Search

Face recognition has become an integral part of modern security processes. This paper introduces an optimization approach for the quantile interval method (QIM), a promising classifier learning technique used in face recognition to create face templates and improve recognition accuracy. Our research offers a three-fold contribution to the field. Firstly, (i) we strengthened the evidence that QIM outperforms other contemporary template creation approaches. For this reason, we investigate seven template creation methods, which include four cluster description-based methods and three estimation-based methods. Further, (ii) we extended testing; we use a nearly four times larger database compared to the previous study, which includes a new set, and we report the recognition performance on this extended database. Additionally, we distinguish between open- and closed-set identification. Thirdly, (iii) we perform an evaluation of the cluster estimation-based method (specifically QIM) with an in-depth analysis of its parameter setup in order to make its implementation feasible. We provide instructions and recommendations for the correct parameter setup. Our research confirms that QIM’s application in template creation improves recognition performance. In the case of automatic application and optimization of QIM parameters, improvement recognition is about 4–10% depending on the dataset. In the case of a too general dataset, QIM also provides an improvement, but the incorporation of QIM into an automated algorithm is not possible, since QIM, in this case, requires manual setting of optimal parameters. This research contributes to the advancement of secure and accurate face recognition systems, paving the way for its adoption in various security applications.

The paper deals with the identification of time-frequency regions describing cyclicality of bank loans before, during and after the 2008 crisis via wavelets. We bring new methods and findings about the short and medium cycles of loans provided to corporates and households in the Euro Area in 2000–2017 using seasonally unadjusted monthly data. We have recognized an impact of the crisis on data volatility which further influences the type of significance testing of wavelet spectrograms. To avoid this influence we propose: (1) an adaptive spectrogram testing based on Torrence and Compo approach and (2) robustness analysis via enhanced spectrogram modelling tested by the MC simulations. Both cross-checked approaches prove the sensitivity of standard wavelet tests on data volatility. The results confirm the usability of the new approaches and show that the crisis in 2008 influenced the cyclical behaviour of both categories of economic sectors, but in a different way.

We investigate the impact of the financial crisis on the access of small and medium-sized enterprises in the Czech Republic to external financing. We apply the non-parametric kernel density estimation on a firm-level measure of financing constraints and evaluate its distribution on a balanced panel of SMEs. We focus on financing constraints related to financial health of companies since they determine the commercial banks’ lending behaviour. Our results reveal that firms were more constrained during the crisis and their financing constraints did not largely improve after the end of financial crisis. We argue that enterprises were financially constrained during the crisis because of reduced cash-flow and cash holdings.

We provide the wavelet analysis of the economic cycle synchronization during the recent financial crisis. However, the global financial crisis caused economic cycles in most European countries to become more strongly synchronized without increasing of the real convergence process. Our contribution is an application of the singular value decomposition to identify and remove the long-term trend including outliers appearing in the year 2007-2010. We found that the historically greater integration provides more highly synchronized cycles in the core Euro area member countries.

The problem of boundary effects for nonparametric kernel regression is considered. We will follow the problem of bandwidth selection for Gasser-Müller estimator especially. There are two ways to avoid the difficulties caused by boundary effects in this work. The first one is to assume the circular design. This idea is effective for smooth periodic regression functions mainly. The second presented method is reflection method for kernel of the second order. The reflection method has an influence on the estimate outside edge points. The method of penalizing functions is used as a bandwidth selector. This work compares both techniques in a simulation study.

The presented paper deals with the identification of cyclical behaviour of business cycle from the time and frequency domain perspective. Herewith, methods for obtaining the growth business cycle are investigated - the first order difference, the unobserved component models, the regression curves and filtration using the Baxter-King, Christiano-Fitzgerald and Hodrick-Prescott filter. In the case of the time domain, the analysis identification of cycle lengths is based on the dating process of the growth business cycle. Thus, the right and left variant of the naive techniques and the Bry-Boschan algorithm are applied. In the case of the frequency domain, the analysis of the cyclical structure trough spectrum estimate via the periodogram and the autoregressive process are suggested. Results from both domain approaches are compared. On their bases, recommendations for the cyclical structure identification of the growth business cycle of the Czech Republic are formulated. In the time domain analysis, the evaluation of the unity results of detrending techniques from the identification turning point points of view is attached. The analyses are done on the quarterly data of the GDP, the total industry excluding construction, the gross capital formation in 1996-2008 and on the final consumption expenditure in 1995-2008.

The authors focus on financial stability in the Eurozone and propose alternative indicator for macro-prudential policy implementation. Rapid growth of the shadow banking system biases increase in loans to identify financial instability. The velocity of money represented money demand in the economy. The empirical part of the paper applies wavelet analysis to identify cyclical co-movements in money velocity and liquidity supply. The authors provide recommendations for policymakers to stabilize global liquidity shocks and reduce procyclical effects by institutions which supply official liquidity in their domestic currency – central banks.

The purpose of our paper is to define rules for decision of existence spurious synchronization of countries within the currency area. We devote this new methodological approach from an empirical research based on the variability of a dynamic correlation (correlation in frequency domain). We analyze the dynamic correlation in full range and in the business cycle frequencies as well. We also consider lags in economic activity co-movements. Contrary to the standard approach we show its insufficiency especially in case of time domain instruments. For this goal GDP values in quarters of the four Visegrad countries and the Eurozone in the period 1997/Q1–2011/Q1 are used.

Presented paper focuses on consideration and evaluation of methodical approaches to analyze cyclical structure character of economic activity in transition economy. As a starting point, work in time domain is applied, which is followed in frequency domain approach. Both approaches are viewed from methodical as well as application point of view and their advantage and disadvantage are discussed. Consequently, time-frequency domain approach is added and applied on real data. On the basis of obtained results recommendation is formulated. All discussed methodical approaches are also considered from the perspective of capability to evaluate behaving of business cycle in time of global economic crisis before/after year 2008. The empirical part of the paper deals with data of gross domestic product in the Czech Republic in 1996/Q1–2010/Q2.

The aim of this paper is to evaluate statistical significance of the business cycle periodicity types in the Czech Republic between 1996/Q1–2008/Q4. Cyclical fluctuation representing a growth business cycle is taken as the result of de-trending input values using filtering techniques, namely Hodrick-Prescott filter and Baxter-King filters. Thereafter, identification of periodicity type is viewed from frequency analysis perspective, i.e. using harmonic analysis. Critical values for the 1%, 5% and 10% risk for the test designed by R. A. Fisher are derived and enumerated. Comparing values of periodogram with critical values of the R. A. Fisher test conclusion about statistic significance of periodicity is fi­nis­hed. Growth business cycle is identified on the basis of quarterly values of industrial production in the Czech Republic 1996/Q1–2008/Q4. Obtained results – the lengths of significance periods – are compared with economic studies.

Presented paper is focused on evaluation of cyclical behaviour of industrial production of the Czech Republic in time and frequency domain. Analyses in both domains are done in the context of commonly used detrending techniques necessary for obtaining growth business cycle. In the case of time domain analysis, the length of existing cycles is calculated on the basis of dating method application, i.e. right and left version of naive rules as well as Bry–Boschan algorithm. In the case of frequency domain, estimate of the spectra using periodogram and autoregressive process with optimum lag are used. In the time domain evaluation of the unity of the results of detrending techniques from the turning points identification is done as well. All analysis are done on the data of industrial production in the Czech Republic in 1996/Q1–2008/Q4.

The purpose of the paper is to identify GDP growth cycle of the Czech Republic by means of turning points identification using alternative statistical method – non-parametric kernel estimate. Special type of estimate, convolution Gasser-Müller one, is used. An advantage of this approach is possibility to estimate derivations of unknown function, which is suitable especially in the case of turning points searching.For identification of growth business cycle type results of nonparametric estimate of regression function is used. Obtained residuals, growth component, are considered as growth cycle type and are ana­lo­gi­cal­ly identified. On the basis of nonparametric estimates of growth component derivations (1st and 2nd) turning points (trough and peak) are identified. At the end, comparison of nonparametric statistical approach with economic turning points identification approach, Canova type, is done.

The aim of the paper is consumer behaviour analysis when purchasing selected commodity groups concerning the effect of price, habit, discount and product characteristics. Analysis proceed from the Czech household marketing research, where 726 households were electronically questioned. As mentioned above, selected factors for the analysis were habit, products‘ characteristics, price and discount actions.Primary aim is to measure the correspondence of selected factors influence on consumer behaviour during purchase decision making process of selected commodity groups. Interpretation is based on two-tier evaluation. First level represents commodity groups distinction by the character of goods and subsequent evaluation of goods characteristics correspondence in accordance with each influencing factor. Second one represents behaviour of commodity group in cross-section of selected factors. For consumer behaviour analysis chi-square test was used. Before its application the data set (responses) was divided according to the ten-point scale into three interval’ groups.

The aim of this paper is factors identification of the decreasing natality trend in the Czech Republic between years 1991–2005. This identification is done with respect to the financial situation and living standard of families.The first step, analysis of natality factor – animation natality, is performed. Animation natality is divi­ded according to the mother family state in the time of the birth. Trend of born in marriage and trend out of marriage are described. Following analysis is focused on decreasing component of natality – number of born in marriage.The second step is time series correlation analysis used for identification and evaluation influence of demographic and economic factors on decreasing component of natality. Based on this analysis, in­fluen­cing factors for regression model describing natality are selected.The last step is formulation and estimation of multiple regression model describing causality between natality in marriage and selected factors.

Presented paper deals with comparison of chosen methods used for the business cycle identification. With respect to this aim nonparametric method (kernel smoothing) and Box-Jenkins methodology were used. This comparison is performed by application on economic activity in USA 1960/Q01–2007/Q01. The residuals are tested by Box-Pierce test. Identified trend is discussed with chosen historical events which affect business cycle in the USA.

Presented article is focused on additive index partition exploitation at the valuation of financial efficiency of the building and loan associations sector between years 1999–2005. Orientation of this paper continue to the article “Selected aspects of the financial efficiency of the building and loan associations” published in Acta Univ. agric. et silvic. Mendel.Brun., 2006, LIV, No. 6, 2006, ISSN pp. 97–105. Main idea is analysis and evaluation of financial results of the Czech building and loan associations sectors between 1999–2005. For this drift pyramidal representation of indicator ROAA is used. Also effect between partial indicators and synthetic indicator ROAA using additive index partition is investigated.

Presented article is focused on evaluation efficiency of the building and loan associations sector between years 1999–2005 and on identification of the fundamental factors which influence the efficiency of the sector. From the efficiency indicators, ROAA, rate margin, commission margin, cost margin and risk margin were used. These indicators were rated from the building and loan associations accounting data. The criterion for the valuation of the building and loan associations sector were average values of the efficiency indicators achieved in the CZ bank sector as unit in given period. For evaluation of influence individual partial margin significance on ROAA correlation analysis was used.

Kernel smoothers belong to the most popular nonparametric functional estimates. They provide a simple way of finding structure in data. The idea of the kernel smoothing can be applied to a simple fixed design regression model. This article is focused on kernel smoothing for fixed design regresion model with three types of estimators, the Gasser-Müller estimator, the Nadaraya-Watson estimator and the local linear estimator. At the end of this article figures for ilustration of desribed estimators on simulated and real data sets are shown.

Kernel smoothing provides a simple way for finding structure in data. The idea of the kernel smoothing can be applied to a simple fixed design regression model and a random design regression model. This article is focused on kernel smoothing for fixed design regression model with using special type of estimator, the Gasser-Müller estimator, and on choice of the bandwidth. At the end of this article figures for ilustration described methods on two data sets are shown. The first data set contains simulated values of function sin(2πx), the second contains January average temperatures measured in Basel 1755–1855.

Kernel smoothers belong to the most popular nonparametric functional estimates. They provide a simple way of finding structure in data. Kernel smoothing can be very well applied on the regression model. In the context of kernel estimates of a regression function, the choice of a kernel from the different points o view can be investigated. The main idea of this paper is to present construction of the optimal kernel and edge optimal kernel by means of the Gegenbauer and Legendre polynomial.

Abstract  Kernel smoothers belong to the most popular nonparametric functional estimates used for describing data structure. They can be applied to the fix design regression model as well as to the random design regression model. The main idea of this paper is to present a construction of the optimum kernel and optimum boundary kernel by means of the Gegenbauer and Legendre polynomials. [ABSTRACT FROM AUTHOR]