This thesis analyzes and assesses changes in the self-similarity of EEG microstate sequences due to acoustic stimuli.
By embedding the microstates into a random walk, an estimate of the Hurst exponent is obtained by means of estimation.
The topic of estimating the Hurst exponent played a significant role in this thesis.
Two methods were introduced for the purpose of estimating the Hurst exponent; one was an established method based on the wavelet transform; the was through the implementation of a convolutional neural network.
Using the two methods, an ANOVA was performed on EEG data recordings of a listening task to assess whether the type of sound was significant.
Ultimately, the results of the analysis proved inconclusive, and further work is needed on the topic.