self-supervised learning approach for inter-subject transfer learning in motor imagery brain-computer interfaces

CATALINA MARÍA GALVÁN; SPIES, RUBEN DANIEL; DIEGO H. MILONE; VICTORIA PETERSON

San Luis

Congreso; Encuentro SAN 2023; 2023

Sociedad Argentina de investigaciones Neurociencia

Reducing calibration time is crucial for enhancing the usability of brain-computerinterfaces based on motor imagery. Due to the high inter-user variability ofelectroencephalography (EEG) signals, a user traditionally has to endure long andtedious calibration sessions to collect enough personalized training data before usingthe system. This need has become even more evident with the advent of deep learningdecoding models, whose performance strongly depends on the volume of data availablefor training. Inter-user transfer learning, where other users’ data is used to train themodel, reduces the required amount of personalized training data. In this context, selfsupervised learning strategies can be used to pretrain the first stages of the model on apretext task and then adapt it to the task of interest through fine-tuning with a few datafrom the target user.Here, we propose a self-supervised learning approach based on a fully convolutionalencoder-decoder network. The reconstruction of EEG segments of a channel is used asthe pretext task. Then an ensemble of the pre-trained encoders per EEG channel,followed by a classification block, conforms the final decoding model. This model isfine-tuned with a small dataset of the target user in the final MI-classification task.