No description available.

Eine Hauptanwendung der Einzelpartikel-Analyse in der Kryo-Elektronenmikroskopie ist die Charakterisierung der dreidimensionalen Struktur makromolekularer Komplexe. Dazu werden zehntausende Bilder verwendet, die verrauschte zweidimensionale Projektionen des Partikels zeigen. Im ersten Schritt werden ein niedrig aufgelöstetes Anfangsmodell rekonstruiert sowie die unbekannten Bildorientierungen geschätzt. Dies ist ein schwieriges inverses Problem mit vielen Unbekannten, einschließlich einer unbekannten Orientierung für jedes Projektionsbild. Ein gutes Anfangsmodell ist entscheidend für den Erfolg des anschließenden Verfeinerungsschrittes. Meine Dissertation stellt zwei neue Algorithmen zur Rekonstruktion eines Anfangsmodells in der Kryo-Elektronenmikroskopie vor, welche auf einer groben Darstellung der Elektronendichte basieren. Die beiden wesentlichen Beiträge meiner Arbeit sind zum einen das Modell, welches die Elektronendichte darstellt, und zum anderen die neuen Rekonstruktionsalgorithmen. Der erste Hauptbeitrag liegt in der Verwendung Gaußscher Mischverteilungen zur Darstellung von Elektrondichten im Rekonstruktionsschritt. Ich verwende kugelförmige Mischungskomponenten mit unbekannten Positionen, Ausdehnungen und Gewichtungen. Diese Darstellung hat viele Vorteile im Vergleich zu einer gitterbasierten Elektronendichte, die andere Rekonstruktionsalgorithmen üblicherweise verwenden. Zum Beispiel benötigt sie wesentlich weniger Parameter, was zu schnelleren und robusteren Algorithmen führt. Der zweite Hauptbeitrag ist die Entwicklung von Markovketten-Monte-Carlo-Verfahren im Rahmen eines Bayes'schen Ansatzes zur Schätzung der Modellparameter. Der erste Algorithmus kann aus dem Gibbs-Sampling, welches Gaußsche Mischverteilungen an Punktwolken anpasst, abgeleitet werden. Dieser Algorithmus wird hier so erweitert, dass er auch mit Bildern, Projektionen sowie unbekannten Drehungen und Verschiebungen funktioniert. Der zweite Algorithmus wählt einen anderen Zugang. Das Vorwärtsmodell nimmt nun Gaußsche Fehler an. Sampling-Algorithmen wie Hamiltonian Monte Carlo (HMC) erlauben es, die Positionen der Mischungskomponenten und die Bildorientierungen zu schätzen. Meine Dissertation zeigt umfassende numerische Experimente mit simulierten und echten Daten, die die vorgestellten Algorithmen in der Praxis testen und mit anderen Rekonstruktionsverfahren vergleichen.

Recent advances in statistical machine translation (SMT) have used dynamic programming (DP) based beam search methods for approximate inference within probabilistic translation models. Despite their success, these methods compromise the probabilistic interpretation of the underlying model thus limiting the application of probabilistically defined decision rules during training and decoding. As an alternative, in this thesis, we propose a novel Monte Carlo sampling approach for theoretically sound approximate probabilistic inference within these models. The distribution we are interested in is the conditional distribution of a log-linear translation model; however, often, there is no tractable way of computing the normalisation term of the model. Instead, a Gibbs sampling approach for phrase-based machine translation models is developed which obviates the need of computing this term yet produces samples from the required distribution. We establish that the sampler effectively explores the distribution defined by a phrase-based models by showing that it converges in a reasonable amount of time to the desired distribution, irrespective of initialisation. Empirical evidence is provided to confirm that the sampler can provide accurate estimates of expectations of functions of interest. The mix of high probability and low probability derivations obtained through sampling is shown to provide a more accurate estimate of expectations than merely using the n-most highly probable derivations. Subsequently, we show that the sampler provides a tractable solution for finding the maximum probability translation in the model. We also present a unified approach to approximating two additional intractable problems: minimum risk training and minimum Bayes risk decoding. Key to our approach is the use of the sampler which allows us to explore the entire probability distribution and maintain a strict probabilistic formulation through the translation pipeline. For these tasks, sampling allies the simplicity of n-best list approaches with the extended view of the distribution that lattice-based approaches benefit from, while avoiding the biases associated with beam search. Our approach is theoretically well-motivated and can give better and more stable results than current state of the art methods.

We present a framework for learning in hidden Markov models with distributed state representations. Within this framework, we derive a learning algorithm based on the Expectation--Maximization (EM) procedure for maximum likelihood estimation. Analogous to the standard Baum-Welch update rules, the M-step of our algorithm is exact and can be solved analytically. However, due to the combinatorial nature of the hidden state representation, the exact E-step is intractable. A simple and tractable mean field approximation is derived. Empirical results on a set of problems suggest that both the mean field approximation and Gibbs sampling are viable alternatives to the computationally expensive exact algorithm.

Submitted by Ronildo Prado (ronisp@ufscar.br) on 2017-08-22T14:32:30Z No. of bitstreams: 1 DissRCA.pdf: 1727058 bytes, checksum: 78d5444e767bf066e768b88a3a9ab535 (MD5) / Approved for entry into archive by Ronildo Prado (ronisp@ufscar.br) on 2017-08-22T14:32:38Z (GMT) No. of bitstreams: 1 DissRCA.pdf: 1727058 bytes, checksum: 78d5444e767bf066e768b88a3a9ab535 (MD5) / Approved for entry into archive by Ronildo Prado (ronisp@ufscar.br) on 2017-08-22T14:32:44Z (GMT) No. of bitstreams: 1 DissRCA.pdf: 1727058 bytes, checksum: 78d5444e767bf066e768b88a3a9ab535 (MD5) / Made available in DSpace on 2017-08-22T14:32:50Z (GMT). No. of bitstreams: 1 DissRCA.pdf: 1727058 bytes, checksum: 78d5444e767bf066e768b88a3a9ab535 (MD5) Previous issue date: 2017-06-02 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / We present the topics and theory of Mixture Models in a context of maximum likelihood and Bayesian inferece. We approach clustering methods in both contexts, with emphasis on the stochastic EM algorithm and the Dirichlet Process Mixture Model. We propose a new method, a modified stochastic EM algorithm, which can be used to estimate the parameters of a mixture model and the number of components. / Apresentamos o tópico e a teoria de Modelos de Mistura de Distribuições, revendo aspectos teóricos e interpretações de tais misturas. Desenvolvemos a teoria dos modelos nos contextos de máxima verossimilhança e de inferência bayesiana. Abordamos métodos de agrupamento já existentes em ambos os contextos, com ênfase em dois métodos, o algoritmo EM estocástico no contexto de máxima verossimilhança e o Modelo de Mistura com Processos de Dirichlet no contexto bayesiano. Propomos um novo método, uma modificação do algoritmo EM Estocástico, que pode ser utilizado para estimar os parâmetros de uma mistura de componentes enquanto permite soluções com número distinto de grupos.

The objective of this study was to estimate genetic and phenotypic parameters, besides the genetic progress for the performance traits post-weaning in a population of Charolais cattle. Components of (co)variance were estimated using an animal model by Bayesian inference method. In the first article was analyzed the average daily gain from weaning to yearling (ADGWY) and weight adjusted to 550 days of age (W550) of 5.897 animals, sired by 181 bulls and 3.897 cows born between 1983 and 1999. The average for direct heritability and standard errors were 0.39±0.04 and 0.48±0.05 for ADGWY and W550, respectively. The genetic correlation between the two traits analyzed and the standard error was 0.60±0.05. The genetic trends for ADGWY and W550 were -0.058g/year and -0.019kg/. In the second article were estimated parameters and genetics trends for visual scores as well as their associations with the average daily gain from weaning to yearling (ADGWY) and weight adjusted to 550 days of age (W550) of 2.964 animals, sired by 145 bulls and 1.820 cows born between 1994 and 2007. The heritability estimates for scores of conformation (C), precocity (P), muscle (M) and size (S) were: 0.13±0.04, 0.23±0.04, 0.16±0.03 and 0.11±0.04, respectively. The genetic correlation between the four visual scores ranged between -0.03 and 0.83. The genetic correlations between ADGWY and W550 with visual scores were negative, ranging from - 0.07 to -0.82. As for the study of genetic progress, the genetic trends for C, P, M and S were: 0.0019, 0.0027, 0.0017 and 0.0011, respectively. / Este estudo teve como objetivo estimar parâmetros genéticos, além do progresso genético para as características de desempenho na pós-desmama em uma população bovina da raça Charolês. Os componentes de (co)variâncias foram estimados utilizando um modelo animal, através do método de inferência Bayesiana. No primeiro capítulo foram analisadas as características ganho médio diário da desmama ao sobreano (GMDDS) e o peso ajustado aos 550 dias de idade (P550) de 5.897 animais, filhos de 181 touros e 3.897 vacas, nascidos entre 1983 e 1999. As médias a posteriori para a herdabilidade direta e os erros padrões foram 0,39±0,04 e 0,48±0,05 para GMDDS e P550, respectivamente. A correlação genética entre as duas características analisadas e o erro padrão foi 0,60±0,05. As tendências genéticas estimadas para GMDDS e P550 foram -0,058g/ano e -0,019kg/ano. No segundo capítulo foram estimados parâmetros e tendências genéticas para os escores visuais, bem como suas associações com o ganho médio diário da desmama ao sobreano (GMDDS) e o peso ajustado aos 550 dias de idade (P550) de 2.964 animais, filhos de 145 touros e 1.820 vacas, nascidos entre 1994 e 2007. As herdabilidades estimadas para os escores de conformação (C), precocidade (P), musculatura (M) e tamanho (T) foram: 0,13±0,04, 0,23±0,04, 0,16±0,03, 0,11±0,04, respectivamente. A correlação genética entre os quatro escores variaram entre - 0,03 e 0,83. Já as correlações genéticas entre GMDDS e P550 com os escores visuais foram todas negativas, variando de -0,07 a -0,82. Quanto ao estudo do progresso genético, as tendências genéticas para C, P, M e T foram: 0,0019, 0,0027, 0,0017 e 0,0011, respectivamente.

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / The objective of this study was to estimate the (co) variances and genetic associations between the visual scores of conformation and precocity at weaning and at yearling, respectively WC, WP and YC, YP with the characteristics of average daily gain (BWG: from birth to weaning and WYG: from weaning to yearling) and rate of weight gain (BWR: from birth to weaning and BYR: from weaning to yearling) for Angus cattle. The (co)variances components were estimated by multi-traits analysis using an animal model by Bayesian Inference Method, assuming a linear model for average daily gain and for rate of weight gain and a non-linear (threshold) for WC, WP, YC and YP. The a posteriori means for direct heritability were 0.12 (WC), 0.17 (WP), 0.15 (BWG and BWR), 0.17 (YC), 0.19 (YP) and 0.16 (WYG and BYR). The genetic correlation between the weaning and yearling scores with the direct effect for the average daily gain and rate of weight gain traits ranged from -0.05 to 0.60 for conformation and -0.15 to 0.60 for precocity; the genetic correlation of the scores with average daily gain and rate of weight gain presented similar behavior. Therefore, correlated response of equal magnitude is expected for scores, if the selection was directed to average daily gain or rate of weight gain. / O objetivo deste estudo foi estimar as (co)variâncias e as associações genéticas entre os escores visuais de conformação e precocidade à desmama ao sobreano, respectivamente CD, PrD e CS, PrS, com as características de ganho médio diário de peso (GMD: do nascimento à desmama; GMS: da desmama ao sobreano) e velocidade de ganho de peso (VD: do nascimento à desmama e VS: da desmama ao sobreano) para bovinos da raça Angus. Os componentes de (co)variâncias foram estimados por um modelo animal tetra-característica usando o Método de Inferência Bayesiana, assumindo um modelo linear para o ganho médio diário e velocidade de ganho de peso e um modelo não-linear (de limiar) para CD, CS, PrD e PrS. As médias a posteriori para a herdabilidade direta foram: 0,12 (CD); 0,17 (PrD); 0,15 (GMD e VD); 0,17 (CS); 0,19 (PrS) e 0,16 (GMS e VS). As correlações genéticas estimadas entre os escores à desmana e ao sobreano com o efeito direto para as caracteristicas de crescimento variaram de -0,05 a 0,60 para conformação e de -0,15 a 0,60 para precocidade; a correlação genética dos escores com ganho médio diário ou com velocidade de ganho de peso apresentou comportamento semelhante. Portanto, resposta correlacionada de igual magnitude é esperada para os escores, se a selação for direcionada para ganho médio diario de peso ou velocidade de ganho de peso.

No description available.

In this thesis, we have proposed a new computing platform called probabilistic spin logic (PSL) based on probabilistic bits (p-bit) using low barrier nanomagnets (LBM) whose thermal barrier is of the order of a kT unlike conventional memory and spin logic devices that rely on high thermal barrier magnets (40-60 kT) to retain stability. p-bits are tunable random number generators (TRNG) analogous to the concept of binary stochastic neurons (BSN) in artificial neural network (ANN) whose output fluctuates between a +1 and -1 states with 50-50 probability at zero input bias and the stochastic output can be tuned by an applied input producing a sigmoidal characteristic response. p-bits can be interconnected by a synapse or weight matrix [J] to build p-circuits for solving a wide variety of complex unconventional problems such as inference, invertible Boolean logic, sampling and optimization. It is important to update the p-bits sequentially for proper operation where each p-bit update is informed of the states of other p-bits that it is connected to and this requires the use of sequencers in digital clocked hardware. But the unique feature of our probabilistic hardware is that they are autonomous that runs without any clocks or sequencers.<br>To ensure the necessary sequential informed update in our autonomous hardware it is important that the synapse delay is much smaller than the neuron fluctuation time.<br>We have demonstrated the notion of this autonomous hardware by SPICE simulation of different designs of low barrier nanomagnet based p-circuits for both symmetrically connected Boltzmann networks and directed acyclic Bayesian networks. It is interesting to note that for Bayesian networks a specific parent to child update order is important and requires specific design rule in the autonomous probabilistic hardware to naturally ensure the specific update order without any clocks. To address the issue of scalability of these autonomous hardware we have also proposed and benchmarked compact models for two different hardware designs against SPICE simulation and have shown that the compact models faithfully mimic the dynamics of the real hardware.<br>

No description available.