Personalized Continual EEG Decoding: Retaining and Transferring Knowledge

TL;DR

This work tackles the problem of large inter-subject variability and catastrophic forgetting in continual MI-EEG decoding under privacy constraints that restrict large centralized pre-training. It introduces the Personalized Continual EEG Decoding (PCED) framework, which combines Euclidean Alignment for fast domain adaptation, an exemplar replay mechanism to preserve previously learned knowledge, and reservoir sampling to manage memory in a subject-incremental setting. The method is formalized for sequential subject learning and validated on the OpenBMI benchmark with 54 subjects, showing superior accuracy and reduced forgetting across multiple backbones compared to baseline approaches. PCED demonstrates a practical, scalable pathway toward personalized, privacy-preserving BCIs that can continually adapt to new users without full retraining or extensive memory overhead.

Abstract

The significant inter-subject variability in electroen-cephalogram (EEG) signals often results in substantial changes to neural network weights as data distributions shift. This variability frequently causes catastrophic forgetting in continual EEG decoding tasks, where previously acquired knowledge is overwritten as new subjects are introduced. While retraining the entire dataset for each new subject can mitigate forgetting, this approach imposes significant computational costs, rendering it impractical for real-world applications. Therefore, an ideal brain-computer interface (BCI) model should incrementally learn new information without requiring complete retraining, thereby reducing computational overhead. Existing EEG decoding meth-ods typically rely on large, centralized source-domain datasets for pre-training to improve model generalization. However, in practical scenarios, data availability is often constrained by privacy concerns. Furthermore, these methods are susceptible to catastrophic forgetting in continual EEG decoding tasks, significantly limiting their utility in long-term learning scenarios. To address these issues, we propose the Personalized Continual EEG Decoding (PCED) framework for continual EEG decoding. The framework uses Euclidean Alignment for fast domain adap-tation, reducing inter-subject variability. To retain knowledge and prevent forgetting, it includes an exemplar replay mechanism that preserves key information from past tasks. A reservoir sampling-based memory management strategy optimizes exemplar storage to handle memory constraints in long-term learning. Experiments on the OpenBMI dataset with 54 subjects show that PCED balances knowledge retention and classification performance, providing an efficient solution for real-world BCI applications.

Figures (2)

• Figure 1: Our personalized incremental learning framework for continual EEG decoding has three key components: (1) EA enables fast adaptation by reducing domain shifts between subjects; (2) memory replay reinforces previous knowledge by replaying stored samples alongside new aligned samples during MI-EEG decoding, preventing forgetting; (3) a reservoir sampling-based memory management strategy maintains a small, dynamic buffer of prior knowledge, using probability scores to efficiently select new samples when memory capacity is exceeded.
• Figure 2: Testing accuracy of different methods at each incremental learning stage for the initial subject in continual EEG decoding.