Reconstructed EEG windows from BCI Competition IV Dataset 1 using EEGReXferNet

This dataset provides reconstructed EEG windows from BCI Competition IV Dataset 1, created using EEGReXferNet, a lightweight AI framework for EEG subspace reconstruction via cross-subject transfer learning and channel-aware embedding. The repository for the framework is available on GitHub: https://github.com/ShanSarkar75/EEGReXferNet/ Dataset Structure: data/ └── BCICIV/ ├── ds1a/ │ └── Recon/c0/ # Reconstructed EEG windows (.npz) ├── ds1b/ │ └── Recon/c0/ ├── ds1f/ │ └── Recon/c0/ └── ds1g/ └── Recon/c0/ Each subject-specific folder contains reconstructed EEG windows in .npz format.Files are organized by subject, model type (A, B, C, D), and window type:Clean windowsNoisy windows (> ±3.5 σ)Note: Due to GitHub file size limitations, reconstructed EEG windows are shared via FigShare. Users are encouraged to download the relevant files before performing model/data analysis or validation. Citation: If you use this data in your research, please cite: @article {Sarkar2025EEGReXferNet, author = {Shantanu Sarkar and Piotr Nabrzyski and Saurabh Prasad and Jose Luis Contreras-Vidal}, title = {‘EEGReXferNet’ – A Lightweight Gen-AI Framework for EEG Subspace Reconstruction via Cross-Subject Transfer Learning and Channel-Aware Embedding}, year = {2025}, note = {Preprint},} Acknowledgment: This work uses data from BCI Competition IV Dataset 1 [1]. References: [1] Blankertz, B., Dornhege, G., Krauledat, M., Müller, K.-R., & Curio, G. (2007). The non-invasive Berlin Brain-Computer Interface: Fast acquisition of effective performance in untrained subjects. NeuroImage, 37(2), 539–550. DOI: https://doi.org/10.1016/j.neuroimage.2007.01.051