|
Genetical algorithms and their use in optimization
Krtek, Jiří ; Dupač, Václav (referee) ; Antoch, Jaromír (advisor)
In the present work we deal with a branch of stochastic optimization algorithms, so called genetic algorithms. In the first chapter we can find description of a run of the genetic algorithm and the main operations which route searching of a feasible solution set, i.e. crossover and mutation. There is not absent a simple example, whereon reader can make sense of the presented operations. There is a short chapter devoted to theory of genetic algorithms which follows section describing various improvements of the basic algorithm, e.g. the Gray code. A real optimization problem is introduced in the third and also the last chapter. We have solved it using the theory of Markov decision processes for modeling a queuing system and by using genetic algorithms for finding optimum. We have also looked for optimum via a specialized algorithm. Both approaches are compared in the end of this chapter. All calculations have been implemented in the Fortran language.
|
|
Cluster analysis methods and their applications in marketing
Dvořák, Marek ; Prášková, Zuzana (referee) ; Vaněček, Pavel (advisor)
In this work we study algorithms for cluster analysis and their application to the real data. In the beginning, the various types of data are presented. We define dissimilarity measures for each type of data and for clusters to be able to do the clustering and evaluate the separation quantitatively. In the Chapter 2, there are described partitioning algorithms and some criteria to determine the optimal number of clusters. A part of this chapter is devoted to the fuzzy cluster analysis which is a generalization of partitioning techniques. Hierarchical algorithms are characterized in Chapter 3 as well as criteria for choosing the appropriate method. In the very end of this chapter, there is a comparison of all the methods in terms of various types of the separation functionals. Archetypal analysis, which is another data mining instrument, is described in Chapter 4. All chapters include illustration examples of usage. The main application part is the last chapter of this diploma thesis and it's based on the lifestyle survey in the Czech republic.
|
|
Bootstrap methods for dependent observations
Petrásek, Jakub ; Kaňková, Vlasta (referee) ; Prášková, Zuzana (advisor)
This Diploma thesis deals with principles, asymptotic properties and comparison of bootstrap methods for dependent observations. In the first chapter principal ideas and benefits of bootstrap method for independent data are introduced. Subsequently, these knowledge are applied to data exhibiting dependency. Block, frequency and sieve bootstrap methods are presented. Afterwards, principle of each method is described in broader context, asymptotic properties are presented and some of them are derived. Strong dependency of block bootstrap method on block length is discussed and algorithms for empirical choice of optimal block length are described. The main aim of this work is to compare discussed methods from theoretical point of view and via simulation study. Eventually, a few examples, which are based on real data sets, are presented. Discussed principles are implemented in software R and software Fortran.
|
|
Vision based localization of mobile robots
Wolf, Petr ; Holan, Tomáš (referee) ; Mráz, František (advisor)
A key feature of a mobile robotic system is the ability to determine its exact position. Solutions using digital cameras to accomplish this task play a major role in the current research. This thesis studies the topic of vision-based localization, proposes a solution and veriffi es its characteristics in series of experiments. An annealed particle filter is used to track line segments of known 3D models. The selected method of evaluation based on control points uses just the necessary part of image data and allows e cient implementation. Real-time performance, resilience to erratic motion and partial occlusion has been achieved, allowing the system to perform motion tracking over large viewpoint changes. A mapping feature is proposed as a possible extension and implemented by combining a particle fi lter with robust optimization.
|
|
The impact and comparison of imaging methods in prenatal diagnosis of congenital defects
Frisová, Veronika ; Roček, Miloslav (advisor) ; Eliáš, Pavel (referee) ; Vaněčková, Manuela (referee)
Traditional two-dimensional (2D) ultrasound has been and remains the mainstay of prenatal screening and diagnosis of congenital structural abnormalities. However, real clinical impact of the newer innovative methods (3D/4D ultrasound and MRI) remains to be unclear. The aim of this study is to evaluate the impact of combined examination by ultrasound and fetal magnetic resonance on prenatal diagnosis of congenital defects and optimization of perinatal management. Both imaging methods were in our department performed on the same day by a single operator specialized in fetal medicine. Therefore, another aim of this study was to compare the importance of 2D and 3D/4D ultrasound and magnetic resonance for exactness of the prenatal diagnosis of congenital defects. 212 pregnancies were examined. 89% of patients were referred for abnormal or suspicious finding at local ultrasound examination and 11% for an increased risk of congenital defects due to an abnormal family history or an abnormal result of previous laboratory testing. Detailed postnatal diagnosis was successfully retrieved in 113 (53%) of referred pregnancies. Only this group of pregnancies with known detailed postnatal diagnosis was used for the evaluation of the impact of combined imaging by ultrasound and magnetic resonance on prenatal diagnosis of...
|
|
Generating of Random Samples with Given Properties and Application to Banking
Voronin, Alexander ; Franěk, Petr (advisor) ; Šmíd, Martin (referee)
The work concerns the searching for the algorithm for generating of the random variables with the given properties. There are made analyses of comparisons of the algorithms, and the optimal algorithm was chosen based on it. Since we focus on generating of random variables of defaults and explanatory variables of defaults, it concentrates mainly on the conservation of the dependence of these variables. Further we are looking for the optimal sample size of the generated samples under conservation of the required properties. And in the last Chapter we have applied the surveyed techniques to the real data.
|
|
High-fast electrophoretic determinations of clinically important compounds in body fluids.
Málková, Klára ; Kvasnicová, Vladimíra (referee) ; Bosáková, Zuzana (advisor)
A capillary electrophoretic procedure employing contactless conductivity detection (C4 D) has been developed for direct determination of the glycerol and mannitol polyalcohols in biological and pharmacological samples. Both glycerol and mannitol are fully separated from the sample matrix within very short times of 3.0 and 3.9 min., respectively, when using the optimized background electrolyte, 60 mM H3BO3 + 30 mM LiOH (pH 9.1). The limits of detection amount to 0.5 µM for glycerol and 0.3 µM for mannitol. The repeatability of the glycerol determination in real biological materials is characterized by the coefficient of variation values, 0.5 % and 3.2 %, for the migration time and the peak area, respectively. The procedure has been used to monitor the free glycerol concentration in adipose tissue microdialyzates. A physiological study has demonstrated that the lipolysis occurring during a sporting action can be stimulated by local application of adrenaline. The procedure has further been utilized to determine mannitol in a pharmacological preparation.
|
|
Portfolio investment for individual investors : (portfolio recommendations for three case studies)
Žigraiová, Diana ; Baxa, Jaromír (referee) ; Pečená, Magda (advisor)
The thesis focuses on the portfolio investment area with respect to individual investors. It discusses their investment possibilities and behavioural aspects that may be the cause of deviations in investors' behaviour from rationality and which as well have the impact on forming their investment objectives. On the three investor case studies two qualitatitive methods of asset allocation are studied, eventually dividing the content of their investment portfolios between stocks and bonds. Additionally, the extension to the traditional stock and bond allocation is performed by means of real estate, commodities and art and antiques and its appropriateness is analyzed for each case study investor. At the very end of the thesis a quantitative mean-variance optimization method of asset allocation is mentioned.
|
|
Technical Analysis of the Futures Contract E-mini Russell 2000
Palidrab, Maroš ; Veselá, Jitka (advisor) ; Málek, Jiří (referee)
The goal of my thesis is to describe futures commodity contract e-mini Russell 2000 and the application of automatic trading system on this contract. Developed trading strategy is tested with technical analysis indicators using Money management for optimization. The conclusion of the thesis answers the question if modeled trading system is profitable and practically used for real trading.
|
|
Analytical evaluation of the pharmaceutical formulations using chromatographic methods II
Sinkoviczová, Magdalena ; Mokrý, Milan (advisor) ; Sochor, Jaroslav (referee)
Analytical drug evaluation by chromatographical methods II. ( Determination of ambroxol in pharmaceutical preparation ) Diploma paper Magdalena Sinkoviczová Charles University in Prague, Pharmaceutical Faculty in Hradec Králové, Department of pharmaceutical chemistry and drug control In this diploma paper is described elaboration of one possible method of HPLC analysis of ambroxol and methylparaben ( and its degradation product, p- hydroxybenzoic acid ) in pharmaceutical preparations. Optimal results are provided by analysis made on column Separon SGX C 18 from Tessek, using mobile phase in composition acetonitril : phosphate buffer ( KH2PO4 0,08 mol/l; pH 5,8 ) 40:60 v/v, with flow rate of 0,5 ml/min. Detection was performed at 254 nm by UV detector. For verification of linearity was used a method of calibration curve, constructed from results from analysis of five solutions with various concentration of components. Precision was verified by analysis of six solutions simulating real composition of pharmaceutical preparation. Validation was performed using propylparaben as internal standard. This metod was used for quantification of MUCOSIN® components. Amount of ambroxol was determined on 7,5642 mg/ml which corresponds to 100,86 % of declaration amount ( 7,5 mg/ml ). The amount of methylparaben...
|
guest ::
