Emotion detection using machine learning on EEG data

Vegter, R. (2020) Emotion detection using machine learning on EEG data. Bachelor's Thesis, Artificial Intelligence.

Preview	Text Bachelor_Thesis_RikVegter.pdf Download (2MB) | Preview
	Text Toestemming.pdf Restricted to Registered users only Download (96kB)

Abstract

Classiyfing emotions based on EEG data has grown in popularity in the last few decades. At the moment, literature trains classifiers on data sets created in a laboratory where emotions are artificially evoked, meaning they present a stimulus and expect a certain emotion. The EEG data during the emotion is then measured and analysed. In order to capture the EEG data of emotions that are naturally evoked, a special type of debate was analysed which evokes emotions for the debaters. The goal of this study was to create a classifier that is able to distinguish \textit{happy} EEG data from \textit{angry} EEG data when training the classifiers based on a data set where the emotions are evoked in a more daily life setting. Time-domain features were extracted from the preprocessed data set. A random forest classifier and KNN classifier were trained and returned accuracies of 92.5 \% and 92.2 \% respectively with parameter optimisation when tested on the training data. The confusion matrices of the classifiers both showed true positives and true negatives of above 90 \%. K-fold cross validation showed accuracies of 65.1 \% for the random forest classifier and 61.2 \% for the KNN classifier.

Item Type:	Thesis (Bachelor's Thesis)
Supervisor name:	Vugt, M.K. van and Kaushik, P.
Degree programme:	Artificial Intelligence
Thesis type:	Bachelor's Thesis
Language:	English
Date Deposited:	03 Aug 2020 13:50
Last Modified:	03 Aug 2020 13:50
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/22977

Actions (login required)

View Item