Este trabalho trata do controle ótimo de um Sistema de Programação de Admissões Hospitalares. Considera-se dois tipos de pacientes: graves (tipo I) a não graves (tipo II), cada um deles com sua própria taxa de chegada e serviço. O objetivo deste trabalho é encontrar uma política ótima de controle para o Sistema sw Programação de Admissões. Isto é, encontra a melhor seqüência de decisões a serem tomadas, nos instantes de decisão, com respeito aos valores das taxas de chegada e de serviço, baseando-se em observações acerca do número de clientes (pacientes) no sistema, a fim de minimizar o custo médio esperado num horizonte infinito de planejamento. A solução apresentada consiste em enfocar o sistema como um Processo Markoviano de Decisão a Tempo Contínuo. É apresentado o enfoque "What-if modelling" como uma ferramenta para melhorar a comprensão e prospecção do sistema.

O objetivo do trabalho é mostrar que a ferramenta de especificação Statecharts, criada para representar o comportamento de sistemas complexos que reagem a eventos, pode também ser utilizada para representar o comportamento dos modelos de desempenho. Um sistema complexo dado será modelado usando as características de Statecharts e através de uma reação a eventos estimulados todas as possíveis configurações do sistema serão geradas. Com este conjunto de todas as possíveis configurações do sistema complexo uma matriz de transições é criada. Esta matriz é a base para calcular probabilidades limite a partir das quais é possível determinar medidas de performance.

The main goal of this work is to present an approach that ?nds an appropriate set of sequential actions for a group of cooperative agents interacting over a constrained environment. This search is considered a complex task for autonomous agents and is not possible to use default reinforcement learning algorithms to learn the adequate policy. In this thesis, a technique that deals with propositionally constrained state spaces and makes use of a Reinforcement Learning algorithm based on Markov Decision Process is proposed. A new model is also presented which formally de?nes this restricted search space. By so doing, this work aims at reducing the overall exploratory need, thus improving the performance of the learning algorithm. To constrain the state space the concept of extended reachability goals is employed. Through them it is possible to de?ne an objective to be preserved during the iteration with the environment and another that de?nes a goal state. In this cooperative environment, the information about the propositions is shared among the agents during its interaction. An architecture to solve problems in such environments is also presented. Experiments to validate the proposed algorithm were performed on different test cases and showed interesting results. A performance evaluation against standard Reinforcement Learning techniques showed that by extending autonomous learning with propositional constraints updated along the learning process can produce faster convergence to adequate policies. The best results achieved present an important reduction over execution time (34,32%) and number of iterations (67.94%). This occurs due to the early state space reduction caused by shared information on state space constraints.

Made available in DSpace on 2016-12-23T14:07:25Z (GMT). No. of bitstreams: 1 Sandra Mara Torres Muller.pdf: 1594751 bytes, checksum: b3abd1c42aa0001991d29eefe9754019 (MD5) Previous issue date: 2006-04-03 / Conselho Nacional de Desenvolvimento Científico e Tecnológico / In this work three incremental adaptation methods for the hidden Markov models (HMM) are studied and implemented, which are based on the Expectation-Maximization (EM), Segmental k-Means and Maximum a Posteriori (MAP) algorithms. These methods, already used in the speech recognition field, are applied here in the electrocardiogram (ECG) segmentation problem. For that, it was used an ECG analysis system able to segment and classify cardiac diseases, like premature ventricular contraction (PVC) and ischemia. The use of these methods allow us to adjust the models to the signal fluctuations commonly met during ambulatory recording. The methods can also be implemented for other kinds of biomedical signals, like electroencephalogram (EEG). / Neste trabalho foram estudadas e implementadas tr^es t¶ecnicas incrementais de adapta»c~ ao de modelos ocultos de Markov (HMM - Hidden Markov Model) baseadas nos algoritmos de treinamento, que s~ ao a esperan»ca da maximiza»c~ ao (expectation maximization - EM), a k-means segmental (segmental k-means) e a m¶aximo a posteriori (Maximum a Posteri- ori -MAP). Essas t¶ecnicas, muito utilizadas em reconhecimento de voz, s~ ao aqui usadas para sinais biom¶edicos, mais precisamente para sinal de eletrocardiograma (ECG). Para tal objetivo, utilizou-se uma plataforma, j¶ a desenvolvida, de segmenta» c~ ao e classi¯ca» c~ ao de ECG, al¶em de detec»c~ oes de anomalias card¶³acas como extra-s¶³stole ventricular (ESV) e isquemia do mioc¶ ardio. Nessa plataforma, os modelos de Markov são empregados na etapa de segmenta»c~ ao do sinal de ECG, tendo em vista a identi¯ca» c~ ao das formas de onda elementares que comp~oem um ciclo card¶³aco. O desenvolvimento dessas t¶ecnicas permite, uma vez que a plataforma esteja funcionando como sistema real, um ajuste aut^ onomo dos modelos µas varia» c~ oes do sinal de ECG ao longo do tempo, assim como a outras varia» c~ oes presentes em um sistema real. As t¶ecnicas foram avaliadas a partir de experimentos usando duas bases de sinais de ECG: QT database e European ST-T database. Os resultados con¯rmam o ganho de desempenho obtido com a adapta»c~ ao, permitindo uma modelagem do sinal ao longo do tempo mais apropriada. As t¶ecnicas desenvolvidas s~ ao indicadas tamb¶em para outros tipos de sinais biom¶edicos, como o sinal de eletroencefalograma (EEG), por exemplo.

Submitted by camilla martins (camillasmmartins@gmail.com) on 2017-02-21T18:59:00Z No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Tese_StatisticalAnalysisMarkov.pdf: 2927895 bytes, checksum: e42c9c41994072a2730f00dce594bce1 (MD5) / Approved for entry into archive by Edisangela Bastos (edisangela@ufpa.br) on 2017-02-22T16:12:58Z (GMT) No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Tese_StatisticalAnalysisMarkov.pdf: 2927895 bytes, checksum: e42c9c41994072a2730f00dce594bce1 (MD5) / Made available in DSpace on 2017-02-22T16:12:58Z (GMT). No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Tese_StatisticalAnalysisMarkov.pdf: 2927895 bytes, checksum: e42c9c41994072a2730f00dce594bce1 (MD5) Previous issue date: 2015-06-26 / CNPq - Conselho Nacional de Desenvolvimento Científico e Tecnológico / The current trends in optical fiber communications are rapidly approaching the physical capacity limit of standard optical fiber. It is becoming increasingly important to efficiently utilize spectral resources wisely to accommodate the ever-increasing Internet traffic demand. However, the rigid and coarse ITU-T grid specifications regarding the spectrum usage restrict the granularity of bandwidth segmentation and allocation, which frequently causes a mismatch between the allocated and the actual requested link bandwidth. This often leads to over provisioning, where usually more resources are provided than necessary. Recently, the concept of elastic optical networks (EONs) has been proposed in order to reduce the waste of spectra resources. In networks with such feature enabled, modulation parameters and central frequencies are not fixed as in the traditional WDM networks: the resources can be allocated with fine granularity, which can adapt to the granularity of the requested bandwidth without over provisioning. This results in more efficient usage of spectral resources. However, elastic optical networks must satisfy dynamic connection add and drop over spectral resources that inevitable results in fragmentation of the spectrum. In EONs, spectrum fragmentation is an important and inevitable problem, because it reduces the spectral efficiency. As consequence, the blocking probability (BP) is increased due to scattered gaps in the optical grid. Currently, several metrics have been proposed in order to quantify a level of spectrum fragmentation. Approximation methods might be used for estimating average blocking probability and some fragmentation measures, but are so far unable to accurately evaluate the influence of different requested connection bandwidths and do not allow in-depth investigation of blocking events and their relation to fragmentation. This thesis presents the analytical study of the effect of fragmentation on requests’ blocking probability.In this study, new definitions for blocking that differentiate between the reasons for the blocking events were introduced. An analytical framework based on Markov modeling was proposed in order to calculate steady-state probabilities for the different blocking events and to analyze fragmentation related problems in elastic optical links under dynamic traffic conditions. Statistical investigations were derived in order to investigate how different allocation request sizes contribute to fragmentation and blocking probability. This work is complemented with the introduction of a new accommodated fragmentation metric that allows better differentiating between very small variations of spectrum occupancy. Moreover, we show to which extend blocking events, due to insufficient amount of available resources, become inevitable and, comparing to the amount of blocking events due to fragmented spectrum, we draw conclusions on the possible gains one can achieve by system defragmentation. We also show how efficient spectrum allocation policies really are in reducing the part of fragmentation that in particular leads to actual blocking events. Simulation experiments are carried out showing good match with our analytical results for blocking probability in a small scale scenario. Simulated blocking probabilities for the different blocking events are provided for a larger scale node- and network-wise operation scenario in elastic optical networks.

Dissertação apresentada para obtenção do grau de Mestre em Estatística Aplicada e Modelação, na Faculdade de Engenharia da Universidade do Porto, sob a orientação da Professora Doutora Filomana Dias d'Almeida

Tese (doutorado) - Pontifícia Universidade Católica do Paraná, Curitiba, 2012 / Bibliografia: p. 111-119 / As falhas de enlace de fibra óptica podem resultar em grande perda de dados em redes de comunicações ópticas de alta velocidade. A resiliência é de importância crítica ao assegurar elevados níveis de disponibilidade e pode torna-se uma questão importante. / ailures of fiber links can result in major loss of data in high speed optical communication networks. Survivability is of critical importance and assuring high levels of availability becomes an important issue. A typical approach to the design of resilien

Dissertação (mestrado) - Pontifícia Universidade Católica do Paraná, Curitiba, 2009 / Bibliografia: f. 53-57 / Este trabalho avalia diferentes estratégias para melhoria de desempenho de um sistema de reconhecimento de caracteres manuscritos baseado em Modelos Escondidos de Markov (MEM). Os algoritmos de Baum-Welch e Viterbi são utilizados para treinamento e classi / This work evaluates different strategies for improvement of a handwriting character recognition system based on HMMs (Hidden Markov Models). The Baum Welch#s and Viterbi#s algorithms are used for model training and testing. Strategies such as the use of m

Apresentamos o teorema de limite hidrodinâmico para o processo de exclusão simples simétrico com taxa lenta no bordo. Neste processo, partículas descrevem passeios aleatórios independentes no espaço {O, 1, , N}, respeitando a regra de exclusão (que afirma que duas partículas não ocupam o mesmo lugar ao mesmo instante). Paralelamente, partículas podem nascer ou morrer nos sítios O e N com taxas proporcionais a N-1 . Com o devido reescalonamento, a densidade de partículas converge para a solução fraca de urna equação diferencial parcial parabólica. Além disso, no primeiro capítulo, apresentamos seções sobre o Teorema de Prohorov, o espaço das funções càdlàg e a métrica de Skorohod definida nesse espaço. / We present the hydrodynamic limit theorem for the simple symmetric exclusion process with slow driven boundary. In this process, particles describe independent random walks in the space {O, 1, , N}, using the exclusion rule (which says that two particles do not occupy the same place at the same time). We also suppose that particles can be born or die on the sites O and N with rates proportional to N -1 . With the right rescaling procedure, the density of particles converges to the weak solution of a parabolic partial differential equation. In the first chapter, we present sections about Prohorov's Theorem, the càdlàg function space and Skorohod's metric defined in this space.

Apresentamos o teorema de limite hidrodinâmico para o processo de exclusão simples simétrico com taxa lenta no bordo. Neste processo, partículas descrevem passeios aleatórios independentes no espaço {O, 1, , N}, respeitando a regra de exclusão (que afirma que duas partículas não ocupam o mesmo lugar ao mesmo instante). Paralelamente, partículas podem nascer ou morrer nos sítios O e N com taxas proporcionais a N-1 . Com o devido reescalonamento, a densidade de partículas converge para a solução fraca de urna equação diferencial parcial parabólica. Além disso, no primeiro capítulo, apresentamos seções sobre o Teorema de Prohorov, o espaço das funções càdlàg e a métrica de Skorohod definida nesse espaço. / We present the hydrodynamic limit theorem for the simple symmetric exclusion process with slow driven boundary. In this process, particles describe independent random walks in the space {O, 1, , N}, using the exclusion rule (which says that two particles do not occupy the same place at the same time). We also suppose that particles can be born or die on the sites O and N with rates proportional to N -1 . With the right rescaling procedure, the density of particles converges to the weak solution of a parabolic partial differential equation. In the first chapter, we present sections about Prohorov's Theorem, the càdlàg function space and Skorohod's metric defined in this space.