Search

A novel feature extraction technique is presented in this paper. The term H-rankgram is coined here due to the similarity in concept with other feature extraction methods like spectrogram. The H-rankgram is two-dimensional feature pattern which shows the change in algebraic complexity (measured using ranks of Hankel matrices) of a given signal at a different scale in time (estimated using phase space reconstruction with different time lags). In general feature extraction techniques could be classified into two broad types: time domain and frequency domain. The proposed technique fits into the first one. The proof of concept for the technique to detect changes in the signal was explained and an effect of additive noise was tested. Application of the technique was demonstrated to classify ECG signals for healthy and ill patients. The results shows that Myocardial infarction is detected most accurately although there is high imbalance in classification accuracy between the classes.

Stronger oscillations in the local Earth magnetic field may have an impact on the course of ischemic heart disease. This effect is individual for every person and depends on the sex, age, living territory, season, capability to adjust to magnetic field fluctuations and health status. Individuals who already have cardiovascular disease are more sensitive to magnetic field fluctuations, therefore further research is needed to make conclusions. Aim: To identify correlations between changes in local Earth magnetic field frequencies and patients with ST segment elevation myocardial infarction (STEMI) and non-ST segment elevation myocardial infarction (NSTEMI) cases per week, cases by sex per week and patients’ blood laboratory parameters (low-density lipoprotein cholesterol and Troponin I level). Methods: A retrospective study of 1667 patients, who were admitted to the Hospital of Lithuanian University of Health Sciences Kaunas Clinics between 1st January and 31st December 2019 due to acute coronary syndrome (STEMI and NSTEMI), were included in the study. Local Earth's time varying magnetic field (TVMF) was measured by a magnetometer located in Baisogala, Lithuania. Data from the magnetometer was collected daily, and weekly averaged. We assessed the correlations between average weekly geomagnetic field strength in six different frequency ranges (Hz) and average number of STEMI and NSTEMI cases per week, cases by sex per week and patients’ blood laboratory parameters (low-density lipoprotein cholesterol and Troponin I level). Results: The presence of a stronger magnetic field in the frequency range of 32-65 Hz was significantly related to the number of cases in the STEMI group during the winter season (r= 0.583, p= 0.036). Tendency towards a positive correlation was found during the winter in the men's group with STEMI in the high frequency range of 32-65 Hz. Low-density lipoprotein cholesterol level correlated positively in the winter STEMI group with the magnetic field in the 32-65 Hz range, and in the autumn STEMI group with the magnetic field in the frequency range of 0-7 Hz. Conclusion: Increased geomagnetic field strength in the high frequency range of 32-65 Hz is associated with a higher number of cases in the STEMI group.

Deep learning-based methods, especially convolutional neural networks, have been developed to automatically process the images of concrete surfaces for crack identification tasks. Although deep learning-based methods claim very high accuracy, they often ignore the complexity of the image collection process. Real-world images are often impacted by complex illumination conditions, shadows, the randomness of crack shapes and sizes, blemishes, and concrete spall. Published literature and available shadow databases are oriented towards images taken in laboratory conditions. In this paper, we explore the complexity of image classification for concrete crack detection in the presence of demanding illumination conditions. Challenges associated with the application of deep learning-based methods for detecting concrete cracks in the presence of shadows are elaborated on in this paper. Novel shadow augmentation techniques are developed to increase the accuracy of automatic detection of concrete cracks.

The urban economists have stressed the importance of various amenities for the attractiveness of urban areas for residents and businesses and built cultural heritage can be considered as one of such amenities, the benefits of which should not be overlooked. This research was aimed to analyze the influence of heritage aspect including the heritage status or features of the building and the historic built environment in general on the real estate prices and development in Kaunas using hedonic price method. Two sets of data were collected for the analysis - general, including heritage buildings and including new construction since 2013. The research has demonstrated that heritage status and the year of construction (as older buildings can be considered having heritage features) have no significant positive influence on the real estate prices. Meanwhile, the location, heritage context and the architectural distinctiveness of new architecture have a direct influence on the real estate prices. The heritage context correlates with architectural quality of new construction as well. This reveals the benefits of heritage context both for the real estate developers and households; however, the study shows the unemployed social-economic potential of historic buildings as generators and maintainers of heritage context.

Hankel matrices are broadly applied to problems such as system identification, the exploration of the onset of chaos in discrete nonlinear dynamical systems, sensitivity to initial conditions in nonlinear systems. The concept of the H-rank was developed in the thesis by analyzing nonlinear dynamical systems, forecasting short time series. The development of the theory of H-ranks is the main value of this work as H-rank based algorithms enable to reveal and/or solve a set of new system identification, forecasting and control problems., Hankelio matricos yra plačiai taikomos sistemų identifikacijos, pereinamųjų procesų evoliucijos į chaotinius režimus diskrečiose netiesinėse sistemose, netiesinių sistemų jautrumo pradinėms sąlygoms uždaviniuose. Darbe plėtojama H-rango koncepcija, tiriant netiesines dinamines sistemas, atliekant trumpų laiko eilučių prognozavimą. Darbo pagrindinė vertė yra H-rango teorijos taikymas, kuris leidžia atrasti ir/ar išskirti visą eilę naujų netiesinių sistemų savybių.

In this paper we present the equations for determining the volume of the pool of peat for a set of given three-dimensional coordinates. The discrete approach is discussed and tested for additive noise also. Such an approach can be successfully applied in estimating the energy value of a pool of peat. By evaluating the sets of each adjacent points the local properties of peat can be carried out through the calculations.

Šiame darbe tirtas Markovo grandinių Monte Karlo (MCMC) metodo taikymas dviems stochastinėms sistemoms modeliuoti. Paprasčiausia vienkanalė aptarnavimo sistema buvo sumodeliuota darant prielaidą, kad paraiškų aptarnavimo trukmės pasiskirstę pagal nežinomą arba sudėtingos formos dėsnį. Pasiūlytas universalus MCMC metodui reikiamos tikslinio tankio aproksimacijos konstravimo būdas. Tai buvo panaudota įkainojant europietiškąjį pasirinkimo sandorį, nes klasikiniai akcijų kainų dinamikos modeliai ne visada gali būti taikomi. MCMC metodo papildymas leidžia jį taikyti bet kokiems empiriniams duomenims neturint prielaidų apie jų pasiskirstymą. Metodo taikymai realizuoti C++ programavimo kalba. Sukurta programa gali būti naudojama europietiškiems pasirinkimo sandoriams įkainoti, kai turimi istoriniai akcijų kainų duomenys. Teorinė aptarnavimo sistema taip pat gali būti modeliuojama tarus, kad paraiškų aptarnavimo trukmių pasiskirstymas nežinomas., The application of Markov chain Monte Carlo (MCMC) technique for two stochastic systems was investigated in this paper. A simple single channel system with a queue was modelled considering the service time distribution is of unknown form. Analysis showed that MCMC is possible to apply and gives adequate results. The custom approach of constructing a proposal density for MCMC was proposed and applied to find the value of a European call option as the classical techniques can not be applied in all the cases. This custom approach leads the enhanced MCMC to be applied to any empirical data having no assumptions about its probability density. Both applications were programmed in C++. This useful tool can be used for pricing the European option of a real stock having its historical prices. Modelling a theoretical queuing system with the distribution of the service time of an unknown form is also implemented. A MATLAB approach for modelling the dynamics of a stock prices with enhanced MCMC can be found in appendix 2 and 3.