Les antibiotiques β-lactamines sont parmi les médicaments antimicrobiens les plus anciens et les plus utilisés. L'enzyme bactérienne β-lactamase hydrolyse l'antibiotique β-lactame en cassant la structure de base "anneau β-lactame". Pour identifier les nouvelles β-lactamases, une étude complète a été réalisée dans diverses bases de données biologiques telles que Human Microbiome Project, env_nr et NCBI nr. L'analyse a révélé que les séquences ancestrales putatives et les recherches de profil HMM jouaient un rôle important dans l'identification de la base de données homologue et métagénomique à distance dans l'enzyme β-lactamase existante comme matière noire. Les larges analyses phylogénétiques des β-lactamases existantes et nouvellement identifiées représentent les nouveaux clades dans les arbres. En outre, l'activité d'hydrolyse des antibiotiques β-lactamines de séquences nouvellement identifiées (provenant d'archées et d'humains) a été étudiée en laboratoire, ce qui montre l'activité de la β-lactamase. La deuxième phase de l'étude a été entreprise pour examiner l'évolution fonctionnelle des β-lactamases. Premièrement, des séquences de protéines ß-lactamase 1155 ont été extraites de la base de données ARG-ANNOT et des valeurs CMI la littérature correspondante. Les résultats ont révélé que l'activité fonctionnelle de la β-lactamase évoluait de manière convergente au sein de la classe moléculaire. La troisième phase de cette thèse représente le développement d'une base de données intégrative de β-lactamases. La base de données publique actuelle de β-lactamases a des informations limitées, par conséquence, une base de données intégrative a été développée. / The β-lactam antibiotics are one of the oldest and widely used antimicrobial drugs. The bacterial enzyme β-lactamase hydrolyzes the β-lactam antibiotic by breaking the core structure “β-lactam ring”. To identify the novel β-lactamases a comprehensive investigation was performed in different biological databases such as Human Microbiome Project, env_nr, and NCBI nr. The analysis revealed that putative ancestral sequences and HMM profile searches played a significant role in the identification of remote homologous and uncovered the existing β-lactamase enzyme in the metagenomic database as dark-matter. The comprehensive phylogenetic analyses of extant and newly identified β-lactamase represent the novel clades in the trees. Further, the β-lactam antibiotic hydrolysis activity of newly identified sequences (from archaea and human) was investigated in laboratory, which shows β-lactamase activity.The second phase of the investigation was undertaken to examine the functional evolution of β-lactamases. First, 1155 β-lactamase protein sequences were retrieved from ARG-ANNOT database and MIC values from the corresponding literature. The results revealed that the functional activity of β-lactamase evolved convergently within the molecular class.The third phase of this thesis presents development of an integrative β-lactamase database. The existing public database of β-lactamase has limited information, therefore, an integrative database was developed.

No description available.

Tato práce se zabývá syntézou řeči z textu. V práci je podán základní teoretický úvod do syntézy řeči z textu. Práce je postavena na MARY TTS systému, který umožňuje využít existujících modulů k vytvoření vlastního systému pro syntézu řeči z textu, a syntéze řeči pomocí skrytých Markovových modelů natrénovaných na vytvořené řečové databázi. Bylo vytvořeno několik jednoduchých programů ulehčujících vytvoření databáze a přidání nového jazyka a hlasu pro MARY TTS systém bylo demonstrováno. Byl vytvořen a publikován modul a hlas pro Český jazyk. Byl popsán a implementován algoritmus pro přepis grafémů na fonémy.

Large variations in writing styles and difficulties in segmenting cursive words are the main reasons for handwritten cursive words recognition for being such a challenging task. An Indian postal document reading system based on a segmentation-free context based stochastic model is presented. The originality of the work resides on a combination of high-level perceptual features with the low-level pixel information considered by the former model and a pruning strategy in the Viterbi decoding to reduce the recognition time. While the low-level information can be easily extracted from the analyzed form, the discriminative power of such information has some limits as describes the shape with less precision. For that reason, we have considered in the framework of an analytical approach, using an implicit segmentation, the implant of high-level information reduced to a lower level. This enrichment can be perceived as a weight at pixel level, assigning an importance to each analyzed pixel based on their perceptual properties. The challenge is to combine the different type of features considering a certain dependence between them. To reduce the decoding time in the Viterbi search, a cumulative threshold mechanism is proposed in a flat lexicon representation. Instead of using a trie representation where the common prefix parts are shared we propose a threshold mechanism in the flat lexicon where based just on a partial Viterbi analysis, we can prune a model and stop the further processing. The cumulative thresholds are based on matching scores calculated at each letter level, allowing a certain dynamic and elasticity to the model. As we are interested in a complete postal address recognition system, we have also focused our attention on digit recognition, proposing different neural and stochastic solutions. To increase the accuracy and robustness of the classifiers a combination scheme is also proposed. The results obtained on different datasets written on Latin and Bengali scripts have shown the interest of the method and the recognition module developed will be integrated in a generic system for the Indian postal automation.

A proliferação de dispositivos móveis e de redes sem fio tem encorajado um crescente interesse em sistemas e serviços baseados na localização de dispositivos portáteis, especialmente em ambientes internos, como edifícios e residências. A localização de um dispositivo portátil é um parâmetro crítico em aplicações baseadas no contexto, as quais requerem grande precisão na estimativa de localização. Entretanto, projetar e desenvolver sistemas de localização em interiores, com crescente precisão na estimação e decrescente custo de instalação, é um problema desafiador. Uma abordagem bastante interessante para satisfazer os requisitos de baixo custo consiste em utilizar as infra-estruturas existentes de redes locais sem fio (WLAN) no padrão IEEE 802.11, que já estão instaladas em muitos ambientes. A maioria das abordagens para localização usando WLAN propostas na literatura é baseada em técnicas probabilísticas, que apresentam bom desempenho e estão cada vez mais populares. Estas técnicas usam um mapa com a informação da potência recebida do sinal (RSSI) juntamente com a freqüência de presença de sinal coletada de múltiplos pontos de acesso Wi-Fi, em diferentes localizações físicas no ambiente. Porém, a informação sobre freqüência de presença de sinal pode ser muito ruidosa devido à natureza imprevisível das falhas de transmissão, as quais podem ocorrer decorrentes de diversos fatores. Este trabalho propõe um novo sistema de localização Wi-Fi, o WBLS (Wireless Based Location System), que não considera a informação sobre freqüência de presença de sinal no processo de estimação, visando eliminar os ruídos a ela associados. O WBLS explora o fato da potência do sinal Wi-Fi variar com a localização e usa um HMM descrito em um grafo onde os nós representam as localizações e as arestas, as probabilidades de transição em função da topologia do ambiente e das velocidades esperadas de um pedestre portando um dispositivo móvel. Investiga-se em que situações a eliminação da informação sobre freqüência de presença de sinal devido a seus ruídos associados aumenta a exatidão da estimativa de localização, apesar do descarte da informação em si. Os experimentos realizados demonstram que a característica mais importante do WBLS é uma particular robustez ao lidar com desligamentos de pontos de acesso, os quais podem ocorrer sem nenhum aviso ou previsão em um ambiente onde pouco controle se tem sobre sua infra-estrutura. / The proliferation of mobile computing devices and wireless networks has fostered a growing interest in location-based systems and services for Portable Wireless Devices, specially in indoor environments. The location of a handheld device is a critical parameter in context-aware applications, which require high degree of accuracy of location estimation. However, designing and developing indoor location systems with increasing estimation accuracy and decreasing cost installation is a challenging problem. A very interesting approach to reach low-cost requirements consists in using the pre-existing IEEE 802.11 wireless local area network (WLAN) infrastructure that is already installed in many places. Most of the WLAN indoor location approaches proposed in the literature are based on probabilistic techniques which show good performance and are becoming increasingly popular. Such approaches use a map of received signal strength information and signal presence frequency collected from multiple Wi-Fi access points at different physical locations in the environment. However, the signal presence frequency information can be very noisy due to the unpredictable nature of transmissions failures, which can be caused by several factors. This work proposes a new probabilistic-based Wi-Fi location system, WBLS (Wireless Based Location System), which doesn\'t take the signal presence frequency information into account in the estimation process, in an attempt to eliminate its associated noise. WBLS exploits the fact that Wi-Fi signal strength vary with location, and uses an HMM on a graph of location nodes whose transition probabilities are a function of the building\'s floor plan and expected speeds of a pedestrian holding a portable device. We investigate if eliminating signal presence frequency information due to its associated noise increases the accuracy of the location estimation, despite the amount of information about the signal presence that is discarded. Experiments show that the most important feature of WBLS is a particular robustness while dealing with access points shutdowns that may happen without any warning in an environment where there is little control over the infrastructure.

A proliferação de dispositivos móveis e de redes sem fio tem encorajado um crescente interesse em sistemas e serviços baseados na localização de dispositivos portáteis, especialmente em ambientes internos, como edifícios e residências. A localização de um dispositivo portátil é um parâmetro crítico em aplicações baseadas no contexto, as quais requerem grande precisão na estimativa de localização. Entretanto, projetar e desenvolver sistemas de localização em interiores, com crescente precisão na estimação e decrescente custo de instalação, é um problema desafiador. Uma abordagem bastante interessante para satisfazer os requisitos de baixo custo consiste em utilizar as infra-estruturas existentes de redes locais sem fio (WLAN) no padrão IEEE 802.11, que já estão instaladas em muitos ambientes. A maioria das abordagens para localização usando WLAN propostas na literatura é baseada em técnicas probabilísticas, que apresentam bom desempenho e estão cada vez mais populares. Estas técnicas usam um mapa com a informação da potência recebida do sinal (RSSI) juntamente com a freqüência de presença de sinal coletada de múltiplos pontos de acesso Wi-Fi, em diferentes localizações físicas no ambiente. Porém, a informação sobre freqüência de presença de sinal pode ser muito ruidosa devido à natureza imprevisível das falhas de transmissão, as quais podem ocorrer decorrentes de diversos fatores. Este trabalho propõe um novo sistema de localização Wi-Fi, o WBLS (Wireless Based Location System), que não considera a informação sobre freqüência de presença de sinal no processo de estimação, visando eliminar os ruídos a ela associados. O WBLS explora o fato da potência do sinal Wi-Fi variar com a localização e usa um HMM descrito em um grafo onde os nós representam as localizações e as arestas, as probabilidades de transição em função da topologia do ambiente e das velocidades esperadas de um pedestre portando um dispositivo móvel. Investiga-se em que situações a eliminação da informação sobre freqüência de presença de sinal devido a seus ruídos associados aumenta a exatidão da estimativa de localização, apesar do descarte da informação em si. Os experimentos realizados demonstram que a característica mais importante do WBLS é uma particular robustez ao lidar com desligamentos de pontos de acesso, os quais podem ocorrer sem nenhum aviso ou previsão em um ambiente onde pouco controle se tem sobre sua infra-estrutura. / The proliferation of mobile computing devices and wireless networks has fostered a growing interest in location-based systems and services for Portable Wireless Devices, specially in indoor environments. The location of a handheld device is a critical parameter in context-aware applications, which require high degree of accuracy of location estimation. However, designing and developing indoor location systems with increasing estimation accuracy and decreasing cost installation is a challenging problem. A very interesting approach to reach low-cost requirements consists in using the pre-existing IEEE 802.11 wireless local area network (WLAN) infrastructure that is already installed in many places. Most of the WLAN indoor location approaches proposed in the literature are based on probabilistic techniques which show good performance and are becoming increasingly popular. Such approaches use a map of received signal strength information and signal presence frequency collected from multiple Wi-Fi access points at different physical locations in the environment. However, the signal presence frequency information can be very noisy due to the unpredictable nature of transmissions failures, which can be caused by several factors. This work proposes a new probabilistic-based Wi-Fi location system, WBLS (Wireless Based Location System), which doesn\'t take the signal presence frequency information into account in the estimation process, in an attempt to eliminate its associated noise. WBLS exploits the fact that Wi-Fi signal strength vary with location, and uses an HMM on a graph of location nodes whose transition probabilities are a function of the building\'s floor plan and expected speeds of a pedestrian holding a portable device. We investigate if eliminating signal presence frequency information due to its associated noise increases the accuracy of the location estimation, despite the amount of information about the signal presence that is discarded. Experiments show that the most important feature of WBLS is a particular robustness while dealing with access points shutdowns that may happen without any warning in an environment where there is little control over the infrastructure.

Acoustic analysis of animal vocalizations has been widely used to identify the presence of individual species, classify vocalizations, identify individuals, and determine gender. In this work automatic identification of speaker and gender of mice from ultrasonic vocalizations and speaker identification of meerkats from their Close calls is investigated. Feature extraction was implemented using Greenwood Function Cepstral Coefficients (GFCC), designed exclusively for extracting features from animal vocalizations. Mice ultrasonic vocalizations were analyzed using Gaussian Mixture Models (GMM) which yielded an accuracy of 78.3% for speaker identification and 93.2% for gender identification. Meerkat speaker identification with Close calls was implemented using Gaussian Mixture Models (GMM) and Hidden Markov Models (HMM), with an accuracy of 90.8% and 94.4% respectively. The results obtained shows these methods indicate the presence of gender and identity information in vocalizations and support the possibility of robust gender identification and individual identification using bioacoustic data sets.

This thesis describes a new optimisation and new heuristics for automatically marking up XML documents, and CEM, a Java implementation, using PPM models. CEM is significantly more general than previous systems, marking up large numbers of hierarchical tags, using n-gram models for large n and a variety of escape methods. Four corpora are discussed, including the bibliography corpus of 14682 bibliographies laid out in seven standard styles using the BibTeX system and marked up in XML with every field from the original BibTeX. Other corpora include the ROCLING Chinese text segmentation corpus, the Computists' Communique corpus and the Reuters' corpus. A detailed examination is presented of the methods of evaluating mark up algorithms, including computation complexity measures and correctness measures from the fields of information retrieval, string processing, machine learning and information theory. A new taxonomy of markup complexities is established and the properties of each taxon are examined in relation to the complexity of marked up documents. The performance of the new heuristics and optimisation are examined using the four corpora.

<p>This dissertation addresses the challenges of giving computers the ability of doing face recognition, i.e. discriminate between different faces. Face recognition systems are commonly trained with a database of face images, becoming “familiar” with the given faces. Many reported methods rely heavily on training database size and representativenes. But collecting training images covering, for instance, a wide range of viewpoints, different expressions and illumination conditions is difficult and costly. Moreover, there may be only one face image per person at low image resolution or quality. In these situations, face recognition techniques usually suffer serious performance drop. Here we present effective algorithms that deal with single image per person database, despite issues with illumination, face expression and pose variation.</p><p>Illumination changes the appearance of a face in images. Thus, we use a new pyramid based fusion method for face recognition under arbitrary unknown lighting. This extended approach with logarithmic transform works efficiently with a single image. The produced image has better contrast at both low and high ranges, i.e. has more visible details than the original one. An improved method works with high dynamic range images, useful for outdoor face images.</p><p>Face expressions also modify the images’ appearance. An extended Hidden Markov Models (HMM) with a flexible encoding scheme treats images as an ensemble of horizontal and vertical strips. Each person is modeled by Joint Multiple Hidden Markov Models (JM-HMMs). This approach offers computational advantages and the good learning ability from just a single sample per class. A fast method simulated JM-HMM functionality is then derived. The new method with abstract observations and a simplified similarity measurement does not require retraining HMMs for new images or subjects. Pose invariant recognition from a single sample image per person was overcome by using the wire frame Candide face model for the synthesis of virtual views. This is one of the support functions of our face recognition system, WAWO. The extensive experiments clearly show that WAWO outperforms the state-of-the-art systems in FERET tests.</p>

This dissertation addresses the challenges of giving computers the ability of doing face recognition, i.e. discriminate between different faces. Face recognition systems are commonly trained with a database of face images, becoming “familiar” with the given faces. Many reported methods rely heavily on training database size and representativenes. But collecting training images covering, for instance, a wide range of viewpoints, different expressions and illumination conditions is difficult and costly. Moreover, there may be only one face image per person at low image resolution or quality. In these situations, face recognition techniques usually suffer serious performance drop. Here we present effective algorithms that deal with single image per person database, despite issues with illumination, face expression and pose variation. Illumination changes the appearance of a face in images. Thus, we use a new pyramid based fusion method for face recognition under arbitrary unknown lighting. This extended approach with logarithmic transform works efficiently with a single image. The produced image has better contrast at both low and high ranges, i.e. has more visible details than the original one. An improved method works with high dynamic range images, useful for outdoor face images. Face expressions also modify the images’ appearance. An extended Hidden Markov Models (HMM) with a flexible encoding scheme treats images as an ensemble of horizontal and vertical strips. Each person is modeled by Joint Multiple Hidden Markov Models (JM-HMMs). This approach offers computational advantages and the good learning ability from just a single sample per class. A fast method simulated JM-HMM functionality is then derived. The new method with abstract observations and a simplified similarity measurement does not require retraining HMMs for new images or subjects. Pose invariant recognition from a single sample image per person was overcome by using the wire frame Candide face model for the synthesis of virtual views. This is one of the support functions of our face recognition system, WAWO. The extensive experiments clearly show that WAWO outperforms the state-of-the-art systems in FERET tests.