System identification deals with the estimation of mathematical models from experimental data. As mathematical models are built for specific purposes, ensuring that the estimated model represents the system with sufficient accuracy is a relevant aspect in system identification. Factors affecting the accuracy of the estimated model include the experimental data, the manner in which the estimation method accounts for prior knowledge about the system, and the uncertainties arising when designing the experiment and initializing the search of the estimation method. As the accuracy of the estimated model depends on factors that can be affected by the user, it is of importance to guarantee that the user decisions are optimal. Hence, it is of interest to explore how to optimally perform an experiment in the system, how to account for prior knowledge about the system and how to deal with uncertainties that can potentially degrade the model accuracy. This thesis is divided into three topics. The first contribution concerns an input design framework for the identification of nonlinear dynamical models. The method designs an input as a realization of a stationary Markov process. As the true system description is uncertain, the resulting optimization problem takes the uncertainty on the true value of the parameters into account. The stationary distribution of the Markov process is designed over a prescribed set of marginal cumulative distribution functions associated with stationary processes. By restricting the input alphabet to be a finite set, the parametrization of the feasible set can be done using graph theoretical tools. Based on the graph theoretical framework, the problem formulation turns out to be convex in the decision variables. The method is then illustrated by an application to model estimation of systems with quantized measurements. The second contribution of this thesis is on Bayesian techniques for input design and estimation of dynamical models. In regards of input design, we explore the application of Bayesian optimization methods to input design for identification of nonlinear dynamical models. By imposing a Gaussian process prior over the scalar cost function of the Fisher information matrix, the method iteratively computes the predictive posterior distribution based on samples of the feasible set. To drive the exploration of this set, a user defined acquisition function computes at every iteration the sample for updating the predictive posterior distribution. In this sense, the method tries to explore the feasible space only on those regions where an improvement in the cost function is expected. Regarding the estimation of dynamical models, this thesis discusses a Bayesian framework to account for prior information about the model parameters when estimating linear time-invariant dynamical models. Specifically, we discuss how to encode information about the model complexity by a prior distribution over the Hankel singular values of the model. Given the prior distribution and the likelihood function, the posterior distribution is approximated by the use of a Metropolis-Hastings sampler. Finally, the existence of the posterior distribution and the correctness of the Metropolis-Hastings sampler is analyzed and established. As the last contribution of this thesis, we study the problem of uncertainty in system identification, with special focus in input design. By adopting a risk theoretical perspective, we show how the uncertainty can be handled in the problems arising in input design. In particular, we introduce the notion of coherent measure of risk and its use in the input design formulation to account for the uncertainty on the true system description. The discussion also introduces the conditional value at risk, which is a risk coherent measure accounting for the mean behavior of the cost function on the undesired cases. The use of risk coherent measures is also employed in application oriented input design, where the input is designed to achieve a prescribed performance in the intended model application. / <p>QC 20161216</p>

Digitized by Kansas State University Libraries

As vendas de uma empresa são determinadas pelo resultado da gestão das variáveis de marketing (produto, preço, distribuição, comunicação e força de vendas) e também pela sua capacidade de reação às oscilações das variáveis incontroláveis do ambiente e às ações da concorrência. Para empresas no setor industrial, venda pessoal tem grande importância em função da necessidade dos clientes no desenvolvimento de relacionamentos. Por isso, venda pessoal em marketing industrial torna-se uma quinta variável de marketing. Existem diferentes propostas para o planejamento desta função, muitas vezes de forma não integrada. Por outro lado, o setor de insumos agropecuários possui particularidades que abrem oportunidade de tornar o planejamento de vendas mais completo. O objetivo central é a proposição de etapas para o planejamento e gestão de vendas que consolide a literatura de planos de força de vendas, com contribuições da Nova Economia Institucional, e que contemple as particularidades deste setor. As organizações de vendas de três empresas foram acompanhada com a intenção de identificar estas particularidades e os desafios identificados foram contrastados aos modelos existentes. Propõe-se uma seqüência com 10 etapas: (1) análise de necessidades do comprador e compreensão do papel de vendas na transação (2) análise ambiental e concorrência, (3) definição de quotas (4) organização de vendas: desenho de territórios, especialização e número de vendedores, (5) modo de governança e remuneração, (6) administração de clientes e informações, (7) recrutamento e seleção, (8) treinamento e motivação, (9) administração de conflitos e (10) formas de controle e auditoria / The sales of a company are determined by the management of the controllable marketing variables (product, price, place, and promotion) and by the ability of the firm of reacting to the uncontrollable marketing variables oscillations and competitor’s actions. Mainly for firms in business-to-business, personal selling has great importance due to the needs costumers have for business relationships. Therefore, in industrial marketing personal selling becomes clearly a fifth marketing variable. The sales literature presents several proposals for sales planning and management, covering different aspects and, very frequently, lacking integration. On the other side, the agricultural input industry has particularities making incomplete or superficial the existing sales planning models. This research aims at proposing a framework for sales planning and management considering the agricultural input industry, and integrating sales research advances and contributions from New Institution Economics. The sales organization of three firms were followed and documented and their critical aspects on sales were contrasted to the existing sales literature and sales management models. It is proposed a sequence with 10 steps: (1) client needs analysis and understanding of the role of sales in marketing;(2) environmental and competitor analysis (3) quota definition (4) sales organization: territory alignment, specialization, and sales force size (5) governance structure and compensation (6) clients management and information (7) recruiting and selection (8)training and motivation (9) conflict management (10) control measures and sales audit.

As correntes subliminares determinam propriedades da membrana neuronal, tais como a resistência de entrada (Rin) e a constante de tempo (tm). Nesta tese, estudamos mecanismos pelos quais duas correntes subliminares (corrente ativada por hiperpolarização, Ih, e corrente de sódio persistente, INaP) determinam Rin e tm em dois tipos de neurônio: neurônio fusiforme do núcleo coclear dorsal e célula piramidal da região CA1 do hipocampo. A tese está dividida em três partes: a primeira estuda como a Ih atua concomitantemente com a corrente de potássio retificadora de entrada (IKIR) para manter Rin estacionária entre neurônios fusiformes com heterogeneidade de disparo (silenciosos, sem disparos espontâneos, e ativos, com disparos espontâneos regulares). Na segunda parte, usa-se uma combinação de modelagem computacional com a técnica experimental de dynamic-clamp em neurônios piramidais de fatias hipocampais para mostrar que a criação de uma região de inclinação negativa na curva I/V (condutância de inclinação negativa) pela ativação rápida da INaP é responsável pelo aumento de Rin e tm e pela amplificação e prolongamento dos potenciais pós-sinápticos das células. Finalmente, a terceira parte estabelece o mecanismo pelo qual a INaP e Ih controlam a tm da célula. Para isso, propomos um novo conceito denominado \"condutância de inclinação dinâmica\" que leva em consideração a cinética das correntes e explica os efeitos observados das cinéticas de Ih e INaP sobre tm. Com base nos resultados, prevemos que uma Ih com cinética rápida atenua e encurta os potenciais pós-sinápticos excitatórios muito mais que uma Ih com cinética lenta. / Subthreshold currents determine the neuronal membrane properties, such as the input resistance (Rin) and the membrane time constant (tm). In this thesis, we studied the mechanisms by which two subthreshold currents (the hyperpolarization-activated current, Ih, and the persistent sodium current, INaP) determine Rin and tm in two types of neurons: the fusiform neuron of the dorsal cochlear nucleus and the pyramidal cell of the CA1 region of the hippocampus. The thesis is divided in three parts: the first part studies how Ih acts concomitantly with the inwardly rectifying potassium current (IKIR) to equalize Rin among fusiform neurons with firing heterogeneity (quiet, without spontaneous firing and active, with regular spontaneous firing). In the second part, we used a combination of computational modeling with the experimental technique dynamic-clamp in pyramidal cells of hippocampal slices to show that the creation of a negative slope region in the I/V curve (negative slope conductance) by the fast activation of the INaP is responsible for the increase of Rin and tm, and for the amplification and prolongation of postsynaptic potentials in these cells. Finally, the third part establishes the mechanism whereby INaP and Ih control tm in the cell. For this, we propose a new concept called \"dynamic slope conductance\", which takes into consideration the current kinetics and explains the observed effects of Ih and INaP kinetics on tm. Based on the results, we predict that an Ih current with fast kinetics attenuates and shortens excitatory postsynaptic potentials strongly than an Ih current with slower kinetics.

by Lo I Fan. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1994. / Includes bibliographical references. / Chapter Chapter 1: --- Introduction / Chapter 1.1 --- The Evolution of Fonts --- p.1 / Chapter 1.2 --- Bitmap Fonts --- p.2 / Chapter 1.3 --- Outline Fonts / Chapter 1.3.1 --- Arc and Vector Form --- p.4 / Chapter 1.3.2 --- Spline Form --- p.4 / Chapter 1.3.3 --- Pros and Cons of Outline Fonts --- p.8 / Chapter 1.4 --- Examples of Outline Fonts / Chapter 1.4.1 --- Adobe's PostScript --- p.9 / Chapter 1.4.2 --- Apple's and Microsoft TrueType / Chapter 1.4.2.1 --- Outline Representation --- p.10 / Chapter 1.4.2.2 --- Rasterisation --- p.12 / Chapter 1.4.2.3 --- Hinting --- p.13 / Chapter 1.5 --- Bold Fonts / Chapter 1.5.1 --- Definition of Bold --- p.15 / Chapter 1.5.2 --- Definition of Auto B oldness --- p.16 / Chapter 1.5.3 --- Auto Boldness by Double Printing --- p.17 / Chapter 1.5.4 --- Auto Boldness by Multi-Master Technique --- p.18 / Chapter 1.6 --- Chinese Fonts / Chapter 1.6.1 --- Chinese Character Sets --- p.19 / Chapter 1.6.2 --- The Subtleties of Chinese Fonts Auto Boldness --- p.21 / Chapter 1.7 --- Project Objective --- p.23 / Chapter 1.8 --- Goals --- p.23 / Chapter Chapter 2: --- Main Ideas of Chinese Font Auto Boldness / Chapter 2.1 --- Prototype of Auto Boldness Driver --- p.24 / Chapter 2.2 --- Design Features of the Prototype Auto Boldness Driver --- p.25 / Chapter 2.3 --- Data Structure and Algorithm of Auto Boldness / Chapter 2.3.1 --- Data Structure of TrueType Character Outline --- p.27 / Chapter 2.3.2 --- Algorithm of Auto Boldness --- p.28 / Chapter 2.3.3 --- Algorithm Description --- p.29 / Chapter 2.4 --- Component Font Auto Boldness --- p.35 / Chapter Chapter 3: --- Language of Auto Boldness / Chapter 3.1 --- Enhancements of TrueType Engine to support Auto Boldness --- p.36 / Chapter 3.2 --- Symmetric Bold Instruction --- p.38 / Chapter 3.3 --- Rotate Bold Instruction --- p.47 / Chapter 3.4 --- Asymmetric B old Instruction --- p.50 / Chapter 3.5 --- Comparison of Bold Instructions --- p.54 / Chapter 3.6 --- Serif Accommodation Instruction --- p.55 / Chapter Chapter 4: --- Shape Parsing and Auto Bold Code Generation / Chapter 4.1 --- Compilation Process and Auto Boldness --- p.62 / Chapter 4.2 --- Shape Lexical Analyzer --- p.64 / Chapter 4.3 --- Shape Token Attributes Evaluation / Chapter 4.3.1 --- line Token --- p.66 / Chapter 4.3.2 --- bezier2 Token --- p.67 / Chapter 4.3.3 --- sharp Token --- p.70 / Chapter 4.3.4 --- concave Token --- p.75 / Chapter 4.3.5 --- convex Token --- p.75 / Chapter 4.4 --- Scope of Shape Parsing --- p.76 / Chapter 4.5 --- Shape Parsing Mechanism --- p.77 / Chapter 4.6 --- Model Grammar Rules / Chapter 4.6.1 --- Grammar Rule Format --- p.81 / Chapter 4.6.2 --- Grammar Rule Item --- p.82 / Chapter 4.6.3 --- Grammar Rule Assignment --- p.83 / Chapter 4.6.4 --- Grammar Rule Condition --- p.83 / Chapter 4.7 --- Auto Boldness Code Generation --- p.84 / Chapter 4.8 --- Program Methodology of Prototype Auto Boldness Driver --- p.86 / Chapter Chapter 5: --- Conclusions / Chapter 5.1 --- Work Achieved --- p.87 / Chapter 5.2 --- The Pros and Cons of Auto Boldness Algorithm --- p.88 / Chapter 5.3 --- Bold Quality Assessments --- p.91 / Chapter 5.3 --- Future Directions --- p.93 / References / Appendix One / Appendix Two

by Tong Po-Cheung. / Bibliography: leaves [v-10]-[v-11] / Thesis (M.Ph.)--Chinese University of Hong Kong, 1983

by Wong Tsz Kin. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1993. / Includes bibliographical references (leaves 45-47). / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Characteristics of Chinese Character --- p.2 / Chapter 1.1.1 --- The Nature of Chinese Language --- p.2 / Chapter 1.1.2 --- The Structure of Chinese Characters --- p.3 / Chapter 1.1.3 --- Basic Writing Strokes --- p.3 / Chapter 1.1.4 --- Writing Stroke Sequencing --- p.3 / Chapter 1.1.5 --- Geographic Structure of Components --- p.4 / Chapter 1.2 --- Stroke Distribution of Chinese Characters --- p.5 / Chapter 1.3 --- Radical --- p.5 / Chapter 1.4 --- Overview --- p.6 / Chapter 1.5 --- Objective --- p.10 / Chapter 2 --- Preprocessing --- p.12 / Chapter 2.1 --- Smoothing and Sampling --- p.12 / Chapter 2.2 --- Interpolation --- p.13 / Chapter 2.3 --- Dehooking --- p.13 / Chapter 2.4 --- Normalization --- p.14 / Chapter 2.5 --- Stroke Segmentation --- p.15 / Chapter 3 --- Preclassification --- p.18 / Chapter 3.1 --- Feature Analysis --- p.18 / Chapter 3.2 --- Radical Detection --- p.20 / Chapter 3.3 --- Description of The Preclassification Component --- p.22 / Chapter 3.4 --- Results and Conclusions --- p.23 / Chapter 4 --- The Recognition Stage --- p.25 / Chapter 4.1 --- Introduction --- p.25 / Chapter 4.2 --- Stroke Match Algorithm --- p.26 / Chapter 4.3 --- Relation Match Stage --- p.30 / Chapter 4.3.1 --- Introduction --- p.30 / Chapter 4.4 --- Final Classification --- p.35 / Chapter 5 --- Results and Conclusions --- p.39 / Chapter 5.1 --- Experiment Results --- p.39 / Chapter 5.2 --- Analysis --- p.39 / Chapter 5.3 --- Conclusions

Yeung Lap Kei. / Thesis (M.Sc.)--Chinese University of Hong Kong, 1992. / Includes bibliographical references (leaves [14-15]). / PREFACE / ABSTRACT / CONTENT / Chapter §1. --- INTRODUCTION / Chapter §1.1 --- The Chinese language --- p.1 / Chapter §1.2 --- Chinese information processing system --- p.2 / Chapter §1.3 --- Chinese character recognition --- p.4 / Chapter §1.4 --- Multi-stage tree classifier Vs Single-stage tree classifier in Chinese character recognition --- p.6 / Chapter §1.5 --- Decision Tree / Chapter §1.5.1 --- Basic Terminology of a decision tree --- p.7 / Chapter §1.5.2 --- Structure design of a decision tree --- p.10 / Chapter §1.6 --- Motivation of the project --- p.12 / Chapter §1.7 --- Objects of the project --- p.14 / Chapter §1.8 --- Development environment --- p.14 / Chapter §2. --- APPROACH 1 - UNSUPERVISED LEARNING --- p.15 / Chapter §3. --- APPROACH 2 - SUPERVISED LEARNING / Chapter §3.1 --- Idea --- p.17 / Chapter §3.2 --- The 3 Corner Code --- p.20 / Chapter §3.3 --- Feature Extraction & Selection --- p.22 / Chapter §3.4 --- Decision at Each Node / Chapter §3.4.1 --- Statistical Linear Discriminant Analysis --- p.22 / Chapter §3.4.2 --- Optimization of the Number of Misclassification --- p.24 / Chapter §3.5 --- Implementation / Chapter §3.5.1 --- Training Data --- p.36 / Chapter §3.5.2 --- Clustering with the Use of SAS --- p.38 / Chapter §3.5.3 --- Building the Decision Trees --- p.42 / Chapter §3.5.4 --- Description of the Classifier --- p.45 / Chapter §3.6 --- Experiments and Testing Result / Chapter §3.6.1 --- Performance Parameters being Measured --- p.47 / Chapter §3.6.2 --- Testing by Resubstitution Method --- p.50 / Chapter §3.6.3 --- Noise Model --- p.52 / Chapter §4. --- POSSIBLE IMPROVEMENT --- p.55 / Chapter §5. --- EXPERIMENTAL RESULTS & THE IMPROVED MULTISTAGE CLASSIFIER / Chapter §5.1 --- Experimental Results --- p.59 / Chapter §5.2 --- Conclusion --- p.70 / Chapter §6. --- IMPROVED MULTISTAGE TREE CLASSIFIER / Chapter §6.1 --- The Optimal Multistage Tree Classifier --- p.72 / Chapter §6.2 --- Performance Analysis --- p.73 / Chapter §7. --- FURTHER DISCRIMINATION BY CONTEXT CONSIDERATION / Chapter §7.1 --- Idea --- p.76 / Chapter §7.2 --- Description of Algorithm --- p.78 / Chapter §7.3 --- Performance Analysis --- p.81 / Chapter §8. --- CONCLUSION / Chapter §8.1 --- Advantage of the Classifier --- p.84 / Chapter §8.2 --- Limitation of the Classifier --- p.85 / Chapter §9. --- AREA OF FUTURE RESEARCH AND IMPROVEMENT / Chapter §9.1 --- Detailed Analysis at Each Terminal Node --- p.86 / Chapter §9.2 --- Improving the Noise Filtering Technique --- p.87 / Chapter §9.3 --- The Use of 4 Corner Code --- p.88 / Chapter §9.4 --- Increase in the Dimension of the Feature Space --- p.90 / Chapter §9.5 --- 1-Tree Protocol with Entropy Reduction --- p.91 / Chapter §9.6 --- The Use of Human Intelligence --- p.92 / APPENDICES / Chapter A.1 --- K-MEANS / Chapter A.2 --- Unsupervised Learning Approach / Chapter A.3 --- Other Algorithms (Maximum Distance & ISODATA) / Chapter A.4 --- Possible Improvement / Chapter A.5 --- Theories on Statistical Discriminant Analysis / Chapter A.6 --- Passage used in Testing the Performance of the Classifier with Context Consideration / Chapter A.7 --- A Partial List of Semantically Related Chinese Characters / Chapter A.8 --- An Example of Misclassification Table / Chapter A.9 --- "Listing of the Program ""CHDIS.C""" / REFERENCE

Typescript (photocopy). / Digitized by Kansas Correctional Industries

Typescript (photocopy). / Digitized by Kansas Correctional Industries