Development of the harmony search algorithm based techniques for removing noise from EEG signals

Electro-encephalogram scanning is a medical imaging technique based on measurement of the electrical activity in human brain by deploying many electrodes to the skull of patient. Inter-neurons electrical activity during the brain functions, provides meaningful information on the health of the braing. Even though EEG scanning is made in appropriate conditions and directed by experts, measured signals are exposed to the external factors or other activities of the body of the patient. Cleaning the signals from the noises are crucial for interpretation. Harmony Search (HS) algorithm is a meta-heuristic algorithm that aims to produce the most appropriate solution by mathematically modeling the musical satisfaction given by the harmony created by the combination of different sounds in the human ear. Over time, many HS variants have been presented to the literature by many researchers in order to better adjust the parameters and change the improvisation mechanism of HS. In this thesis, the performance of HS variants on noise removal from EEG signals which is a big data optimization problem and common benchmark functions is discussed. Besides the existing HS variants, a new variant called Trigonometric Harmony Search (TRI-HS) is developed and its performance is evaluated on dedicated problems with other variants.