Active Semi-supervised Transfer Learning (ASTL) for Offline BCI Calibration
This work addresses the calibration burden in BCI systems for users by enabling the use of unlabeled auxiliary data, offering an incremental improvement over existing transfer learning methods.
The paper tackles the problem of reducing subject-specific calibration data in brain-computer interfaces (BCI) by proposing Active Semi-supervised Transfer Learning (ASTL), which integrates active learning, semi-supervised learning, and transfer learning to handle unlabeled auxiliary data. It demonstrates that ASTL achieves consistently good performance across subjects and EEG headsets, outperforming state-of-the-art approaches.
Single-trial classification of event-related potentials in electroencephalogram (EEG) signals is a very important paradigm of brain-computer interface (BCI). Because of individual differences, usually some subject-specific calibration data are required to tailor the classifier for each subject. Transfer learning has been extensively used to reduce such calibration data requirement, by making use of auxiliary data from similar/relevant subjects/tasks. However, all previous research assumes that all auxiliary data have been labeled. This paper considers a more general scenario, in which part of the auxiliary data could be unlabeled. We propose active semi-supervised transfer learning (ASTL) for offline BCI calibration, which integrates active learning, semi-supervised learning, and transfer learning. Using a visual evoked potential oddball task and three different EEG headsets, we demonstrate that ASTL can achieve consistently good performance across subjects and headsets, and it outperforms some state-of-the-art approaches in the literature.