Ziwei Wang

Ziwei Wang

Ziwei Wang 王紫薇

Ph.D. Candidate at Huazhong University of Science and Technology (HUST), supervised by Prof. Dongrui Wu.

My research focuses on data-efficient, robust, and generalizable EEG decoding under data scarcity and distribution shifts, including cross-subject, cross-dataset, cross-modality, and cross-species settings.

I am supported by the Youth Student Fundamental Research Project from NSFC and the Doctoral Student Program of the Young S&T Talents Cultivation Project from CAST, with total funding of 340,000 RMB (~48,000 USD).

Deep Learning Brain-Computer Interfaces Rehabilitation and healthcare foundation models
Google Scholar GitHub CV Email
Research Focus

EEG decoding under data scarcity and distribution shift, spanning cross-subject, cross-dataset, cross-modality, and cross-species transfer.

Research Highlights

9 first-author papers, including 7 Q1/Top/CAA-A journals such as National Science Review, IEEE JBHI, IEEE TBME, Neural Networks, and Knowledge-Based Systems.

Honors

Ant Group InTech Scholarship, National Scholarship ×3, IEEE CIS Student Grant, Merit Student Pacesetter at HUST, and Huanau Top-10 BCI Awards in China.

Service

Reviewer for IEEE TFS, IEEE JBHI, IEEE TNSRE, KBS, IEEE TBIOM, JNE, Scientific Reports, ICONIP, and IEEE SMC; recognized as an IOP Trusted Reviewer.

Research Vision

My long-term goal is to build reliable EEG foundation models for real-world BCI and smart healthcare.

I am particularly interested in three directions:

  • Data-efficient learning: learning robust EEG representations from limited labeled data.
  • Generalizable modeling: improving transfer across subjects, datasets, modalities, and even species.
  • Knowledge-driven intelligence: integrating neuroscience priors and signal processing knowledge into data generation and decoding models.

Selected Contributions

  • Cross-species EEG transfer for seizure detection: proposed ResizeNet+MSA to transfer knowledge from canine EEG to human seizure detection, enabling cross-species and cross-modality generalization under limited target labels.
  • Efficient EEG decoding architectures: developed DBConformer, a dual-branch convolutional Transformer that improves EEG decoding accuracy with a lightweight design.
  • Knowledge-driven EEG augmentation: introduced Channel Reflection (CR), DWTaug, and HHTaug to improve decoding robustness under limited training data.
  • Representation learning for BCIs: developed CKD, MVCNet, CST, and TASA-SDS for contrastive learning/transfer learning in EEG decoding.

News

  • 06 / 2026 — CKD accepted by IEEE TBME.
  • 04 / 2026 — We have released the CHSZ dataset, an EEG dataset collected from 27 children for epileptic seizure detection, for public download and use. Please refer to our TASA-SDS and CST papers for details of data processing.
  • 03 / 2026 — Our survey on brain signal generation is available on arXiv. Special thanks to Tiki 🐱 for kindly providing her photo for the figures.
  • 02 / 2026 — Selected for the Top 10 Advances in Brain–Computer Interfaces in China (Huanau Award).
  • 12 / 2025 — Supported by the Doctoral Student Program of the Young S&T Talents Cultivation Project from CAST (40,000 RMB).
  • 12 / 2025 — Supported by the Youth Student Fundamental Research Project from NSFC (300,000 RMB).
  • 10 / 2025 — DBConformer accepted by IEEE JBHI.
  • 09 / 2025 — Awarded the Ant Group InTech Scholarship.
  • 07 / 2025 — MVCNet accepted by Knowledge-Based Systems.
  • 03 / 2025 — CST accepted by National Science Review.
  • 02 / 2025 — CSDA accepted by Knowledge-Based Systems.
  • 08 / 2024 — Selected for the Top 10 Highlights in Brain–Computer Interfaces in China (Huanau Award).
  • 04 / 2024 — CR accepted by Neural Networks.

Representative Publications

CST
National Science Review 2025
Canine EEG helps human: Cross-species and cross-modality epileptic seizure detection via multi-space alignment
Z. Wang, S. Li, and D. Wu*
National Science Review, vol. 12, no. 6, p. nwaf086, 2025. CAS Q1 Top
A cross-species and cross-modality transfer framework for seizure detection, showing that canine EEG can help human EEG analysis.
Paper Code Media News
DBConformer
IEEE JBHI 2025
DBConformer: Dual-branch convolutional Transformer for EEG decoding
Z. Wang, H. Wang, T. Jia, X. He, S. Li, and D. Wu*
IEEE Journal of Biomedical and Health Informatics, vol. 30, no. 5, pp. 4134–4147, 2026. CAS Q1 Top
A dual-branch temporal-spatial model for EEG decoding that improves performance while remaining parameter-efficient.
Paper Code
CKD
IEEE TBME 2026
CKD: Contrastive knowledge distillation for cross-dataset EEG classification
Z. Wang, X. He, H. Wang, and D. Wu*
IEEE Trans. on Biomedical Engineering, Early access, 2026. DOI: 10.1109/TBME.2026.3701548. CAA-A
A two-stage contrastive knowledge distillation framework for cross-dataset EEG decoding
Paper Code
Channel Reflection
Neural Networks 2024
Channel reflection: Knowledge-driven data augmentation for EEG-based BCIs
Z. Wang†, S. Li†, J. Luo, J. Liu, and D. Wu*
Neural Networks, vol. 176, p. 106351, 2024. CAS Q1 Top
A knowledge-driven spatial augmentation strategy that leverages EEG channel symmetry to improve decoding accuracy and robustness.
Paper Code Media
CSDA
Knowledge-Based Systems 2025
Time-frequency transform based EEG data augmentation for brain-computer interfaces
Z. Wang, S. Li, X. Chen, and D. Wu*
Knowledge-Based Systems, vol. 311, p. 113074, 2025. CAS Q1 Top
A time-frequency augmentation framework that improves EEG decoding performance under limited training data.
Paper Code
MVCNet
Knowledge-Based Systems 2025
MVCNet: Multi-view contrastive network for motor imagery classification
Z. Wang, S. Li, X. Chen, and D. Wu*
Knowledge-Based Systems, vol. 328, p. 114205, 2025. CAS Q1 Top
A multi-view contrastive learning framework designed to improve representation learning for motor imagery classification.
Paper Code
TASA
Journal of Neural Engineering 2023
Unsupervised domain adaptation for cross-patient seizure classification
Z. Wang, W. Zhang, S. Li, X. Chen, and D. Wu*
Journal of Neural Engineering, vol. 20, no. 6, p. 066002, 2023. CAS Q2 Top
An unsupervised domain adaptation framework for seizure detection across patients under distribution shifts.
Paper Code

View full publication list →

Awards

  • 2025 Ant Group InTech Scholarship (10 awardees worldwide; 2 in Digital Medicine)
  • 2025 National Scholarship (Ph.D.)
  • 2024 National Scholarship (Ph.D.)
  • 2021 National Scholarship (Undergraduate)
  • 2025 Merit Student Pacesetter (highest student honor at HUST)
  • 2022 IEEE Computational Intelligence Society Student Grant (5 awardees worldwide)
  • 2025 Top 10 Advances in Brain–Computer Interfaces in China (Huanau Award)
  • 2024 Top 10 Highlights in Brain–Computer Interfaces in China (Huanau Award)
  • 2023 National First Prize in World Robot Contest
  • 2025 National Second Prize in World Robot Contest
  • 2021 Outstanding Graduate of Hunan Province

Education

  • 09 / 2021 – Present — Ph.D. candidate, Huazhong University of Science and Technology
  • 09 / 2017 – 06 / 2021 — B.Eng. in Measurement and Control Technology and Instrumentation, Central South University

Talks

  • 11 / 2023 — ICONIP Tutorial: Transfer learning for EEG-based brain–computer interfaces
  • 05 / 2025 — CSSC Oral: Cross-species and cross-modality seizure detection via multi-space alignment
  • 09 / 2024 — Alibaba Cloud Yunqi Conference Oral: EEG-based automatic seizure detection
  • 12 / 2024 — China Brain–Computer Intelligence Conference Poster
  • 12 / 2025 — SAAC 2025 Poster: DBConformer
  • 12 / 2024 — SAAC 2024 Poster: CR

Internships

Alibaba Cloud, China
10 / 2022 – 04 / 2023

  • Designed five augmentation operators based on N-grams and TF-IDF for anomaly-aware data augmentation.
  • Proposed a SparseAttention module for long-sequence forecasting.
  • Designed a domain-generalized mixture-of-experts model for robust fault prediction under temporal and device-level shifts.