A3E: Aligned and Augmented Adversarial Ensemble for Accurate, Robust and Privacy-Preserving EEG Decoding

TL;DR

EEG-based BCIs struggle with data scarcity and variability, adversarial vulnerability, and privacy concerns. The authors introduce A3E, a framework combining data alignment, data augmentation, adversarial training, and ensemble learning, and integrate it into three privacy-protecting transfer-learning scenarios: centralized source-free transfer learning, federated source-free transfer learning, and source data perturbation. Across three public MI EEG datasets, A3E consistently outperforms more than 10 baselines and even surpasses privacy-unaware transfer approaches, demonstrating simultaneous accuracy, robustness, and privacy protection. This work provides a practical, scalable approach to real-world EEG decoding, enabling robust BCI performance without compromising user privacy.

Abstract

An electroencephalogram (EEG) based brain-computer interface (BCI) enables direct communication between the brain and external devices. However, EEG-based BCIs face at least three major challenges in real-world applications: data scarcity and individual differences, adversarial vulnerability, and data privacy. While previous studies have addressed one or two of these issues, simultaneous accommodation of all three challenges remains challenging and unexplored. This paper fills this gap, by proposing an Aligned and Augmented Adversarial Ensemble (A3E) algorithm and integrating it into three privacy protection scenarios (centralized source-free transfer, federated source-free transfer, and source data perturbation), achieving simultaneously accurate decoding, adversarial robustness, and privacy protection of EEG-based BCIs. Experiments on three public EEG datasets demonstrated that our proposed approach outperformed over 10 classic and state-of-the-art approaches in both accuracy and robustness in all three privacy-preserving scenarios, even outperforming state-of-the-art transfer learning approaches that do not consider privacy protection at all. This is the first time that three major challenges in EEG-based BCIs can be addressed simultaneously, significantly improving the practicalness of EEG decoding in real-world BCIs.

Figures (11)

• Figure 1: Flowchart of a closed-loop EEG-based BCI system.
• Figure 2: Flowchart of a closed-loop EEG-based BCI system using source model or data.
• Figure 3: Illustration of the centralized source-free transfer learning scenario.
• Figure 4: Illustration of the federated source-free transfer learning scenario.
• Figure 5: Illustration of the source data perturbation scenario.