Submitted by EVERTON SIMÕES DA MOTTA (evmotta08@gmail.com) on 2017-01-05T13:08:55Z No. of bitstreams: 1 DissertacaoFinalCorrigida-v5.pdf: 7335404 bytes, checksum: fa4ba2ff69cd4d2dd72b1b5cb475111b (MD5) / Approved for entry into archive by Juliano Benedito Ferreira (julianoferreira@reitoria.unesp.br) on 2017-01-09T17:01:51Z (GMT) No. of bitstreams: 1 motta_es_me_rcla.pdf: 7335404 bytes, checksum: fa4ba2ff69cd4d2dd72b1b5cb475111b (MD5) / Made available in DSpace on 2017-01-09T17:01:51Z (GMT). No. of bitstreams: 1 motta_es_me_rcla.pdf: 7335404 bytes, checksum: fa4ba2ff69cd4d2dd72b1b5cb475111b (MD5) Previous issue date: 2016-12-22 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / Sistemas de captura de movimentos humanos vêm sendo cada dia mais estudados, tanto pela área de Visão Computacional, quanto por grandes empresas do setor de entretenimento. São sistemas capazes de rastrear a posição e orientação das articulações do corpo e sua trajetória no espaço durante um intervalo de tempo. São utilizados em diversas aplicações, tais como em jogos digitais, animação de personagens virtuais para cinema e televisão, reconhecimento gestual, medicina de reabilitação, e outras. O surgimento de novos dispositivos de baixo custo e boa resolução que fornecem informações de profundidade, tem motivado novas pesquisas para a utilização dos mesmos. No entanto, sistemas que se baseiam somente em informações de profundidade (geralmente sistemas em tempo real) não apresentam uma alta acurácia no rastreamento do movimento. Considerando este contexto, o presente trabalho teve como objetivo principal, o desenvolvimento de um método de captura de movimentos humanos utilizando único sensor RGB-D, combinando informações de textura da imagem e de profundidade, que são associados a um esqueleto virtual, conseguindo-se uma maior acurácia em comparação com os métodos baseados apenas em profundidade. Tal método não visa aplicações de tempo real, mas sim, uma maior acurácia em comparação com os métodos baseados apenas em profundidade. / Human motion capture systems are being increasingly studied, in the area of computer vision and also by major entertainment industries. These systems are able to track the position and orientation of joints of the body and its trajectory in space over a period of time. They are used in various applications such as digital games, animation of virtual characters for film and television, gesture recognition, medical rehabilitation, etc. The emergence of new low-cost and good resolution devices that provide depth information has prompted new research. However, systems that are based only on depth information (usually real-time systems) do not present a high accuracy in movement tracking. Considering this context, this thesis project presents the development of a method for capturing human movements using only one RGB-D sensor, combining captured texture from the image and depth information in order to obtain a higher accuracy, which are associated with a virtual skeleton. This method is not intended for real-time applications, but those that require greater accuracy, such as the animation of virtual characters in video and medical rehabilitation.

SkePU 3 is a framework for parallel program execution that uses higher order functions called skeletons, which provide a layer of abstraction between user code and the parallel implementation it provides through its backends. The backend that enables SkePU to run on an HPC cluster has a slowdown of a factor two. This reduces the viability of SkePU as an alternative for HPC, and as such, warrants an investigation. Programs written in SkePU are sequential-looking, single-source C++ programs where skeleton calls can transparently execute on multiple different types of processing units, such as CPU cores, GPUs and clusters, using different backends. In this thesis, a strategy for improving the performance of SkePU on clusters is presented, and with it, the design and implementation of a new cluster backend that is simpler and more closely integrated with the non-cluster SkePU code base. Runtime measurements are made, which show that the new cluster backend sees a relative speedup of about a factor of two, which effectively eliminates the slowdown.

This thesis deals with the important problem of parallelising sequential code. Despite the importance of parallelism in modern computing, writing parallel software still relies on many low-level and often error-prone approaches. These low-level approaches can lead to serious execution problems such as deadlocks and race conditions. Due to the non-deterministic behaviour of most parallel programs, testing parallel software can be both tedious and time-consuming. A way of providing guarantees of correctness for parallel programs would therefore provide significant benefit. Moreover, even if we ignore the problem of correctness, achieving good speedups is not straightforward, since this generally involves rewriting a program to consider a (possibly large) number of alternative parallelisations. This thesis argues that new languages and frameworks are needed. These language and frameworks must not only support high-level parallel programming constructs, but must also provide predictable cost models for these parallel constructs. Moreover, they need to be built around solid, well-understood theories that ensure that: (a) changes to the source code will not change the functional behaviour of a program, and (b) the speedup obtained by doing the necessary changes is predictable. Algorithmic skeletons are parametric implementations of common patterns of parallelism that provide good abstractions for creating new high-level languages, and also support frameworks for parallel computing that satisfy the correctness and predictability requirements that we require. This thesis presents a new type-based framework, based on the connection between structured parallelism and structured patterns of recursion, that provides parallel structures as type abstractions that can be used to statically parallelise a program. Specifically, this thesis exploits hylomorphisms as a single, unifying construct to represent the functional behaviour of parallel programs, and to perform correct code rewritings between alternative parallel implementations, represented as algorithmic skeletons. This thesis also defines a mechanism for deriving cost models for parallel constructs from a queue-based operational semantics. In this way, we can provide strong static guarantees about the correctness of a parallel program, while simultaneously achieving predictable speedups.

No description available.

It has been discovered during the past few years that demosponges of the order Verongida such as Aplysina cavernicola exhibit chitin-based skeletons. Verongida sponges are well known to produce bioactive brominated tyrosine derivatives. We could recently demonstrate that brominated compounds do not exclusively occur in the cellular matrix but also in the skeletons of the marine sponges Aplysina cavernicola and Ianthella basta. Our measurements imply that these yet unknown compounds are strongly, possibly covalently bound to the sponge skeletons. In the present work, we determined the skeletal amino acid composition of the demosponge A. cavernicola especially with respect to the presence of halogenated amino acids. The investigations of the skeletons before and after MeOH extraction confirmed that only a small amount of the brominated skeleton-bound compounds dissolves in MeOH. The main part of the brominated compounds is strongly attached to the skeletons but can be extracted for example by using Ba(OH)2. Various halogenated tyrosine derivatives were identified by GC-MS and LC-MS in these Ba(OH)2 extracts of the skeletons.

Demosponges of the order Verongida such as Ianthella basta exhibit skeletons containing spongin, a collagenous protein, and chitin. Moreover, Verongida sponges are well known to produce bioactive brominated tyrosine derivatives. We recently demonstrated that brominated compounds do not only occur in the cellular matrix but also in the skeletons of the marine sponges Aplysina cavernicola and I. basta. Further investigations revealed the amino acid composition of the skeletons of A. cavernicola including the presence of several halogenated amino acids. In the present work, we investigated the skeletal amino acid composition of the demosponge I. basta, which belongs to the Ianthellidae family, and compared it with that of A. cavernicola from the Aplysinidae family. Seventeen proteinogenic and five non-proteinogenic amino acids were detected in I. basta. Abundantly occurring amino acids like glycine and hydroxyproline show the similarity of I. basta and A. cavernicola and confirm the collagenous nature of their sponging fibers. We also detected nine halogenated tyrosines as an integral part of I. basta skeletons. Since both sponges contain a broad variety of halogenated amino acids, this seems to be characteristic for Verongida sponges. The observed differences of the amino acid composition confirm that spongin exhibits a certain degree of variability even among the members of the order Verongida.

This study focuses on the attitudes of museumgoers toward the exhibition of human remains in modern Swedish museums. More specifically, it deals with how their attitudes toward remains are shaped and informed by museums’ materiality and institutionalized authority, whether they think of remains as humans or objects, as well as how these dead bodies ultimately become culturally meaningful to us who are still alive. Through the use of ethnographic field methods including go-along interviews and participant observation, the behaviors and opinions of museumgoers are recorded. With the help of Emmanuel Levinas’ ethical phenomenology and Annemarie Mol’s theory of multiple ontology, the cultural background against which these attitudes have taken shape is examined. The study shows that museumgoers are generally positive toward the exhibition of human remains in museums—an attitude which is influenced by the history, scientific authority, and carefully designed materiality of the museum. Among the perceived benefits of exhibiting human remains, visitors cite the ability of the remains to arouse their curiosity and serve as links to the past, as well as provide material proof that validates the museum’s claims to knowledge. However, this positive attitude is dependent on whether the remains are treated and displayed with respect. This call to treat the dead respectfully can be seen as a universal reaction to the ”face-to-face encounter” as described by Emmanuel Levinas, but at the same time, museumgoers’ interpretation of respect is culturally contingent and heavily influenced by contemporary values such as individuality, scientific objectivity, and equality. Furthermore, the perceived need to treat remains respectfully is directly tied to the perceived humanity of the remains. This is in turn dependent on how close the remains are to us in terms of appearance and temporal distance. By focusing on museumgoers instead of professionals, and through using ethnographic fieldwork to note opinions and their cultural backgrounds, this study attempts to add a fresh perspective and new knowledge to what is currently one of the hottest debates in museology: whether remains even belong in museums. Further, by recognizing that no opinion is formed in a vacuum, the narrow question of displaying death can tell us something bigger about the norms and values of Sweden today.

As processors' clock-speed flattened out in the early 2000s, multi-core processors became more prevalent and so did parallel programming. However this programming paradigm introduces additional complexities, and to combat this, the SkePU framework was created. SkePU does this by offering a single-threaded interface which executes the user's code in parallel in accordance to a chosen computational pattern. Furthermore it allows the user themselves to decide which parallel backend should perform the execution, be it OpenMP, CUDA or OpenCL. This modular approach of SkePU thus allows for different hardware to be used without changing the code, and it currently supports CPUs, GPUs and clusters. This thesis presents a new so-called SkePU-backend made for clusters, using the communication library GPI. It demonstrates that the new backend is able to scale better and handle workload imbalances better than the existing SkePU-cluster-backend. This is achieved despite it performing worse at low node amounts, indicating that it requires less scaling overhead. Its weaknesses are also analyzed, partially from a design point of view, and clear solutions are presented, combined with a discussion as to why they arose in the first place.

[pt] A sísmica é uma importante ferramenta utilizada no processo de exploração de petróleo e gás natural. A partir dos estudos sísmicos é possível obter informações referentes a probabilidade de encontrar situações favoráveis a acumulação de hidrocarbonetos. O presente trabalho visa integrar os dados adquiridos através deste método geofísico a um modelo de simulação de canais baseados em esqueletos em um ambiente deposicional turbidítico, e também apresentar a modelagem de tais canais condicionados a localização de um poço. / [en] The use of seismic data is an important tool in oil and gas research. It can show us the probability of having a high concentration of hydrocarbon in a possible reservoir. This work intends to condition skeleton-based modeling of channels reservoir in a turbidite system to seismic data. We also present such modeling process constraint by a well previously defined.

Modelling objects and simulating them do not always map to each other, and often requires defining additional information outside the scope of the original model to achieve an accurate simulation. For example: cables in \textit{AGX Dynamics} (a simulation library from Algoryx AB) are entirely defined by its physical parameters (e.g. Young's modulus, stiffness, etc.), radius, and the route through which the cables run. This thesis explores two approaches to closing the gap between the modelling of a cable and the creation of one in AGX Dynamics through evaluating current methods applied to generating a route and radius from a mesh. Two methods are identified as being useful in generating a route for a cable from a mesh: one which is a surface simplification algorithm, creating approximations of models using non-manifold meshes with radii defined at each vertex, and another method which creates a skeleton from a model using the surface's curvature to gradually shrink the model into a zero-volume shape. Both methods are evaluated using two different approaches to measuring the closeness to the original mesh from the results: using the metric introduced in the surface simplification method applied along the route, and measuring the mean distance from each point on the surface to the route. We show a clear advantage in the first method's inherent way of approximating the radius of the model but also its lack of detail. We also demonstrate that the second method produces more detailed skeletons, but in turn has issues with skewed routes which do not follow the original mesh. Both methods have their own advantages and disadvantages and with improvements to both radius calculations or adaptions to the fundamental algorithms, they could provide a great way of creating AGX cables from mesh models.