In this study, an ensemble speaker and speaking environment modeling (ESSEM) approach is proposed to characterize environments in order to enhance performance robustness of automatic speech recognition (ASR) systems under adverse conditions. The ESSEM process comprises two stages, the offline and online phases. In the offline phase, we prepare an ensemble speaker and speaking environment space formed by a collection of super-vectors. Each super-vector consists of the entire set of means from all the Gaussian mixture components of a set of hidden Markov Models that characterizes a particular environment. In the online phase, with the ensemble environment space prepared in the offline phase, we estimate the super-vector for a new testing environment based on a stochastic matching criterion. A series of techniques is proposed to further improve the original ESSEM approach on both offline and online phases. For the offline phase, we focus on methods to enhance the construction and coverage of the environment space. We first demonstrate environment clustering and environment partitioning algorithms to well structure the environment space; then, we propose a discriminative training algorithm to enhance discrimination across environment super-vectors and therefore broaden the coverage of the ensemble environment space. For the online phase, we study methods to increase the efficiency and precision in estimating the target super-vector for the testing condition. To enhance the efficiency, we incorporate dimensionality reduction techniques to reduce the complexity of the original environment space. To improve the precision, we first study different forms of mapping function and propose a weighted N-best information technique; then, we propose cohort selection, environment space adaptation and multiple cluster matching algorithms to facilitate the environment characterization. We evaluate the proposed ESSEM framework on the Aurora-2 connected digit recognition task. Experimental results verify that the original ESSEM approach already provides clear improvement over a baseline system without environment compensation. Moreover, the performance of ESSEM can be further enhanced by using the proposed offline and online algorithms. A significant improvement of 16.08% word error rate reduction is achieved by ESSEM with optimal offline and online configuration over our best baseline system on the Aurora-2 task.

Visualization could be defined as the graphical presentation of information, in which the main aim is to improve the user&rsquo / s perception. In all phases of the disaster management, decision makers come across huge data sets with spatio-temporal content. It is hard to deal with these sets in order to find answers to the main question of &ldquo / How can we decrease the losses due to disasters?&rdquo / , which is at the core of the disaster management concept. To furnish this aim, disaster risk information has to be transparent and clearly stated to the public, decision makers and disaster managers. This might be more sophisticated than the calculation of the risk. Taking precautions before a disaster to reduce the causalities and lossess engendered by natural disasters is relatively cheaper, and more importantly, better than cure. To achieve enhanced preparations for all kinds of disasters, visualization is quite an important tool for decision support and risk communication. The basic aim of this research is to propose a conceptual framework, with the consideration of all stakeholders related to the disaster management issue to have a better risk communication, and to guide the design, implementation and integration of the 3D urban modeling tools into disaster risk visualization. Moreover, an empirical methodology is also developed for the generation of visualization solutions through the design, and employment of the tool for disaster management framework. The proposed framework has three main phases .These are the definition of visualization components, object representation, and needs assessment. A new LoD hierarchy with indoor is proposed to visualize all the possible 3D urban disaster situations in the first phase. Then, a decision rule with eight attributes is proposed in the second phase to establish a link between the hazard type and the LoD needed in a 3D urban model for visualization. This decision rule is applied in a proposed three-level hierarchycal structure. The assessed objects of these three levels are urban, sub-urban zone and building. Moreover, a method to define the needed sub-urban zone is proposed. Finally, different 3D urban modelling methods are analyzed to define the data and process needs of possible 3D urban disaster visualization situations. Two natural hazard cases are studied within the scope of this dissertation to assess the operability of the proposed framework. These implementations involve one earthquake and one tsunami case. Special attention is paid to finding one specific sample for two modelling viewpoints, namely static and dynamic. The first applications of the proposed framework with all the related features prove quite promising.

Submitted by Cláudia Bueno (claudiamoura18@gmail.com) on 2016-05-12T17:45:14Z No. of bitstreams: 2 Dissertação - Danillo Pires Valverde - 2016: 5970913 bytes, checksum: db2323393b3cc67f95df905f6a4600e9 (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) / Approved for entry into archive by Luciana Ferreira (lucgeral@gmail.com) on 2016-05-13T11:02:15Z (GMT) No. of bitstreams: 2 Dissertação - Danillo Pires Valverde - 2016: 5970913 bytes, checksum: db2323393b3cc67f95df905f6a4600e9 (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) / Made available in DSpace on 2016-05-13T11:02:15Z (GMT). No. of bitstreams: 2 Dissertação - Danillo Pires Valverde - 2016: 5970913 bytes, checksum: db2323393b3cc67f95df905f6a4600e9 (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) Previous issue date: 2016-03-01 / Conselho Nacional de Pesquisa e Desenvolvimento Científico e Tecnológico - CNPq / In this work we realize a study of the aminobenzoic acid family (Abz) in neutral, anionic and cationic conformations combining a Sequential QM/MM method with mean field approximation (ASEC) with Free Energy Gradient method (FEG), for the solvents: cyclohexane, acetonitrile and water. In all cases, the Abz molecules type present a conformation in aqueous solution has greater geometry distortion compared to the gas phase. According to our results, there is just one cationic conformation in aqueous solution for ortho. For para and meta, anionic molecules the geometry is more stable planar conformation, which is a saddle point in gas phase. For neutral ortho, the rotamer I is the more energetically stable structure. In anionic conformations it is found the largest number of hydrogen bonds in aqueous solution. In acetonitrile and water, the energy stability of conformations can be explained by hydrogen bonds and solute molecule dipole moment. In all cases, the peak of the first absorption band is in good agreement with experimental results. Our results turn out to be more close to experimental result if explicit solvent molecules are included. The theoretical absorption spectra has similar behaviour to those obtained experimentally and the absorption band shifts are well described with ASEC, mainly those involving the neutral, anionic and cationic conformations in water. / Neste trabalho é realizado o estudo da família de ácidos aminobenzóicos (Abz) nas conformações neutra, aniônica e catiônica combinando o uso do método QM/MM sequencial com a aproximação de um campo eletrostático médio (ASEC) juntamente com o método de gradiente de energia Livre (FEG) para os solventes: ciclohexano, acetonitrila e água. Em todas as moléculas estudadas, a conformação em solução aquosa é a que apresenta maiores mudanças geométricas com respeito a fase gasosa. De acordo com os nossos resultados, existe apenas uma conformação catiônica em solução aquosa para o caso do orto. Nas moléculas aniônicas do p-Abz e do m-Abz a geometria mais estável é a conformação planar, que em fase gasosa corresponde a um ponto de sela. No caso do orto neutro, o rotâmero I é a estrutura mais favorável energeticamente. As conformações aniônicas em geral efetuam mais ligações de hidrogênios LH em meio aquoso. Em acetonitrila e água, a estabilidade energética é bem explicada pelas ligações de hidrogênio e pelo momento de dipolo da molécula do soluto. O pico da 1ª banda de absorção em todas as moléculas estudadas utilizando o ASEC se encontra em bom acordo com o experimental, mas com a inclusão de algumas moléculas explícitas do solvente os resultados obtidos se aproximam ainda mais do valor experimental. Os espectros de absorção teóricos possuem comportamento similar aos obtidos experimentalmente e os deslocamentos da banda de absorção são bem descrito com ASEC, principalmente os que envolvem as conformações neutra, aniônica e catiônica em água.

As is now generally accepted, climate change and environmental degradation has largely been triggered by carbon emissions and energy modeling for policy analysis has therefore attained renewed urgency. It is important for governments to satisfy emission targets and timetables set down by international agreements without disregarding macroeconomic concerns and restrictions. In this study, we present a large-scale nonlinear optimization model that allows the analysis of macroeconomic and multi-sectoral energy policies in respect of technological and environmental options and scenarios. The model consists of a detailed representation of energy activities and disaggregates the rest of the economy into five main sectors. Economy-wide solutions are obtained by computing a utility maximizing aggregate consumption bundle on the part of a representative household. Intersectoral and foreign transaction balances are maintained using a modified accounting matrix. The model also computes the impact on macroeconomic variables of greenhouse gas (GHG) emission strategies and abatement schemes. As such the model is capable of producing solutions that can be used to benchmark regulatory instruments and policies. Several scenarios are presented for the case of Turkey in which the impact of a nuclear power programme and power generation coupled with carbon-capture-and-storage schemes are investigated as well as setting quotas on total and sectoral GHG emissions.