FasterEA-FML for EEG: Federated Meta-learning with Faster Euclidean Space Data Alignment
摘要
A major challenge in electroencephalogram-based brain-computer interfaces (EEG-based BCIs) is to manage individual differences in EEG. The paper presents FasterEA, which reduces time by randomly selecting EEG trials to compute the reference matrix quickly. The FasterEA-FML framework combines FasterEA and Federated Meta-Learning to accelerate the alignment process and adaptation speed of subjects. The experimental results demonstrate that FasterEA significantly improves computational speed while maintaining high accuracy comparable to EA. Additionally, FasterEA-FML outperforms FedAvg in processing EEG signals from new subjects.