|
|
Study of influence of toxical and nutritional elements on cell metabolism using combination of Raman spectroscopy and Laser-Induced breakdown spectroscopy
Mazura, Martin ; Hrdlička, Aleš (referee) ; Prochazka, David (advisor)
In this work on-line discrimination of six bacteria strains by means of the Multivariate discrimination analysis (MVDA) is presented. Principal components analysis (PCA) was selected as most suitable technique. Two analytical methods – Laser-Induced breakdown spectroscopy (LIBS) and Raman spectroscopy were equipped for chemical analysis of bacteria strains. In order to obtain the best possible bacteria strains differentiation, the data from both methods was analyzed in two ways separately and together. The data from both measurements was normalized separately and afterwards combined in one data frame for simultaneous analysis. This data frame contained information from both analytical methods. Moreover an influence of cultivation period for each bacteria strain was studied. It was determined that Raman spectroscopy is able to discriminate two bacteria strains and LIBS even four bacteria strains. Using combination of both methods the complete discrimination was achieved. From information of first principal component it was determined that most valuable information in LIBS data is not included in different elemental composition but rather in influence of matrix effect. Solely the LIBS was utilized for studying the effect of cultivation period. It was possible to observe transformation of four bacteria strains within 24 hours. Based on results of this work it is possible to assume that combination of Raman spectroscopy and LIBS, because of the complementary information, is suitable for fast discrimination of different bacteria species and strains. Moreover it was determined that LIBS is able to observe the transformation caused by cultivation period.
|
|
|
Video Retrieval
Černý, Petr ; Mlích, Jozef (referee) ; Chmelař, Petr (advisor)
This thesis summarizes the information retrieval theory, the relational model basic and focuses on the data indexing in relational database systems. The thesis focuses on multimedia data searching. It includes description of automatic multimedia data content extraction and multimedia data indexing. Practical part discusses design and solution implementation for improving query effectivity for multidimensional vector similarity which describes multimedia data. Thesis final part discusses experiments with this solution.
|
|
|
Joint EEG-fMRI analysis based on heuristic model
Janeček, David ; Kremláček, Jan (referee) ; Labounek, René (advisor)
The master thesis deals with the joint EEG-fMRI analysis based on a heuristic model that describes the relationship between changes in blood flow in active brain areas and in the electrical activity of neurons. This work also discusses various methods of extracting of useful information from the EEG and their influence on the final result of joined analysis. There were tested averaging methods of electrodes interest, decomposition by principal components analysis and decomposition by independent component analysis. Methods of averaging and decomposition by PCA give similar results, but information about a stimulus vector can not be extracted. Using ICA decomposition, we are able to obtain information relating to the certain stimulation, but there is the problem in the final interpretation and selection of the right components in a blind search for variability coupled with the experiment. It was found out that although components calculated from the time sequence EEG are independent for each to other, their spectrum shifts are correlated. This spectral dependence was eliminated by PCA / ICA decomposition from vectors of spectrum shifts. For this method, each component brings new information about brain activity. The results of the heuristic approach were compared with the results of the joined analysis based on the relative and absolute power approach from frequency bands of interest. And the similarity between activation maps was founded, especially for the heuristic model and the relative power from the gamma band (20-40Hz).
|
|
|
Biometric fingerprint liveness detection
Váňa, Tomáš ; Vítek, Martin (referee) ; Smital, Lukáš (advisor)
This master‘s thesis deals with biometric fingerprint liveness detection. The theoretical part of the work describes fingerprint recognition biometric systems, fingerprint liveness detection issues and methods for fingerprint liveness detection. The practical part of the work describes proposed set of discriminant features and preprocessing of fingerprint image. Proposed approach using neural network to detect a liveness. The algorithm is tested on LivDet database comprising real and fake images acquired with tree sensors. Classification performance approximately 93% was obtained.
|
|
|
Sleep stage classification based on Hjorth descriptors of EEG signals
Kupková, Kristýna ; Ronzhina, Marina (referee) ; Kozumplík, Jiří (advisor)
This bachelor thesis is focused on the distinction between sleep stages from EEG signals. In its first part the classical method of visual classification of sleep stages is introduced, the second part introduces an automated method for sleep stage scoring. It is a method that uses the three parameters of Hjorth to create a vector space, in which, on the basis of similarity of formed shapes, different stages of sleep could be distinguished. Parameters of Hjorth are calculated from the whole EEG signal, and also from its bands. In the next section of this thesis a principle component analysis is performed. The principle components are placed into a vector space analogously with parameters of Hjorth and the character of formed objects is observed.
|
|
|
Image data processing using principal component analysis (PCA)
Solnický, Jan ; Archalous, Tomáš (referee) ; Rychtárik, Milan (advisor)
This project deals with using of principal component analysis (PCA) in image processing and its aim is introduce mathematical apparatus of principal component analysis and possibility of its using in image processing. Project contains instructions how to compress images with using PCA and also how to convert colour image to grayscale intensity image. There are shown how to use PCA to denoising operation in wavelet spectrum. Project includes results of that operations and their evaluation.
|
|
|
Detection of ischemia in ECG
Tichý, Pavel ; Smital, Lukáš (referee) ; Ronzhina, Marina (advisor)
This paper describes the manifestations of ischemia in the ECG signals and summarizes some methods allowing automatic detection of ischemia. Morphological features were then calculated from ECG signals available from UBMI and statistically evaluated to select features appropriate for further automatic classification. Multilayer feedforward neural network was used for classification of heart beats. The neural network was designed in Matlab. Classification performance up to 99.9% was obtained on available dataset.
|
|
|
Dimensionality reduction of statistical dataset
Sabo, Adam ; Kosová, Petra (referee) ; Hrabec, Pavel (advisor)
This thesis introduces methods which are used to reduce dimensionality and their subsequent application to selected sets of sports statistical data. The first part of the thesis deals with the theoretical apparatus of mathematical statistics, in particular with the Principal Component Analysis and its alternative - the Factor Analysis. The second part provides a brief explanation of the terms related to the selected sets of football statistics where these methods are applied. The third part introduces the results of the application of both methods to statistical files. Data obtained through calculations performed in Python programming language are organized and interpreted by means of graphs and tables.
|
|
|
Video Feature for Classification
Behúň, Kamil ; Herout, Adam (referee) ; Hradiš, Michal (advisor)
This thesis compares hand-designed features with features learned by feature learning methods in video classification. The features learned by Principal Component Analysis whitening, Independent subspace analysis and Sparse Autoencoders were tested in a standard Bag of Visual Word classification paradigm replacing hand-designed features (e.g. SIFT, HOG, HOF). The classification performance was measured on Human Motion DataBase and YouTube Action Data Set. Learned features showed better performance than the hand-desined features. The combination of hand-designed features and learned features by Multiple Kernel Learning method showed even better performance, including cases when hand-designed features and learned features achieved not so good performance separately.
|
|
|
Determination of Factors Affecting Wage Differentiation in the EU Countries
Kocurová, Tamara
This thesis aims to identify the determinants of wage differentiation in the EU countries using panel data analysis. The literature review provides an overview of the theoretical background of the topic, previous studies on wage differences, the current situation of wage differentiation in the European Union, and the potential determinants and consequences of wage differentiation. The analysis reveals that wages are influenced by various potential variables. However, the application of factor analysis led to the identification of significant factors, namely the economic strength of a country, the level of digitization and working conditions, investment, and the unemployment rate. These findings are consistent with the existing literature. The conclusion offers recommendations for increasing wages based on the research results.
|
guest ::
