Classification of Numerical Thought Using Electroencephalography

Capoccia, Marco (2024) Classification of Numerical Thought Using Electroencephalography. Bachelor's Thesis, Artificial Intelligence.

Preview	Text Thesis-Final.pdf Download (1MB) | Preview
	Text Toestemming.pdf Restricted to Registered users only Download (183kB)

Abstract

This paper centers on the possibility of classifying numerical thoughts using electroencephalography (EEG) and machine learning. I can prove considerable challenges to the differentiation of EEG patterns related to different numerical thoughts, despite the very promising combination of EEG and artificial intelligence in various cognitive and clinical functions or applications. I applied stringent statistical analysis and machine learning models, including CNN, SVM, KNN and a simple feed forward neural network, always reaching the same conclusion: inability to do better than random guessing. High p-values in all statistical comparisons indicate no significant difference between the means of EEG signals related to different numerical thoughts, therefore implying the limitations of the discriminative power of EEG data and the applied tools for their analysis. The complexity of EEG signals and the need for more sophisticated data acquisition, preprocessing techniques, and modeling approaches in order to enhance sensitivity and specificity in EEG-based cognitive state classification are therefore underlined by the study. These parameters should be optimized in future research to enhance the reliability of decoding numerical cognition from EEG data.

Item Type:	Thesis (Bachelor's Thesis)
Supervisor name:	Vortmann, L.
Degree programme:	Artificial Intelligence
Thesis type:	Bachelor's Thesis
Language:	English
Date Deposited:	12 Nov 2024 09:23
Last Modified:	20 Dec 2024 13:13
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/34344

Actions (login required)

View Item