<p>In ordinary image coding, the same image quality is obtained in all parts of an image. If it is known that there is only one viewer, and where in the image that viewer is focusing, the quality can be degraded in other parts of the image without incurring any perceptible coding artefacts. This master's thesispresents a coding scheme where an image is segmented into homogeneous regions which are then separately coded, and where knowledge about the user's focus point is used to obtain further data reduction. It is concluded that the coding performance does not quite reach the levels attained when applying focus-based quality degradation to coding schemes not based on segmentation.</p>

<p>In vitro-test för upptäckt och behandling av tumör eller mikrometastaser har de senaste 30 åren gjort stora framsteg tack vare immunokemi och nya framgångsrika cellodlings- tekniker som bättre reproducerar celltillväxt i tre dimensioner (3D) och det omgivande stromat (multicellulär tumörsfäroidodling). TS1-218 scFv (single chain variable fragment) är en monoklonal antikropp som har affinitet till ett protein tillhörande cytoskelettet (cytokeratin). Av TS1-218 har skapats olika varianter (en dimer TS1-218₂ och en mutant HE1-Q) med syftet att öka affinitet och retentionstid på platsen för dess verkan. I det här projektet försökte vi att testa och jämföra egenskaper hos alla 3 joderade antikropparna genom att inkubera odlade Hela Hep 2 tumörcellssfäroider med dessa antikroppar. Alla tre antikroppsvarianter visade god förmåga att penetrera sfäroider och att binda deras epitop i cytokeratin 8. Försöken visade att det fanns affinitetsskillnader mellan TS1-218 monomer, dimer och mutant vilket visade sig som olika inbindningsförmåga till sfäroiderna.</p>

<p>Does the country people live in affect the probability of them experiencing happiness? Can a country variable in an ordinal regression model be affected when microeconomic and macroeconomic factors are added to the model? The possible outcomes are either that the country variable affects less when the additional predictors are added to the model, or that they stay the same. The micro data is collected from the European Social Survey database, the macro data is collected from the World Bank. The country variable becomes less substantial when additional variables are added to the model. The variable with the most influence over expected happiness apart from the country variable is whether the individual often socializes with friends or not. It’s statistically significant that the supervened variables make the country variable less volatile in some cases.</p>

This thesis presents both an analysis and simulations of a passively controlled variable-speed generator system, which can be applied for renewable energy sources such as wind turbines. Parallel connected passive/external resistors and inductors are connected to the slip rings of a wound rotor induction machine to provide an acceptable speed operation range, while maintaining high efficiency of the generator system. Two generators, of 80 kW and 186 kW ratings, have been tested and compared to their simulations and good correlation has been obtained. It is shown by both modelling and by laboratory tests that the steady-state power characteristic is well suited to the application and the efficiency compares well with similar rated machines in which either a fixed speed is required or power electronic converters provide the adjustable speed control. Thus, the tested systems are comparable in energy capture while being lower in costs and being both more robust and more reliable. In consequence, it is a more practical solution than power electronics for remote locations. The dynamic results indicate that the generator is dynamically stable following three types of transient conditions: connections to the grid at non-synchronous speed; sudden decreases and increases in applied torque to simulate wind gusts; cyclic torque changes to demonstrate tower shadow effects. Unreasonable transients and undamped conditions have been neither predicted nor observed. Finally, it is proposed that the external elements could be developed to be linked to the rotor circuit without slip rings enabling a complete passive and brushless system. / Graduation date: 2000

Due to the increasing demand of application-specific and/or multi-functional materials, it is necessary to integrate material design and product design. To support such design integration, this paper proposes a methodology to establish the relationships of both material design variables and product design variables. These variables include the required system performances and/or other evaluation criteria, and the relevant system loadings and attributes, where the attributes include both the product structural attributes and the material properties. This is achieved by modeling the behaviors of the product and those of the used material, and identifying the dependencies between the relevant design variables from the behaviors. The variable relationships can then be used to solve various design problems, such as design evaluation, evaluation and optimization of critical design variables, and so on. A design case study is also conducted to illustrate the proposed methodology and its usefulness. / Singapore-MIT Alliance (SMA)

This thesis is an effort to develop generalized dynamic models for systems with time-varying time delays. Unlike the simple time-delay model characterized by a transportation lag in the case of a fixed time delay, time-varying delays exhibit quite different characteristics, making the development of easy to use models a difficult endeavor. First an algorithm is developed to predict the actual input-output behavior when the input signal is directly fed into a device that characterizes the time-varying delay. Input-output behaviour generated with this algorithm serves as the truth model for subsequent approximate model development. Simulation results for different classes of delay and different inputs were obtained using the truth model. The input functions were limited to steps, ramps and sinusoids. This limited class of inputs and delays defines the scope of this thesis and the results are to be interpreted as such. The methodology adopted to identify the basic underpinnings of models was system identification where input-output data came from the truth model. Models for the aforementioned classes of inputs and delays were then derived using elementary system identification tools. These models were then carefully analyzed to extract trends by changing the delay parameter. A satisfactory trend was observed in the case of linearly varying time delay. A generalized model for the linearly varying time delay with step and polynomial inputs was developed. An attempt was also made at developing a generalized model for sinusoidally time-varying time delays. This study proposes a model for linearly time-varying time delay, whose structure is not surprisingly also dependent on the class of inputs. It is shown that the derived model reduces to the well known model in the case of a fixed delay.

Statisticians are used to model interactions between variables on the basis of observed data. In a lot of emerging fields, like bioinformatics, they are confronted with datasets having thousands of variables, a lot of noise, non-linear dependencies and, only, tens of samples. The detection of functional relationships, when such uncertainty is contained in data, constitutes a major challenge. Our work focuses on variable selection and network inference from datasets having many variables and few samples (high variable-to-sample ratio), such as microarray data. Variable selection is the topic of machine learning whose objective is to select, among a set of input variables, those that lead to the best predictive model. The application of variable selection methods to gene expression data allows, for example, to improve cancer diagnosis and prognosis by identifying a new molecular signature of the disease. Network inference consists in representing the dependencies between the variables of a dataset by a graph. Hence, when applied to microarray data, network inference can reverse-engineer the transcriptional regulatory network of cell in view of discovering new drug targets to cure diseases. In this work, two original tools are proposed MASSIVE (Matrix of Average Sub-Subset Information for Variable Elimination) a new method of feature selection and MRNET (Minimum Redundancy NETwork), a new algorithm of network inference. Both tools rely on the computation of mutual information, an information-theoretic measure of dependency. More precisely, MASSIVE and MRNET use approximations of the mutual information between a subset of variables and a target variable based on combinations of mutual informations between sub-subsets of variables and the target. The used approximations allow to estimate a series of low variate densities instead of one large multivariate density. Low variate densities are well-suited for dealing with high variable-to-sample ratio datasets, since they are rather cheap in terms of computational cost and they do not require a large amount of samples in order to be estimated accurately. Numerous experimental results show the competitiveness of these new approaches. Finally, our thesis has led to a freely available source code of MASSIVE and an open-source R and Bioconductor package of network inference.

Generalizing an idea of I. Vekua [1] who, in order to construct theory of plates and shells, fields of displacements, strains and stresses of threedimensional theory of linear elasticity expands into the orthogonal Fourier-series by Legendre Polynomials with respect to the variable along thickness, and then leaves only first N + 1, N = 0, 1, ..., terms, in the bar model under consideration all above quantities have been expanded into orthogonal double Fourier-series by Legendre Polynomials with respect to the variables along thickness, and width of the bar, and then first (Nsub(3) + 1)(Nsub(2) + 1), Nsub(3), Nsub(2) = 0, 1,..., terms have been left. This case will be called (Nsub(3), Nsub(2)) approximation. Both in general (Nsub(3), Nsub(2)) and in particular (0,0) (1,0) cases of approximation, the question of wellposedness of initial and boundary value problems, existence and uniqueness of solutions have been investigated. The cases when variable cross-section turns into segments of straight line, and points have been also considered. Such bars will be called cusped bars (see also [2]).

When polymer power cables were introduced into mediumvoltage applications in the late 1960s, the degradationphenomenon later referred to aswater treeingwas still unknown. However, this changedwithin a relatively short space of time. Just 5-10 years afterthese first polymer applications, faults began to appear thatcould be attributed to the new phenomenon-water treeing. Sincethen, cable manufacturers have been improving their designsgradually, and cables produced today are assumed to have muchlower susceptibility to water trees. However, water treeing isstill assumed to be the most important degradation process inthese applications. There are huge amounts of polymer cable installations thatare degraded by water trees. Since many of these cables areinstalled in the same grid nets, one cable fault oftengenerates another. If two or more cables fail at the same time,the consequences can be extensive. A non-destructive diagnosticmethod makes it possible to assess cables and take action priorto failures. This thesis presents the results of research work on thedevelopment of a measurement technique and diagnostic methodfor the assessment of medium voltage XLPE cables in the field.The method used is called high voltage dielectric spectroscopy,and involves the measurement of capacitance and loss as afunction of frequency at high voltages. The method is used in ashort frequency range of approximately one decade in thelow-frequency region (normally from 1 down to 0.1 Hz). Thevoltage levels used initially are all below the service-voltagelevel. If the cable is assessed to be free from water trees, aslightly increased voltage can be used. Based on dielectric-spectroscopy measurements, specificwater-tree analysis and breakdown tests, a set of measurementprocedures and criteria that take into account differences incable designs has been proposed for the assessment of mediumvoltage XLPE cables. The characteristic responses of XLPEcables (in both non and water-tree-deteriorated cases) make itpossible to separate the cable response from the influence ofaccessories. Field experience shows thatwater-tree-deteriorated cables with relatively low breakdownstrengths can remain in service for many years. <b>Keywords:</b>dielectric spectroscopy, high voltage,variable frequency, diagnostics, non-destructive diagnostics,cables, medium voltage, polymer, XLPE, water trees, watertreeing

Does the country people live in affect the probability of them experiencing happiness? Can a country variable in an ordinal regression model be affected when microeconomic and macroeconomic factors are added to the model? The possible outcomes are either that the country variable affects less when the additional predictors are added to the model, or that they stay the same. The micro data is collected from the European Social Survey database, the macro data is collected from the World Bank. The country variable becomes less substantial when additional variables are added to the model. The variable with the most influence over expected happiness apart from the country variable is whether the individual often socializes with friends or not. It’s statistically significant that the supervened variables make the country variable less volatile in some cases.