Machine learning for signal processing

The aim of this work is to analyze the artificial neural network (ANN), which may help identifying patients with medical illnesses examining peaks of the electroencephalograms (EEG). Definition of the signal is presented in the first section. During this work the main attention is focused on the theory of EEG signals and their peaks, which are described in section two. In section three definition of machine learning is presented along with the theory and performance of the ANN. In this work Python programming language was used for implementing the ANN. Also trial version of the mathematical symbolic computation program Mathematica 11 was used for testing purposes. Both these programs and a brief introduction to the Wolfram programming language and used functions are introduced in section four. The last section holds the practical tests which were done using the programmed ANN. These tests were done with real preprocessed EEG data. Two parameters were used to train the neural network. It was the upslope and downslope of EEG data peaks. There were two diagnoses: patients with Rolandic epilepsy and patients with brain damage which were analyzed. Conclusions and test results are presented in the last section