Skip to content
DraconiticResearch
Search Sources API
197,768 studies31 sourcesUpdated 3 min ago
Peer-reviewedCC BYFundamental Research

PAT: Privacy-Preserving Adversarial Transfer for Accurate, Robust and Privacy-Preserving EEG Decoding

  • Xiaoqing Chen1,2,3
  • Tianwang Jia1,2
  • Dongrui Wu1,2
  • Dongrui Wu1,2,3
  1. 1Hubei Key Laboratory of Brain-inspired Intelligent Systems, School of Artificial Intelligence and Automation, Huazhong University of Science and Technology, Wuhan, 430074, China
  2. 2Shenzhen Huazhong University of Science and Technology Research Institute, Shenzhen, 518000, China
  3. 3Zhongguancun Academy, Beijing, 100094, China
Open at sourcePDFDOICite

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: inaccurate decoding, poor robustness of EEG decoding models, and challenges in user privacy protection. While previous studies have addressed one or two of these issues, simultaneously accommodating all three challenges remains challenging and unexplored. This paper demonstrates for the first time that achieving simultaneously accurate, robust and privacy-preserving EEG decoding is feasible by proposing Privacy-preserving Adversarial Transfer (PAT). PAT integrates adversarial training into privacy-preserving transfer learning, achieving simultaneously accurate decoding, adversarial robustness, and privacy protection of EEG-based BCIs. Experiments on five public EEG datasets and three privacy-preserving scenarios (centralized source-free transfer, federated source-free transfer, and transfer with privacy-preserved source data) demonstrated that our proposed approach outperformed over 10 classic and state-of-the-art approaches in both accuracy and robustness. PAT even outperformed state-of-the-art transfer learning approaches that do not consider privacy protection. This is the first time three major challenges in EEG-based BCIs can be handled simultaneously. We hope to promote future research on more accurate, robust, and privacy-preserving EEG decoding.

Related

  1. Federated Spiking GANs with Rényi Differential Privacy for privacy-preserving synthetic EEG generation in brain-computer interfaces.
    Shouvik Paul, Garima Bajwa
    Frontiers in human neurosciencePeer-reviewedCC BYFull textNeuroscienceEurope PMCPMC
  2. Two Kinds of Mental Privacy: What Speech Decoding Does and Does Not Threaten.
    Marcello Ienca
    The Hastings Center reportPeer-reviewedCC BYFull textNeuroscienceEurope PMCPMC
  3. FedEdgeR: federated and privacy-preserving edgeR for differential gene expression analysis
    X. Song, Z. Wei
    bioRxivPreprintCC BYFull textBioinformatics & computational biologybioRxiv
  4. Agricultural visual recognition: a novel privacy-preserving framework based on 4D chaos and Mamba networks
    Hai Yu, Yutao Xia, Yandong Ding +3
    Frontiers in plant sciencePeer-reviewedCC BYFull textPlant biologyEurope PMCCrossref
  5. MediNet: simplifying federated and privacy-preserving AI deployment in healthcare
    Ramon Mateo-Navarro, David Sarrat-González, Dolors Pelegrí-Sisó +3
    BioinformaticsPeer-reviewedCC BY 4.0Full textBioinformatics & computational biologyCrossrefPMC

Cite

BibTeXRIS
BibTeX
@article{chen2026privacypreserving,
  title = {{PAT: Privacy-Preserving Adversarial Transfer for Accurate, Robust and Privacy-Preserving EEG Decoding}},
  author = {Xiaoqing Chen and Tianwang Jia and Dongrui Wu and Dongrui Wu},
  journal = {Fundamental Research},
  year = {2026},
  month = jun,
  doi = {10.1016/j.fmre.2026.04.034},
  url = {https://doi.org/10.1016/j.fmre.2026.04.034},
  note = {CC BY},
}
RIS
TY  - JOUR
TI  - PAT: Privacy-Preserving Adversarial Transfer for Accurate, Robust and Privacy-Preserving EEG Decoding
AU  - Xiaoqing Chen
AU  - Tianwang Jia
AU  - Dongrui Wu
AU  - Dongrui Wu
PY  - 2026
DA  - 2026/06/01
JO  - Fundamental Research
DO  - 10.1016/j.fmre.2026.04.034
UR  - https://doi.org/10.1016/j.fmre.2026.04.034
AB  - 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: inaccurate decoding, poor robustness of EEG decoding models, and challenges in user privacy protection. While previous studies have addressed one or two of these issues, simultaneously accommodating all three challenges remains challenging and unexplored. This paper demonstrates for the first time that achieving simultaneously accurate, robust and privacy-preserving EEG decoding is feasible by proposing Privacy-preserving Adversarial Transfer (PAT). PAT integrates adversarial training into privacy-preserving transfer learning, achieving simultaneously accurate decoding, adversarial robustness, and privacy protection of EEG-based BCIs. Experiments on five public EEG datasets and three privacy-preserving scenarios (centralized source-free transfer, federated source-free transfer, and transfer with privacy-preserved source data) demonstrated that our proposed approach outperformed over 10 classic and state-of-the-art approaches in both accuracy and robustness. PAT even outperformed state-of-the-art transfer learning approaches that do not consider privacy protection. This is the first time three major challenges in EEG-based BCIs can be handled simultaneously. We hope to promote future research on more accurate, robust, and privacy-preserving EEG decoding.
SN  - 2096-9457
SN  - 2667-3258
ER  -