La gestion de la qualité des services dans les réseaux IP est devenue une opération complexe et coûteuse pour les operateurs. Les approches de gestion actuelles se sont montionnées peu efficaces par l’importante des interventions humaines necéssaires. La réponse à ce problème est clairement la conception de nouvelles approches de gestion autonome des réseaux. Cette thèse présente un ensemble de solutions qui contribuent à la proposition et la réalisation d’une telle approche. La première partie de cette thèse examine l’état de l’art des approches de gestion des réseaux. Dans la deuxième partie, nous proposons une démarche de mesure de la QoE (Quality of Experience) et une extension du modèle CIM (Common Information Model) pour permettre d’intégrer les métriques de cette démarche. Dans la partie suivante, nous proposons une extension de l’architecture des équipements réseaux vers plus d’autonomie de prise de décision. Dans les parties suiventes, nous proposons une architecture de gestion autonome (ANEMA: Autonomic NEtwork Management Architecture) instrumentée par les politiques comportementales, les objectifs et les fonctions d’utilité. En exploitant un ensemble de politiques de différents niveaux d’abstraction, ANEMA permet au réseau IP d'atteindre un certain niveau d'autonomie dans la gestion de la qualité de ses services. / The quality of services management in IP networks became a very complex and expensive operation from the operators. The actual management approaches are mentioned inefficient because they are mainly based on direct human intervention. It is clear that the response to this problem is to specify autonomic network management approach. This thesis contributes to the specification and the realisation of such approach. The first part of this thesis addresses the state of art of network management approaches. In the second part, we propose a QoE (Quality of Experience) measuring method and a CIM (Common Information Model) extension to model the concepts and the metrics of this method. In the third part, we propose an extension of the network equipment architecture towards more autonomic decision making. In the following parts, we propose an Autonomic NEtwork Management Architecture (ANEMA) based on Behavioral policies, Goal and Utility function. By exploiting several policy description forms described in different abstraction levels, ANEMA allows the IP network to achieve a certain autonomy level in quality of services management.

This paper compares two technologies, Public Key Infrastructure (PKI) and Virtual Private Network (VPN). PKI and VPN are two approaches currently in use to resolve the problem of securing data in computer networks. Making this comparison difficult is the lack of available data. Additionally, an organization will make their decision based on circumstances unique to their information security needs. Therefore, this paper will illustrate a method using a utility function and the Analytic Hierarchy Process (AHP) to determine which technology is better under a hypothetical set of circumstances. This paper will explain each technology, establish parameters for a hypothetical comparison, and discuss the capabilities and limitations of both technologies. / Master of Arts

Performance measurement of Non-Profit Organizations (NPOs) is of increasing importance for aid agencies, policy-makers and donors. A widely used benchmark for measuring the efficiency of NPOs is the overhead cost ratio, consisting of the total money spent on administration and fundraising relative to the budget. Donors generally favor a lower overhead cost ratio as it ensures that more Money directly reaches beneficiaries. Unlike fundraising expenses, administrative costs do not contribute to advertising the actions of an NPO even though they account for a significant proportion of overhead cost. Reducing administrative expenses is a logical consequence from a financial viewpoint, but might negatively affect NPOs through the resulting administrative capacities. This phenomenon is known as "Nonprofit Starvation Cycle" This work provides an analytical framework for analyzing NPO decision making concerning administrative costs. The paper provides answers to important research questions on the optimal level of administrative spending, the influencing factors and the effects of available information on NPOs. The research shows that focusing on financial performance measurements can result in reduced utility created for NPOs. Less transparency often leads to increased utility for NPOs, but more transparency can increase NPOs' utility if the information available exceeds a certain threshold. Fluctuating donations are challenging for NPOs' planning and may Impact administrative capacities negatively.

We analyze which utility function would best represent the Brazilian representative investor with a one-month investment horizon who has to allocate his wealth across three main asset classes (bonds, equities, and risk free). To do this, we compute the optimal portfolio weights by considering four different specifications for the utility function: (i) mean-variance, (ii) constant relative risk aversion (expected utility functions), (iii) ambiguity aversion, and (iv) loss aversion (non-expected utility functions). We compare the optimal portfolio weights to the empirical portfolio - computed by considering the market value of all the assets in our sample - using the Mahalanobis distance. Our results indicate that the traditional utility function, the mean-variance utility, should not be used to represent the behavior of the Brazilian investor. All other utilities are statistically equal and could be used to compute optimal portfolios for the Brazilian investor. However, the constant relative risk aversion (CRRA) and the ambiguity aversion functions are only justified for extremely high levels of risk aversion. As the loss averse function showed the lowest Mahalanobis distance, we propose that the Brazilian investor is best represented by a utility function that incorporates aversion to losses, in which the decrease of utility caused by a loss is much greater than the increase caused by a gain of equal magnitude. Moreover, this different impact of gains and losses on the investor\'s utility leads individuals to behave as investors with high risk aversion and justifies the fact that loss-aversion preferences have also been widely used to explain why the high risk premium might be consistent with high levels of risk aversion. / Analisamos qual função utilidade representa melhor o investidor representativo brasileiro que aloca sua riqueza em três principais classes de ativos (títulos, ações e livre de risco) e com um horizonte de investimento de um mês. Para isso, calculamos os pesos ótimos do portfólio considerando quatro especificações diferentes para a função utilidade: (i) média-variância, (ii) aversão relativa ao risco constante (funções utilidade esperadas), (iii) aversão à ambiguidade, (iv) aversão à perdas (funções utilidade não esperadas). Comparamos os pesos do portfólio ótimo com o portfólio empírico - calculado considerando o valor de mercado de todos os ativos em nossa amostra - usando a distância de Mahalanobis. Nossos resultados indicam que a função utilidade tradicional de média-variância não deve ser utilizada para representar o comportamento do investidor brasileiro. Todas as demais especificações de função utilidade são estatisticamente iguais e podem ser utilizadas para computar o portfólio ótimo do investidor brasileiro. Entretanto, as funções CRRA e de aversão à ambiguidade são justificadas apenas com níveis extremamente elevados de aversão ao risco. Como o portfólio ótimo com função utilidade do tipo aversão à perdas apresentou a menor distância de Mahalanobis, propomos que o investidor brasileiro é melhor representado por uma função que incorpora aversão à perdas, em que a diminuição da utilidade causada por uma perda é muito maior do que o aumento causado por um ganho de igual magnitude. Além disso, esse impacto diferente de ganhos e perdas na utilidade do investidor leva os indivíduos a comportar-se como investidores com grande aversão ao risco e justifica o fato de que as preferências de aversão à perdas também foram amplamente utilizadas para explicar por que o prêmio de risco pode ser consistente com altos níveis de aversão ao risco.

Sharpe ratio is extensively used in performance of portfolio. However, it is based on assumption that return follows normal distribution. In other words, when return in asset is not normal distribution, the Sharpe ratio is not meaningful. This research focuses on Generalized Sharpe ratio with different distribution in eight indexes from 2001/12/31 to 2010/12/31. We try to find a suitable levy process to fit our data. Instead of Normal distribution assumption, we use Jump diffusion, Variance Gamma, Normal Inverse Gaussian, Hyperbolic, Generalized Hyperbolic, as our distribution to solve stylized fact like skewness and kurtosis. Compared the difference between standard Sharpe ratio and Generalized Sharpe ratio, we come to these conclusions: first of all, Generalized Hyperbolic is better levy process to fit our eight indexes. Second, Sharpe ratio under GH levy process has low autocorrelation, and it present that modified Sharpe ratio is more elastic. Third, Generalized Sharpe ratio can uncover the strategy that fund manager manipulate Sharpe ratio. At last, Generalized Sharpe ratio have better predict than standard Sharpe ratio. Keywords: Sharpe ratio, Levy process, GH distribution, portfolio, utility function

On-line negotiation and bargaining systems can work effectively on the Internet based on the prerequisite that user utility functions are known while undergoing transactions. However, this prerequisite is hard to meet due to the variety and anonymous nature of Internet surfing. Therefore, how to rapidly and precisely construct a user¡¦s utility function is an essential issue. This research proposes a radial basis function (RBF) network, a neural network, to model a user¡¦s utility function in order to rapidly and precisely model user utility function. We verify the feasibility of the method through experiments, and compare the performance of RBF networks in prediction performance, time expenses, and subjects¡¦ perceptions with the Multiple Regression (MR), SMARTS, and SMARTER methods. The results show that the RBF network method is feasible in these criteria. Not only the RBF network needs less time to construct the users¡¦ utility function than the SMARTS method does, but also it can model user utility functions more precisely than the MR, SMARTS, and SMARTER methods.

We analyze which utility function would best represent the Brazilian representative investor with a one-month investment horizon who has to allocate his wealth across three main asset classes (bonds, equities, and risk free). To do this, we compute the optimal portfolio weights by considering four different specifications for the utility function: (i) mean-variance, (ii) constant relative risk aversion (expected utility functions), (iii) ambiguity aversion, and (iv) loss aversion (non-expected utility functions). We compare the optimal portfolio weights to the empirical portfolio - computed by considering the market value of all the assets in our sample - using the Mahalanobis distance. Our results indicate that the traditional utility function, the mean-variance utility, should not be used to represent the behavior of the Brazilian investor. All other utilities are statistically equal and could be used to compute optimal portfolios for the Brazilian investor. However, the constant relative risk aversion (CRRA) and the ambiguity aversion functions are only justified for extremely high levels of risk aversion. As the loss averse function showed the lowest Mahalanobis distance, we propose that the Brazilian investor is best represented by a utility function that incorporates aversion to losses, in which the decrease of utility caused by a loss is much greater than the increase caused by a gain of equal magnitude. Moreover, this different impact of gains and losses on the investor\'s utility leads individuals to behave as investors with high risk aversion and justifies the fact that loss-aversion preferences have also been widely used to explain why the high risk premium might be consistent with high levels of risk aversion. / Analisamos qual função utilidade representa melhor o investidor representativo brasileiro que aloca sua riqueza em três principais classes de ativos (títulos, ações e livre de risco) e com um horizonte de investimento de um mês. Para isso, calculamos os pesos ótimos do portfólio considerando quatro especificações diferentes para a função utilidade: (i) média-variância, (ii) aversão relativa ao risco constante (funções utilidade esperadas), (iii) aversão à ambiguidade, (iv) aversão à perdas (funções utilidade não esperadas). Comparamos os pesos do portfólio ótimo com o portfólio empírico - calculado considerando o valor de mercado de todos os ativos em nossa amostra - usando a distância de Mahalanobis. Nossos resultados indicam que a função utilidade tradicional de média-variância não deve ser utilizada para representar o comportamento do investidor brasileiro. Todas as demais especificações de função utilidade são estatisticamente iguais e podem ser utilizadas para computar o portfólio ótimo do investidor brasileiro. Entretanto, as funções CRRA e de aversão à ambiguidade são justificadas apenas com níveis extremamente elevados de aversão ao risco. Como o portfólio ótimo com função utilidade do tipo aversão à perdas apresentou a menor distância de Mahalanobis, propomos que o investidor brasileiro é melhor representado por uma função que incorpora aversão à perdas, em que a diminuição da utilidade causada por uma perda é muito maior do que o aumento causado por um ganho de igual magnitude. Além disso, esse impacto diferente de ganhos e perdas na utilidade do investidor leva os indivíduos a comportar-se como investidores com grande aversão ao risco e justifica o fato de que as preferências de aversão à perdas também foram amplamente utilizadas para explicar por que o prêmio de risco pode ser consistente com altos níveis de aversão ao risco.

There had been a number of researches that investigated on the security market without transaction costs. The focus of this research is in the area that when the security market with transaction costs is fair and in such fair market how one chooses a suitable portfolio to optimize the financial goal. The research approach adopted in this thesis includes linear algebra and elementary probability. The thesis provides evidence that we can maximize expected utility function to achieve our goal (maximize expected return under certain risk tolerance). The main conclusions drawn from this study are under certain conditions the security market is arbitrage-free, and we can always find an optimal portfolio maximizing certain expected utility function.

We define a novel algorithm based on utility functions for dynamically allocating tasks to mobile robots in a multi-robot system. The algorithm attempts to maximize the performance of the mobile robot while minimizing inter-robot communications. The algorithm takes into consideration the proximity of the mobile robot to the task, the priority of the task, the capability required by the task, the capabilities of the mobile robot, and the rarity of the capability within the population of mobile robots. We evaluate the proposed algorithm in a simulation study and compare it to alternative approaches, including the contract net protocol, an approach based on the knapsack problem, and random task selection. We find that our algorithm outperforms the alternatives in most metrics measured including percent of tasks complete, distance traveled per completed task, fairness of execution, number of communications, and utility achieved.

of the master thesis Title: Modeling of risk aversion Author: František Navrátil Department: Department of Probability and Mathematical Statistics Supervisor: Doc. RNDr. Petr Lachout, CSc. Abstract: The thesis discusses various theories that are able to model investor's subjective attitude to risk. The goal of the thesis is to clearly recapitulate possible mathematical approaches and to apply them in a real situation. One of the ways to tackle the problem is to use expected utility theory and a specific shape of a utility function. Another way is to choose a suitable risk measure. Especially useful for the modelling of risk aversion is the class of spectral risk measures that enables investor to choose a risk spectrum that meets his perception of risk. The thesis contains basic definitions concerning stochastic programming - a theory essential to solve the related optimization problems. Keywords: Risk aversion, utility function, probability constraint.