Domain Adaptation for Super Resolution in Consumer EEG Products
Name
opsahl-sopsahl-meng-eecs-2026-thesis.pdf
Size
6.23 MB
Format
Adobe PDF
Checksum (MD5)
dc263637bf6ca2817bb0c8d64c3d6108
Author(s)
Opsahl, Simon D.
Advisor(s)
Monardo, Vince
Ahmed, Subutai
Date Issued
February 2026
Publisher
Massachusetts Institute of Technology
Abstract
Data from consumer EEG devices often suffers from poor spatial resolution, which can limit accuracy on practical applications of the signal. In this thesis, we evaluate whether augmenting spatial resolution can be separated from the downstream classification. We find that general spatial super-resolution is not a viable method of data augmentation because of differences between tasks, paradigms and devices on diversity and importance of spatial information. We develop a formulaic method of scoring channels that is tied to downstream classification accuracies. Ultimately, we claim that super-resolution and other data-augmentation strategies should never be decoupled from the task to which they will be applied.
MIT Department
Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science
Terms of Use
In Copyright - Educational Use Permitted
Copyright retained by author(s)
Persistent DSpace Link
