A master thesis entitled “A study and comparison of the performance and effectiveness of data mining algorithms on biosensor signals” was discussed by the researcher Nagham Ahmed Abboud at the Faculty of Mechanical and Electrical Engineering,
under the supervision of Prof. Majd Al-Deen Al-Ali.
The judging panel comprised the following professors: Prof. Abdullah Ghandour, Prof. Yuser Sayyed Suleiman Al-Atassi, and Prof. Majd Al-Deen Al-Ali
Data mining techniques were applied to cardiac biosensor signals, starting from processing the data and extracting features related to the physiological state of the participants, up to building deep learning models for detecting psychological stress conditions.
The researcher used Multi-Layer Perceptron (MLP) and attention-mechanism in neural networks to classify stress states based on features extracted from biosensor signals.
The models were evaluated using a range of metrics, including accuracy, recall, precision, and F1, with results obtained using deep learning techniques compared to those obtained using traditional algorithms such as decision trees and random forests. ROC curves and confusion matrices were also used to analyze performance on the test data.
The study demonstrates the significant effectiveness of data mining techniques in improving the detection of psychological stress using biosensor signals.
In addition, it provides scientific evidence of the effectiveness of data mining and neural network techniques in improving the performance of psychological stress detection using biosensor signals
.
At the end of the discussion, the researcher was awarded a master’s degree with a very good grade.
