Machine Learning-Based Comparison of EEG Signals During Associative Equivalence Learning with Visual Stimuli of Varying Complexity

Előnézet	PDF (disszertáció) Available under License Creative Commons Attribution Non-commercial. Download (5MB) | Előnézet
Előnézet	PDF (tézisfüzet) Available under License Creative Commons Attribution Non-commercial. Download (85kB) | Előnézet
Előnézet	PDF (tézisfüzet) Available under License Creative Commons Attribution Non-commercial. Download (79kB) | Előnézet
Előnézet	PDF (melléklet) Download (232kB) | Előnézet
Előnézet	PDF (kapcsolódó saját publikációk) Download (3MB) | Előnézet

Absztrakt (kivonat) idegen nyelven

Associative learning is a fundamental cognitive process that enables individuals to form connections between stimuli and responses. This thesis investigates the neural mechanisms underlying associative learning using electroencephalography (EEG) and machine learning techniques. The study compares two tasks: the original Rutgers Acquired Equivalence Test (RAET) and a simplified, feature-reduced version called Polygon. By analyzing cortical activity, the research aims to understand how stimulus complexity and verbalizability influence learning processes. EEG signals were recorded from participants performing both tasks and preprocessed using Independent Component Analysis (ICA) to isolate neural components. Machine learning classifiers, including Long Short-Term Memory (LSTM) networks and Support Vector Classification (SVC), were employed to detect differences in brain activity between the two tasks. The analysis focused on distinct brain regions, such as the frontal, temporal, occipital, and parietal lobes, to assess their roles in learning, attention, and decision-making. Key findings reveal that the frontal region shows the most significant differences between the RAET and Polygon tasks. These differences are linked to varying levels of attention, memory load, and decision-making processes influenced by stimulus complexity. In contrast, the parietal region exhibited minimal variation, suggesting its role is less affected by stimulus features. Although limited in sample size, this study demonstrates the effectiveness of combining EEG with machine learning to uncover cortical activity patterns in associative learning. The findings provide a foundation for future research and highlight the potential of these methods to advance our understanding of cognitive processes.

Tétel nézet