The microstructure of the brain at the cellular level provides crucial information for the understanding of the function of the brain. A large volume of high-resolution brain image data from 3D microscopy is an essential resource to study detailed microstructures of the brain. Accordingly, we have worked on obtaining high-resolution image data of entire mouse brains using the Knife-Edge Scanning Microscope (KESM). Furthermore, to disseminate these high-resolution whole mouse brain data sets to the neuroscience research community, we developed a web-based brain atlas, the KESM Brain Atlas (KESMBA). To visualize the data sets in 3D while using only a standard web browser, we employed distance attenuation and Google Maps API. The KESMBA is a powerful tool to analyze and share the KESM mouse brain data sets, but the image loading was slow because of the number of raster image (PNG) tiles and the file size. Moreover, since Google Maps API is governed by a commercial license, it does not provide enough flexibility for customization, extension, and mirroring. To solve these issues, we designed and developed a new KESM mouse brain atlas that uses a vector graphics format called Scalable Vector Graphics (SVG) instead of PNG, and OpenLayers API instead of Google Maps API. The SVG-based KESMBA using OpenLayers allows faster navigation and exploration of the KESM data, and more overlay of layers with the 4 times reduced file size compared to PNG tiles. Due to the reduced file size, the SVG-based KESMBA using OpenLayers is 2.45 times faster than the original atlas. By enhancing the performance, the users can more easily access the KESM data. We expect the SVG-based KESMBA to accelerate new discoveries in neuroscience.

No description available.

A limited set of Processing Element (PE) pairs in a parallel computer cover the internal communications of scalable parallel programs. We take advantage of this property using the concept of Sparse Flat Neighborhood Networks (Sparse FNNs). Sparse FNNs are network designs that provide single-switch latency and full wire bandwidth for each specified PE pair, despite using relatively few network interfaces per PE and switches that have far fewer ports than there are PEs. This dissertation discusses the design problem, runtime support, and working prototype (KASY0) for Sparse FNNs. KASY0 not only demonstrated the claimed properties, but also set world records for its price/performance and performance on a specific application. Parallel supercomputers execute many portions of an application simultaneously. For scalable programs, the more PEs the system has, the greater the potential speedup. Portions executing on different PEs may be able to work independently for short periods, but the performance desired might not be achieved due to delays in communication between PEs. The set of PE pairs that will communicate often is both predictable and small relative to the number of possible PE pairings. This sparseness property can be exploited in the design and implementation of networks for massively parallel supercomputers. The sparseness of communicating pairs is rooted in the fact that each of the human-designed communication patterns commonly used in parallel programs has the property that the number of communicating pairs grows relatively slowly as the number of PEs is increased. Additionally, the number of pairs in the union of all communication patterns used in a suite of parallel programs grows surprisingly slowly due to pair synergy: the same pair often appears in multiple communication patterns. Detailed analysis of communication patterns clearly shows that the number of PE pairs actually communicating is very sparse, although the structure of the sparseness can be complex.

We propose a new ensemble algorithm: the meta-boosting algorithm. This algorithm enables the original Adaboost algorithm to improve the decisions made by different WeakLearners utilizing the meta-learning approach. Better accuracy results are achieved since this algorithm reduces both bias and variance. However, higher accuracy also brings higher computational complexity, especially on big data. We then propose the parallelized meta-boosting algorithm: Parallelized-Meta-Learning (PML) using the MapReduce programming paradigm on Hadoop. The experimental results on the Amazon EC2 cloud computing infrastructure show that PML reduces the computation complexity enormously while retaining lower error rates than the results on a single computer. As we know MapReduce has its inherent weakness that it cannot directly support iterations in an algorithm, our approach is a win-win method, since it not only overcomes this weakness, but also secures good accuracy performance. The comparison between this approach and a contemporary algorithm AdaBoost.PL is also performed.

Current grid computing architectures have been based on cluster management and batch queuing systems, extended to a distributed, federated domain. These have shown shortcomings in terms of scalability, stability, and modularity. To address these problems, this dissertation applies architectural styles from the Internet and Web to the domain of generic computational grids. Using the REST style, a flexible model for grid resource interaction is developed which removes the need for any centralised services or specific protocols, thereby allowing a range of implementations and layering of further functionality. The context for resource interaction is a generalisation and formalisation of the Condor ClassAd match-making mechanism. This set theoretic model is described in depth, including the advantages and features which it realises. This RESTful style is also motivated by operational experience with existing grid infrastructures, and the design, operation, and performance of a proto-RESTful grid middleware package named DIRAC. This package was designed to provide for the LHCb particle physics experiment’s “off-line” computational infrastructure, and was first exercised during a 6 month data challenge which utilised over 670 years of CPU time and produced 98 TB of data through 300,000 tasks executed at computing centres around the world. The design of DIRAC and performance measures from the data challenge are reported. The main contribution of this work is the development of a REST model for grid resource interaction. In particular, it allows resource templating for scheduling queues which provide a novel distributed and scalable approach to resource scheduling on the grid.

Fourth Generation (4G) cellular technology Third Generation Partnership Project (3GPP) Long Term Evolution (LTE) offers high data rate capabilities to mobile users; and, operators are trying to deliver a true mobile broadband experience over LTE networks. Mobile TV and Video on Demand (VoD) are expected to be the main revenue generators in the near future [36] and efficient video streaming over wireless is the key to enabling this. 3GPP recommends the use of H.264 baseline profiles for all video based services in Third Generation (3G) Universal Mobile Telecommunication System (UMTS) networks. However, LTE networks need to support mobile devices with different display resolution requirements like small resolution mobile phones and high resolution laptops. Scalable Video Coding (SVC) is required to achieve this goal. Feasibility study of SVC for LTE is one of the main agenda of 3GPP Release10. SVC enhances H.264 with a set of new profiles and encoding tools that may be used to produce scalable bit streams. Efficient adaptation methods for SVC video transmission over LTE networks are proposed in this thesis. Advantages of SVC over H.264 are analyzed using real time use cases of mobile video streaming. Further, we study the cross layer adaptation and scheduling schemes for delivering SVC video streams most efficiently to the users in LTE networks in unicast and multicast transmissions. We propose SVC based video streaming scheme for unicast and multicast transmissions in the downlink direction, with dynamic adaptations and a scheduling scheme based on channel quality information from users. Simulation results indicate improved video quality for more number of users in the coverage area and efficient spectrum usage with the proposed methods.

Resumo: Os movimentos de massa na Serra do Mar são alvos de muito interesse e estudo devido à alta incidência e aos conseqüentes impactos ambientais e econômicos. Portanto, reunir várias informações sobre movimentos de massa, e disponibilizá-los como um produto de fácil aquisição e de navegabilidade para o conhecimento de alunos de graduação, é o desafio desse trabalho. Como área de aplicação, escolheu-se o município de Cubatão, pela significativa incidência de escorregamentos e pela importância da região no contexto nacional, frente a aspectos econômicos e ambientais. Desta forma, o objetivo principal é produzir um Atlas Multimídia sobre os processos de movimentos de massa, como contribuição ao ensino de graduação, nas áreas de Ciências da Terra e Engenharia e avaliar a aplicabilidade da linguagem SVG (Scalable Vectorial Graphics) na produção de mapas interativos. Foi efetuada uma consulta a docentes e alunos da área de Ciências da Terra, sobre o interesse na utilização de material didático multimídia como apoio ao ensino e aprendizado e os resultados obtidos mostraram o claro interesse deste público no desenvolvimento do material .O Atlas interativo elaborado, voltado especificamente a escorregamentos, usando interfaces gráficas na linguagem SVG,mostrou que é possível a publicação de aplicações cartográficas interativas utilizando o SVG. Acredita-se que a utilização deste tipo de material como apoio ao ensino e aprendizado seja uma tendência crescente, impulsionada pelos avanços tecnológicos e pela forte inserção da informática em todos os setores de atividade. Inseridos neste contexto, os atuais alunos, desde muito cedo, vivem a informática nos seus cotidianos e a inserção destes recursos em sala de aula torna os cursos mais atrativos para este público. / Abstract: Mass movements of Serra do Mar are the subject of much interest and study due to high incidence, the consequent environmental and economic impacts. For these reasons, to get some information about mass movements and make them available as a product of easy acquisition and navigability for the knowledge of undergraduate students, it is the challenge of this work. The study area was the Cubatão city by the significant incidence of slip and the importance of the region in the national context, in face of economic and environmental aspects. Thus, the main goal is to produce an Atlas Multimedia about processes of mass movements, to contribute to the education of undergraduate students in Earth Sciences and Engineering areas and evaluate the applicability of the SVG (Scalable Vectorial Graphics) language in the production of interactive maps. It was done a questionnaire for the purpose of gathering information from teachers and students in Earth Sciences area, about the interest in the use of multimedia educational materials to support teaching and learning. The results showed clear interest of the public in developing the material. The Interactive Atlas prepared, specifically to landslide, using graphical interfaces in the SVG language, showed that it is possible to publish interactive mapping applications using SVG. It was believed that the use of such material to support teaching and learning is a growing trend, driven by technological advances and the strong integration of computer science technology in all sectors of this activity. In this context, the current students, from an early age, living in their everyday computing and the integration of these resources in the classroom make the courses more appealing to the public. / Orientador: Paulina Setti Riedel / Coorientador: Cristhiane da Silva Ramos / Banca: Teresa Galloti Florenzano / Banca: Diana Sarita Hamburger / Banca: Rosângela Doin de Almeida / Banca: Paulo Milton Barbosa Landim / Possui anexo 1 CD_ROM com o título: Atlas multimídia sobre movimentos de massa na sera do Mar paulista: município de Cubatão/SP / Doutor

abstract: With the rise of social media, hundreds of millions of people spend countless hours all over the globe on social media to connect, interact, share, and create user-generated data. This rich environment provides tremendous opportunities for many different players to easily and effectively reach out to people, interact with them, influence them, or get their opinions. There are two pieces of information that attract most attention on social media sites, including user preferences and interactions. Businesses and organizations use this information to better understand and therefore provide customized services to social media users. This data can be used for different purposes such as, targeted advertisement, product recommendation, or even opinion mining. Social media sites use this information to better serve their users. Despite the importance of personal information, in many cases people do not reveal this information to the public. Predicting the hidden or missing information is a common response to this challenge. In this thesis, we address the problem of predicting user attributes and future or missing links using an egocentric approach. The current research proposes novel concepts and approaches to better understand social media users in twofold including, a) their attributes, preferences, and interests, and b) their future or missing connections and interactions. More specifically, the contributions of this dissertation are (1) proposing a framework to study social media users through their attributes and link information, (2) proposing a scalable algorithm to predict user preferences; and (3) proposing a novel approach to predict attributes and links with limited information. The proposed algorithms use an egocentric approach to improve the state of the art algorithms in two directions. First by improving the prediction accuracy, and second, by increasing the scalability of the algorithms. / Dissertation/Thesis / Doctoral Dissertation Computer Science 2014

Neste trabalho projetamos e construímos um terminal de baixo custo para o sistema de automação de laboratórios em desenvolvimento no L.I.E. Para tanto, discutimos as diversas arquiteturas empregadas nos terminais buscando obter subsídios com vistas à escolha da mais adequada as aplicações cientificas que são típicas desse microcomputador. Também é feito um estudo comparativo dos diversos dispositivos empregados como terminais, bem como das funções que desempenham nas disposições convencionais dos centros computacionais. Procurou-se implementar um terminal de custo extremamente reduzido, mas capaz de ser facilmente expandido pela incorporação de módulos adicionais. Também foi objetivo do trabalho prover um projeto sólido para permitir facilmente uma vasta gama de alterações, variando desde a substituição de alguns componentes até a troca de todo o sistema operacional residente. Para tanto houve grande cuidado na documentação dos esquemas e da programação. O dispositivo construído mostrou-se bastante versátil, confiável e baixíssimo custo / In the present work we compare several devices used as terminals, as well as the roles they play on a conventional computing centre. Their various architectural aspects are also investigated, providing the ground basis for a terminal biased towards scientific aplications, which are typical of the microcomputer system developed at L.I.E.. We wanted a very low cost traminal that still retained the capability of beeing expanded by the inclusion of additional modules. A solid Project, able to handle the most severe alterations, ranging from the replacement of some ICs to the replacement of the entire resident operational system was aso wanted. To accomplish these goals a great care was taken on the Project documentation of the software as well as of the hardware. The device that was eventually build proved to be very versatile, reliable and low cost one

Artificial Neural Networks (ANNs) appear increasingly and routinely to gain popularity today, as they are being used in several diverse research fields and many different contexts, which may range from biological simulations and experiments on artificial neuronal models to machine learning models intended for industrial and engineering applications. One example is the recent success of Deep Learning architectures (e.g., Deep Belief Networks [DBN]), which appear in the spotlight of machine learning research, as they are capable of delivering state-of-the-art results in many domains. While the performance of such ANN architectures is greatly affected by their scale, their capacity for scalability both for training and during execution is limited by the increased power consumption and communication overheads, implicitly posing a limiting factor on their real-time performance. The on-going work on the design and construction of spike-based neuromorphic platforms offers an alternative for running large-scale neural networks, such as DBNs, with significantly lower power consumption and lower latencies, but has to overcome the hardware limitations and model specialisations imposed by these type of circuits. SpiNNaker is a novel massively parallel fully programmable and scalable architecture designed to enable real-time spiking neural network (SNN) simulations. These properties render SpiNNaker quite an attractive neuromorphic exploration platform for running large-scale ANNs, however, it is necessary to investigate thoroughly both its power requirements as well as its communication latencies. This research focusses on around two main aspects. First, it aims at characterising the power requirements and communication latencies of the SpiNNaker platform while running large-scale SNN simulations. The results of this investigation lead to the derivation of a power estimation model for the SpiNNaker system, a reduction of the overall power requirements and the characterisation of the intra- and inter-chip spike latencies. Then it focuses on a full characterisation of spiking DBNs, by developing a set of case studies in order to determine the impact of (a) the hardware bit precision; (b) the input noise; (c) weight variation; and (d) combinations of these on the classification performance of spiking DBNs for the problem of handwritten digit recognition. The results demonstrate that spiking DBNs can be realised on limited precision hardware platforms without drastic performance loss, and thus offer an excellent compromise between accuracy and low-power, low-latency execution. These studies intend to provide important guidelines for informing current and future efforts around developing custom large-scale digital and mixed-signal spiking neural network platforms.