Scientific machine learning (SciML) is a new branch of AI research at the edge of scientific computing (Sci) and machine learning (ML). It deals with efficient amalgamation of data-driven algorithms along with scientific computing to discover the dynamics of the time-evolving process. The output of such algorithms is represented in the form of a governing equation(s) (e.g., ordinary differential equation(s), ODE(s)), which one can solve then for any time point and, thus, obtain a rigorous prediction.  In this thesis, we present a methodology on how to incorporate the SciML approach in the context of clinical trials to predict IPF disease progression in the form of governing equation. Our proposed methodology also quantifies the uncertainties associated with the model by fitting 95\% high density interval (HDI) for the ODE parameters and 95\% posterior prediction interval for posterior predicted samples. We have also investigated the possibility of predicting later outcomes by using the observations collected at early phase of the study. We were successful in combining ML techniques, statistical methodologies and scientific computing tools such as bootstrap sampling, cubic spline interpolation, Bayesian inference and sparse identification of nonlinear dynamics (SINDy) to discover the dynamics behind the efficacy outcome as well as in quantifying the uncertainty of the parameters of the governing equation in the form of 95 \% HDI intervals. We compared the resulting model with the existed disease progression model described by the Weibull function. Based on the mean squared error (MSE) criterion between our ODE approximated values and population means of respective datasets, we achieved the least possible MSE of 0.133,0.089,0.213 and 0.057. After comparing these MSE values with the MSE values obtained after using Weibull function, for the third dataset and pooled dataset, our ODE model performed better in reducing error than the Weibull baseline model by 7.5\% and 8.1\%, respectively. Whereas for the first and second datasets, the Weibull model performed better in reducing errors by 1.5\% and 1.2\%, respectively. Comparing the overall performance in terms of MSE, our proposed model approximates the population means better in all the cases except for the first and second datasets, assuming the latter case's error margin is very small. Also, in terms of interpretation, our dynamical system model contains the mechanistic elements that can explain the decay/acceleration rate of the efficacy endpoint, which is missing in the Weibull model. However, our approach had a limitation in predicting final outcomes using a model derived from  24, 36, 48 weeks observations with good accuracy where as on the contrast, the Weibull model do not possess the predicting capability. However, the extrapolated trend based on 60 weeks of data was found to be close to population mean and the ODE model built on 72 weeks of data. Finally we highlight potential questions for the future work.

The emperical pseudopotential method in the large basis approach was used to calculate the electronic bandstructures of bulk semiconductor materials and layered semiconductor heterostructures. The crucial continuous atomic form factor potentials needed to carry out such calculations were determined by using Levenberg-Marquardt optimization in order to obtain optimal cubic spline interpolations of the potentials. The optimized potentials were not constrained by any particular functional form (such as a linear combination of Gaussians) and had better convergence properties for the optimization. It was demonstrated that the results obtained in this work could potentially lead to better agreement between calculated and empirically determined band gaps via optimization / Physics / M. Sc. (Physics)

Cette thèse a pour objectif d'étudier et d'analyser une nouvelle stratégie de compression, dont le principe consiste à compresser conjointement des données issues de plusieurs modalités, en utilisant un codeur unique. Cette approche est appelée « Compression Multimodale ». Dans ce contexte, une image et un signal audio peuvent être compressés conjointement et uniquement par un codeur d'image (e.g. un standard), sans la nécessité d'intégrer un codec audio. L'idée de base développée dans cette thèse consiste à insérer les échantillons d'un signal en remplacement de certains pixels de l'image « porteuse » tout en préservant la qualité de l'information après le processus de codage et de décodage. Cette technique ne doit pas être confondue aux techniques de tatouage ou de stéganographie puisqu'il ne s'agit pas de dissimuler une information dans une autre. En Compression Multimodale, l'objectif majeur est, d'une part, l'amélioration des performances de la compression en termes de débit-distorsion et d'autre part, l'optimisation de l'utilisation des ressources matérielles d'un système embarqué donné (e.g. accélération du temps d'encodage/décodage). Tout au long de ce rapport, nous allons étudier et analyser des variantes de la Compression Multimodale dont le noyau consiste à élaborer des fonctions de mélange et de séparation, en amont du codage et de séparation. Une validation est effectuée sur des images et des signaux usuels ainsi que sur des données spécifiques telles que les images et signaux biomédicaux. Ce travail sera conclu par une extension vers la vidéo de la stratégie de la Compression Multimodale / The objective of this thesis is to study and analyze a new compression strategy, whose principle is to compress the data together from multiple modalities by using a single encoder. This approach is called “Multimodal Compression” during which, an image and an audio signal is compressed together by a single image encoder (e.g. a standard), without the need for an integrating audio codec. The basic idea developed in this thesis is to insert samples of a signal by replacing some pixels of the "carrier's image” while preserving the quality of information after the process of encoding and decoding. This technique should not be confused with techniques like watermarking or stéganographie, since Multimodal Compression does not conceal any information with another. Two main objectives of Multimodal Compression are to improve the compression performance in terms of rate-distortion and to optimize the use of material resources of a given embedded system (e.g. acceleration of encoding/decoding time). In this report we study and analyze the variations of Multimodal Compression whose core function is to develop mixing and separation prior to coding and separation. Images and common signals as well as specific data such as biomedical images and signals are validated. This work is concluded by discussing the video of the strategy of Multimodal Compression

Made available in DSpace on 2016-03-22T17:26:24Z (GMT). No. of bitstreams: 1 Wagner Guimaraes Al-Alam.pdf: 2995290 bytes, checksum: 9902eafe02c0b5318a99f1e796dc399f (MD5) Previous issue date: 2010-01-15 / Companies working with bulk materials have appropriate locations for storage during the development of the production and storage of the final product, known as warehouses or storehouses. The values of stocks need to be periodically validated by comparing the control of receipts the and the physical situation (removal of the volume stored in the company). In this context, the calculation of physical inventory as the volume of bulk present in the warehouses is usually done manually with low credibility and prone to errors. The current audit procedures on the contents of warehouses involve inaccurate estimates, and often require emptying the warehouse. Considering the use of technologies which enable the electronic measurement of distances, angles, and automatic controls on actuators enabling mechanical movements on the supporting structures, we sought to develop a system capable of providing both computing solutions, and technology for the problem of calculation of irregular relief (products stocked in warehouses). The Automated Auditing Warehouse SisA3 intends to make this process automatic, fast and precise, without the need for emptying warehouses or having contact the products. To achieve this goal, we developed an integrated system composed of: (i) a scanner equipment, consoling the hybrid prototype of hardware and software called DigSisA3, in order to the measurement of points of relief non-uniform, formed by the products in stock, and (ii) a method for calculating the volume iCone, which combines techniques of scientific visualization, numerical interpolation points and iterative calculation of volume. The parallelization of the prototype iCone was also developed in order to satisfy the test of agility and performance of the method iCone in the audit process. The development for multiprocessor, multi-core, and distributed architectures was done over the DGM (Geometric Distributed Machine), which provides the formalities to ensure creation, management and application processing parallel and / or distributed scientific computing, with emphasis on the exploitation of data parallelism and synchronization steps. The prototype of software iCone was functionally validated, including analysis of error in the method. The analysis of performance in the prototype p-iCone showed satisfactory results. The development of this work strengthens the system SisA3, enabling automatic and reliable measurement of inventories, including broad market application / Empresas que trabalham com produtos a granel possuem locais para estocagem, durante o desenvolvimento do processo produtivo e no armazenamento do produto final, denominados armaz´ens ou silos. Os valores dos estoques devem ser validados periodicamente atrav´es da comparac¸ ao dos estoques fiscal (controle das notas fiscais) e f´ısico (levantamento do volume estocado na empresa). Neste contexto, o c´alculo do estoque f´ısico, ou seja, o volume de gran´eis presentes nos armaz´ens, ´e geralmente efetuado de forma manual e com baixa credibilidade, desta forma com propens ao a erros. Os atuais processos de auditoria no conte´udo de silos, al´em de envolverem estimativas inexatas, est ao frequentemente baseados no esvaziamento do silo. Considerando o uso de tecnologias que viabilizam a medic¸ ao eletr onica de dist ancias, angulos, e controles autom´aticos sobre atuadores que possibilitam movimentos mec anicos sobre estruturas de suporte, buscou-se o desenvolvimento de um sistema capaz de prover tanto soluc¸ oes computacionais, quanto tecnol´ogicas para o problema de c´alculo do volume de relevos irregulares, no caso dos produtos estocados nos armaz´ens. O Sistema Automatizado de Auditoria em Armaz´ens (SisA3) pretende tornar este processo autom´atico, r´apido e preciso, sem a necessidade de esvaziamento ou contato com os produtos. Para alcanc¸ar este objetivo, tem-se um sistema integrado composto de: (i) um equipamento digitalizador, consolidando o prot´otipo h´ıbrido de hardware e software denominado Dig-SisA3 , para a medic¸ ao de pontos do relevo n ao-uniforme, formado pelos produtos estocados; e (ii) m´etodo para o c´alculo do volume (iCone), que combina t´ecnicas de visualizac¸ ao cient´ıfica, interpolac¸ ao num´erica de pontos e c´alculo iterativo de volume. Al´em disto, introduz-se a paralelizac¸ ao do prot´otipo iCone, para diminuir o tempo da obtenc¸ ao dos resultados do m´etodo iCone no processo de auditoria. A an´alise sobre as perspectivas em arquiteturas multiprocessadas, multi-core e paralela distribu´ıda, utiliza o ambiente D-GM (Distributed Geometric Machine), a qual prov e os formalismos para garantir criac¸ ao, gerenciamento e processamento de aplicac¸ oes paralelas e/ou distribu´ıdas da computac¸ ao cient´ıfica, com enfase na explorac¸ ao do paralelismo de dados e nas etapas de sincronizac¸ oes. O prot´otipo de software iCone apresenta-se funcionalmente validado, incluindo an´alise de erro na execuc¸ ao do m´etodo. As an´alises de desempenho no prot´otipo p-iCone apresentaram resultados satisfat´orios. O desenvolvimento deste trabalho consolida o sistema SisA3, viabilizando aferic¸ ao autom´atica e confi´avel de estoques, incluindo ampla aplicac¸ ao no mercado

The emperical pseudopotential method in the large basis approach was used to calculate the electronic bandstructures of bulk semiconductor materials and layered semiconductor heterostructures. The crucial continuous atomic form factor potentials needed to carry out such calculations were determined by using Levenberg-Marquardt optimization in order to obtain optimal cubic spline interpolations of the potentials. The optimized potentials were not constrained by any particular functional form (such as a linear combination of Gaussians) and had better convergence properties for the optimization. It was demonstrated that the results obtained in this work could potentially lead to better agreement between calculated and empirically determined band gaps via optimization / Physics / M. Sc. (Physics)

Turbulent phenomena are found in both natural (e.g. the Earth's oceans, the Sun's corona) and artificial (e.g. flows through pipes, the plasma in a tokamak device) settings; evidence suggests that turbulence is usually the normal behaviour in most cases. Turbulence has been studied extensively for more than a century, but a complete and consistent theoretical description of it has not yet been proposed. It is in this context that the motion of particles under the influence of turbulent fields is studied in this work, with direct numerical simulations. The thesis is structured in three main parts. The first part describes the tools that are used. Methods of integrating particle trajectories are presented, together with a discussion of the properties that these methods should have. The simulation of magnetohydrodynamic (MHD) turbulence is discussed, while also introducing fundamental concepts of fluid turbulence. Particle trajectory integration requires information that is not readily available from simulations of turbulent flows, so the interpolation methods needed to adapt the fluid simulation results are constructed as well. The second part is dedicated to the study of two MHD problems. Simulations of Kolmogorov flow in incompressible MHD are presented and discussed, and also simulations of the dynamo effect in compressible MHD. These two scenarios are chosen because large scale structures are formed spontaneously by the turbulent flow, and there is an interest in studying particle transport in the presence of structures. Studies of particle transport are discussed in the third part. The properties of the overall approach are first analyzed in detail, for stationary predefined fields. Focus is placed on the qualitative properties of the different methods presented. Charged article transport in frozen turbulent fields is then studied. Results concerning transport of particles in fully developed, time-evolving, turbulent fields are presented in the final chapter.<p><p><p>\ / Doctorat en Sciences / info:eu-repo/semantics/nonPublished