META-EEG: Meta-learning-based class-relevant EEG representation learning for zero-calibration brain-computer interfaces

Transfer learning for motor imagery-based brain-computer interfaces (MI-BCIs) struggles with inter-subject variability, hindering its generalization to new users. This paper proposes an advanced implicit transfer learning framework, META-EEG, designed to overcome the challenge arising from inter-subject variability. By incorporating gradient-based meta-learning with an intermittent freezing strategy, META-EEG ensures efficient feature representation learning, providing a robust zero-calibration solution. A comparative analysis reveals that META-EEG significantly outperforms all the baseline methods and competing methods on three different public datasets. Moreover, we demonstrate the efficiency of the proposed model through a neurophysiological and feature-representational analysis. With its robustness and superior performance on challenging datasets, META-EEG provides an effective solution for calibration-free MI-EEG classification, facilitating broader usability. •We propose a meta-learning-based zero-calibration EEG feature learning framework.•We construct meta-tasks robust to unseen subjects in meta-training.•We design intermittent freezing to learn class-relevant EEG features efficiently.•It shows effectiveness in motor imagery EEG classification for new stroke patients.•It provides an effective solution for zero-calibration BCI with broader usability.