in the presence of compatibiliser. In the case of 90/10 PS/SBR composites, the addition of short Nylon-6 fibres at 1 wt.% loading improved the tensile modulus, flexural properties and impact strength while the tensile strength was marginally reduced. The surface treated fibers along with compatibiliser at 0.5 wt.% improved the tensile properties, flexural properties and impact strength. DMA reveale that the storage modulus at room temperature was better for composites containing untreated fibre and the compatibiliser. In the case of 90/22 PS/WTR blends, 1 wt.% unmodified fibre and 0.5 wt.% RFL-coated fibres improved tensile modulus, flexural properties and impact strength. Tensile strength was improved marginally. The surface treatment of Nylon fibre and the addition of compatibiliser at 0.5 wt.% enhanced the tensile properties, flexural properties and impact strength. The dynamic mechanical analysis showed that the storage modulus at room temperature was better for untreated fibre composites in conjunction with the compatibiliser. The thermal stability of PS/NR was studied by TGA. Thermal stability of the blends improved with dynamic vulcanisation and with the incorporation of RFL-coated Nylon fibres. The untreated and partially hydrolyzed fibre composites in conjunction with the compatibiliser enhanced the thermal stability. Kinetic studies showed that the degradation of the blends and the composites followed first order kinetics.

Compositional Data Analysis with R

R from http://www.r-project.org/ is ‘GNU S’ – a language and environment for statistical computing and graphics. The environment in which many classical and modern statistical techniques have been implemented, but many are supplied as packages. There are 8 standard packages and many more are available through the cran family of Internet sites http://cran.r-project.org . We started to develop a library of functions in R to support the analysis of mixtures and our goal is a MixeR package for compositional data analysis that provides support for operations on compositions: perturbation and power multiplication, subcomposition with or without residuals, centering of the data, computing Aitchison’s, Euclidean, Bhattacharyya distances, compositional Kullback-Leibler divergence etc. graphical presentation of compositions in ternary diagrams and tetrahedrons with additional features: barycenter, geometric mean of the data set, the percentiles lines, marking and coloring of subsets of the data set, theirs geometric means, notation of individual data in the set . . . dealing with zeros and missing values in compositional data sets with R procedures for simple and multiplicative replacement strategy, the time series analysis of compositional data. We’ll present the current status of MixeR development and illustrate its use on selected data sets

Active micromachined integrated terahertz circuits

Schottky barrier diodes have been integrated into on-chip rectangular waveguides. Two novel techniques have been developed to fabricate diodes with posts suitable for integration into waveguides. One technique produces diodes with anode diameters of the order of microns with post heights from 90 to 125 microns and the second technique produces sub-micron anodes with post heights around 20 microns. A method has been developed to incorporate these structures into a rectangular waveguide and provide a top contact onto the anode which could be used as an I.F. output in a mixer circuit. Devices have been fabricated and D.C. characterized.

Mixing characterisation for a serpentine microchannel equipped with embedded barriers

This paper describes the design, simulation, fabrication and experimental analysis of a passive micromixer for the mixing of biological solvents. The mixer consists of a T-junction, followed by a serpentine microchannel. the serpentine has three arcs, each equipped with circular barriers that are patterned as two opposing triangles. >The barriers are engineered to induce periodic perturbations in the flow field and enhance the mixing. CFD (Computational Fluid Dynamics) method is applied to optimise the geometric variables of the mixer before fabrication. The mixer is made from PDMS (Polydimethylsiloxane) using photo- and soft-lithography techniques. Experimental measurements are performed using yellow and blue food dyes as the mixing fluids. The mixing is measured by analysing the composition of the flow's colour across the outlet channel. The performance of the mixer is examined in a wide range of flow rates from 0.5 to 10 µl/min. Mixing efficiencies of higher than 99.4% are obtained in the experiments confirming the results of numerical simulations. The proposed mixer can be employed as a part of lab-on-a-chip for biomedical applications.

Lab-on-a-chip : a component view

Miniaturization is being increasingly applied to biological and chemical analysis processes. Lab-on-a-chip systems are direct creation of the advancement in the miniaturization of these processes. They offer a host of exciting applications in several areas including clinical diagnostics, food and environmental analysis, and drug discovery and delivery studies. This paper reviews lab-on-a-chip systems from their components perspective. It provides a categorization of the standard functional components found in lab-on-a-chip devices together with an overview of the latest trends and developments related to lab-on-a-chip technologies and their application in nanobiotechnology. The functional components include: injector, transporter, preparator, mixer, reactor, separator, detector, controller, and power supply. The components are represented by appropriate symbols allowing designers to present their lab-on-a-chip products in a standard manner. Definition and role of each functional component are included and complemented with examples of existing work. Through the approach presented in this paper, it is hoped that modularity and technology transfer in lab-on-a-chip systems can be further facilitated and their application in nanobiotechnology be expanded.

Collaborative authorship in film production: Walter Murch and film editing

Filmmaking is frequently cited as the most collaborative of all arts, yet for the most part, mainstream and scholarly literature have received films as the creative voice of just one artist – the director. The reasons for this are many: general ignorance of how films are made; the hijacking of film theory by literary theory, and the continuing popularity of the myth of the Romantic Artist as solitary genius are some of them. The case for collaborative authorship has gained momentum since the 1980s as studies on the production of individual films, actors, production companies and the history of the film industry as a whole have proliferated and drawn attention to the disparities between how films are perceived and how they are actually made. This article analyses collaboration in film production culture through examination of the role of the film editor. Concentrating specifically on the film/sound editor and mixer Walter Murch, it examines his role as a collaborative author in his early work with director Francis Ford Coppola and his later work with English director Anthony Minghella.

Fluid mixing in droplet-based microfluidics with a serpentine microchannel

The hydrodynamics and mixing process within droplets travelling along a three dimensional serpentine microchannel are studied using a computational fluid dynamics simulation based on the volume-of-fluid approach. The fluid mixing within the droplet follows symmetric circulations in the straight section, which generates axial mixing. In the winding section, the asymmetric circulations lead to the reorientation of the fluids within the droplet, thus enhancing the mixing efficiency. The mixing performance is controlled by the spatial distribution of the mixing components and the circulation period within the droplet. The best mixing occurs when the droplet size is comparable with the channel width. When the droplet size is less than two times the channel width, the asymmetric circulations make it easy for the fluid to distribute in the axial direction, which leads to a fast mixing process. For larger droplets, the long circulation period becomes more significant, which causes lower mixing efficiency.

Low cost BIST techniques for linear and non-linear analog circuits

With the ever increasing demands for high complexity consumer electronic products, market pressures demand faster product development and lower cost. SoCbased design can provide the required design flexibility and speed by allowing the use of IP cores. However, testing costs in the SoC environment can reach a substantial percent of the total production cost. Analog testing costs may dominate the total test cost, as testing of analog circuits usually require functional verification of the circuit and special testing procedures. For RF analog circuits commonly used in wireless applications, testing is further complicated because of the high frequencies involved. In summary, reducing analog test cost is of major importance in the electronic industry today. BIST techniques for analog circuits, though potentially able to solve the analog test cost problem, have some limitations. Some techniques are circuit dependent, requiring reconfiguration of the circuit being tested, and are generally not usable in RF circuits. In the SoC environment, as processing and memory resources are available, they could be used in the test. However, the overhead for adding additional AD and DA converters may be too costly for most systems, and analog routing of signals may not be feasible and may introduce signal distortion. In this work a simple and low cost digitizer is used instead of an ADC in order to enable analog testing strategies to be implemented in a SoC environment. Thanks to the low analog area overhead of the converter, multiple analog test points can be observed and specific analog test strategies can be enabled. As the digitizer is always connected to the analog test point, it is not necessary to include muxes and switches that would degrade the signal path. For RF analog circuits, this is specially useful, as the circuit impedance is fixed and the influence of the digitizer can be accounted for in the design phase. Thanks to the simplicity of the converter, it is able to reach higher frequencies, and enables the implementation of low cost RF test strategies. The digitizer has been applied successfully in the testing of both low frequency and RF analog circuits. Also, as testing is based on frequency-domain characteristics, nonlinear characteristics like intermodulation products can also be evaluated. Specifically, practical results were obtained for prototyped base band filters and a 100MHz mixer. The application of the converter for noise figure evaluation was also addressed, and experimental results for low frequency amplifiers using conventional opamps were obtained. The proposed method is able to enhance the testability of current mixed-signal designs, being suitable for the SoC environment used in many industrial products nowadays.

Desenvolvimento de elastômeros termoplásticos a partir de SBR epoxidada e polipropileno

Os elastômeros termoplásticos vulcanizados (TPVs) na sua maioria constituídos por borrachas apolares (EPDM, NR) e poliolefinas (PP, PE), apresentam a vantagem de serem processáveis como termoplásticos e serem facilmente reciclados. No entanto, apresentam desvantagens no que se refere à sua baixa resistência a óleos, combustíveis e graxas em relação à borracha termofixa. Este trabalho, teve como objetivo estudar a obtenção de TPVs com propriedades mecânicas adequadas e resistência a óleos e solventes orgânicos, a partir da borracha comercial SBR 1502 parcialmente epoxidada. Esta, por ter a estrutura química de sua cadeia principal modificada pela introdução de grupos epóxidos, deve apresentar melhor resistência a óleos e solventes. Os TPVs foram obtidos em misturador fechado acoplado a um reômetro Haake, na temperatura de 1800C e velocidade de rotor de 75 rpm, vulcanizados dinamicamente com o sistema Bismaleimida/peróxido de dicumila. Foram caracterizados quanto às suas propriedades mecânicas por medidas tensão-deformação, medidas mecânicas dinâmicas, inchamento em ciclohexano, THF e óleo IRM 903, dureza. A morfologia foi determinada por microscopia eletrônica de varredura, MEV. Foram analisados os fatores que influenciam as propriedades dos TPVs, tais como composição (relação PP/SBR), teor de BMI, grau de epoxidação da borracha, uso de agente compatibilizante. O TPV na composição PP/SBR 40/60, esta epoxidada em 40 mol % e contendo o agente compatibilizante Vestenamer adicionado na forma de blenda (borracha/agente compatibilizante) apresentou a melhor resposta em termos de tensão-deformação na ruptura. Os TPVs com a SBR epoxidada em 70% apresentaram melhor resistência a óleo e solventes. Os fatores, potencialmente, capazes de influenciar a dureza dos TPVs também foram avaliados. Neste particular, verificou-se que o tipo de poliolefina, bem como o uso de plastificante são os fatores que mais influenciam a dureza dos TPVs.

Uma interface para o aumento da faixa de freqüências de operação de FPAAS

O crescente avanço nas mais diversas áreas da eletrônica, desde instrumentação em baixa freqüência até telecomunicações operando em freqüências muito elevadas, e a necessidade de soluções baratas em curto espaço de tempo que acompanhem a demanda de mercado, torna a procura por circuitos programáveis, tanto digitais como analógicos, um ponto comum em diversas pesquisas. Os dispositivos digitais programáveis, que têm como grande representante os Field Programmable Gate Arrays (FPGAs), vêm apresentando um elevado e contínuo crescimento em termos de complexidade, desempenho e número de transistores integrados, já há várias décadas. O desenvolvimento de dispositivos analógicos programáveis (Field Programmable Analog Arrays – FPAAs), entretanto, esbarra em dois pontos fundamentais que tornam sua evolução um tanto latente: a estreita largura de banda alcançada, conseqüência da necessidade de um grande número de chaves de programação e reconfiguração, e a elevada área consumida por componentes analógicos como resistores e capacitores, quando integrados em processos VLSI Este trabalho apresenta uma proposta para aumentar a faixa de freqüências das aplicações passíveis de serem utilizadas tanto em FPAAs comerciais quanto em outros FPAAs, através da utilização de uma interface de translação e seleção de sinais, mantendo características de programabilidade do FPAA em questão, sem aumentar em muito sua potência consumida. A proposta, a simulação e a implementação da interface são apresentadas ao longo desta dissertação. Resultados de simulação e resultados práticos obtidos comprovam a eficácia da proposta.

Desenvolvimento de bioblendas poliméricas a partir de gelatina bovina e de tilápia com amido de milho

The natural raw materials acquired special importance beside the mineral raw materials with the need for using alternative sources to oil, because they can be used to produce biopolymers. Gelatin, produced from the denaturation of collagen, and starch, an abundant polysaccharide in various plants, are examples of biopolymers which have several technological applications, especially in films. The objective of this work is to produce polymeric bioblends with gelatin and corn starch using two types of gelatin: commercial bovine gelatin and gelatin produced from mechanically separated flesh of tilapia (Oreochromis niloticus). For the extraction of tilapia gelatin 3 distinct pretreatments, followed by extraction in distilled water under heating were performed. The properties of gelatin extracted were similar to bovine gelatin, and the differences can be explained by the difference in extraction processes and sources. Blends of commercial gelatin and starch were produced in an internal mixer from a Haake torque rheometer, to study the behavior of the gelatin mixture with starch, thus, the same compositions were processed by twin screw extrusion, to define the mixing parameters. Subsequently, the extrusion of blends of tilapia gelatin and corn starch was carried out in the same twin screw extruder. The physico-chemical, rheological and morphological properties of the blends with thermoplastic starch and gelatin were studied. It was found that various properties vary linearly with increasing concentration of the components. The blends produced are immiscible, and among the two gelatins, tilapia gelatin showed a better interfacial adhesion with the corn starch. Regarding the morphology, gelatins formed the dispersed phase in all compositions studied, even in compositions rich in starch. Can be concluded that the procedure for tilapia gelatin extraction is feasible and advantageous, and the increasing in its scale to a reactor of 30 liters is possible, with a satisfactory yield. The bioblends of bovine gelatin/corn starch and tilapia gelatin/corn starch were successfully produced, and the processing conditions were appropriate

Compósitos de matriz metálica à base níquel com adição de TaC e NbC produzidos via metalurgia do pó

Carbide reinforced metallic alloys potentially improve some important mechanical properties required for the overall use of important engineering materials such as steel and nickel. Nevertheless, improved performance is achieved not only by composition enhancement but also by adequate processing techniques, such as novel sintering methods in the case of powder metallurgy. The method minimizes energy losses in addition to providing uniform heating during sintering. Thus, the general objective of this study was to evaluate the density, hardness, flexural strength, dilatometric behavior and to analyze the microstructure of metal matrix composites based nickel with addition of carbides of tantalum and / or niobium when sintered in a conventional furnace and Plasma assisted debinding and sintering (PADS). Initially, were defineds best parameters of granulation, screening and mixing procedure. After, mixtures of carbonyl Ni and 5%, 10% and 15 wt.% NbC and TaC were prepared in a Y-type mixer under wet conditions during 60 minutes. The mixtures were then dried and granulated using 1.5 wt. % paraffin diluted in hexane. Granulates were cold pressed under 600 MPa. Paraffin was then removed from the pressed pellets during a pre-sintering process carried out in a tubular furnace at 500 °C during 30 min. The heating rate was 3 ºC/min. The pellets were then sintered using either a plasma assisted reactor or a conventional resistive tubular furnace. For both methods, the heating rate was set to 8 ºC/min up to 1150 °C. The holding time was 60 minutes. The microstructure of the sintered samples was evaluated by SEM. Brinell hardness tests were also carried out. The results revealed that higher density and higher hardness values were observed in the plasma-assisted sintered samples. Hardness increased with the concentration of carbides in the Ni-matrix. The flexural strength also increased by adding the carbides. The decline was larger for the sample with addition of 5% 5% TaC and NbC. In general, compositions containing added carbide 10% showed less porous and more uniform distribution of carbides in the nickel matrix microstructural appearance. Thus, both added carbide and plasma sintering improved density, hardness, flexural strength and microstructural appearance of the composites

Um olhar estético sobre o telespetáculo esportivo :contribuições para o ensino do esporte na escola

The sportive television spectacle shows up like a media phenomenon which evidence performances and techniques that further on the community realized in the quotidian sportive practices into a show and play mixer, that more and more, sensibility the viewers. In this scenery , this research is about the relation between sportive television spectacle and viewer by the aesthetics perspective, and discussed the sport in the Physical Education context with the focus in the contributions to teaching in the school. The research, to the methodological view, uses the discuss analysis and realizes a qualitative study in two interdependent moments. In the first, analysis the dada collated in the Education, sport and television workshop. In the second, turns about TV images of sportive television spectacle to appreciated, to support a discuss of these phenomenon s aesthetics elements, looking for the relation established between the sportive television spectacle and viewer. The analysis s emphasis is in the sensations by sportive television spectacle, as well in the aesthetic elements problems promoted by this experience, from the two basics concepts, to know: the aesthetic and the synaesthetic perception. The analysis s corpus is composed by Esporte Espetacular TV program. Identify, in the sportive television spectacle appreciation, four big discuss axis of great importance between sportive television spectacle and viewer understanding: the relation space and time, the belonging feeling, the language interlacing, and beauty models. Understand in the analysis that, in the sportive TV scenery, occurs a construction image that seduces and involves the viewer through the aesthetics elements. From that discuss axis reflect about teaching sport in the school, main concern to the materialization a sport model only. Aims unfolding about space and time construction yet, and about the multiplicity language for to work in the school. Positions that no absolute rules, but that check the TV sport by the seduction promoted in the viewers through the aesthetics elements transmit