|
|
Using of neural network for detection of heart rhythm disturbances from ECG data and accelerometer signal
Aleksandrenko, Borys ; Ředina, Richard (referee) ; Bulková, Veronika (advisor)
This bachelor's thesis addresses the issue of detecting heart rhythm disorders from EKG and accelerometer signals using machine learning. First, an analysis of the possibilities for detecting heart rhythm disorders from these signals was conducted through a theoretical review. In the next part, a methodology was proposed for detecting two rhythm disorders: inappropriate sinus tachycardia and chronotropic incompetence. The methodology was further supplemented with adaptive filtering of EKG signals using signals from the accelerometer. In the third part of the thesis, a database of samples was created for training machine learning models proposed in the methodology. The next section included the description and implementation of the models. In the fifth part of the thesis, an application for detecting heart rhythm disorders using the proposed methodology was developed in the Python programming language. Finally, a discussion and evaluation of the results were conducted.
|
|
|
Forex forecasting with Support vector regression and Long short-term memory recurrent neural network
Bodický, Michal ; Šíla, Jan (advisor) ; Krištoufek, Ladislav (referee)
In the last years, the field of machine learning boomed. That led to its numerous forecasting applications on prices of Foreign exchange market. Re- searchers there mostly compare neural networks to linear model baselines. The contribution of this thesis consists of a comprehensive performance com- parison between two promising machine learning methods, Support vector regression (SVR) and Long short-term memory recurrent neural network (LSTM RNN), in the forecasting of six highly traded currency pairs on one minute univariate time series data. First, it analyses methods' performances in the forecasting of one step ahead price while varying input dimensions of these methods. Next, it examines how methods perform in longer forecasts, that enabled by using a recurrent version of SVR. In the first analysis, LSTM RNN outperforms SVR in most of the cases several times. Performance of SVR is robust to varying input while LSTM RNN's performance fluctuates across dimensions. In the second analysis, LSTM RNN beats SVR mostly by order of magnitude. With increasing forecasting horizon, SVR's performance gets worse and LSTM RNN's performance remains stable. 1
|
guest ::
