Complete training-data provenance and Braindecode software citation
Browse filesREADME-only metadata audit, 2026-11-09. new task-head scope; Braindecode Zenodo citation. No weights, config, license, or upstream paper citations changed.
README.md
CHANGED
|
@@ -59,3 +59,43 @@ accessions, Hub ids otherwise). `nm`/`on` ids are datasets on [NEMAR](https://ne
|
|
| 59 |
- self-supervised (alignment + masked reconstruction), 58 ch, 256 Hz, 4 s; 90% of each dataset's samples for training, 10% validation; PhysionetMI: all 109 subjects, MI+ME runs; HGD: train and test files of all 14 subjects; TSU: all 35 subjects/40 targets
|
| 60 |
- Sources: Wang et al. NeurIPS 2024 (proceedings PDF 4540d267...-Paper-Conference.pdf) p.6, Table 1 and text; Wang et al. NeurIPS 2024, Sec. 3 (pretraining strategy); https://github.com/BINE022/EEGPT README (Pretrained Models); https://github.com/BINE022/EEGPT/blob/main/datasets/pretrain/prepare_pretrain_dataset.py.
|
| 61 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 59 |
- self-supervised (alignment + masked reconstruction), 58 ch, 256 Hz, 4 s; 90% of each dataset's samples for training, 10% validation; PhysionetMI: all 109 subjects, MI+ME runs; HGD: train and test files of all 14 subjects; TSU: all 35 subjects/40 targets
|
| 60 |
- Sources: Wang et al. NeurIPS 2024 (proceedings PDF 4540d267...-Paper-Conference.pdf) p.6, Table 1 and text; Wang et al. NeurIPS 2024, Sec. 3 (pretraining strategy); https://github.com/BINE022/EEGPT README (Pretrained Models); https://github.com/BINE022/EEGPT/blob/main/datasets/pretrain/prepare_pretrain_dataset.py.
|
| 61 |
|
| 62 |
+
|
| 63 |
+
|
| 64 |
+
## Pretrained components
|
| 65 |
+
|
| 66 |
+
The released EEGPT encoder is pretrained; the task probe/head and the default channel projection are newly initialized in braindecode. Train those components for the downstream task rather than treating this as a pretrained classifier.
|
| 67 |
+
|
| 68 |
+
|
| 69 |
+
## Cite Braindecode
|
| 70 |
+
|
| 71 |
+
If you use this braindecode port in scientific work, cite the software on Zenodo,
|
| 72 |
+
[doi:10.5281/zenodo.17699192](https://doi.org/10.5281/zenodo.17699192),
|
| 73 |
+
and the original model and datasets cited above.
|
| 74 |
+
|
| 75 |
+
```bibtex
|
| 76 |
+
@software{braindecode,
|
| 77 |
+
author = {Aristimunha, Bruno and
|
| 78 |
+
Guetschel, Pierre and
|
| 79 |
+
Wimpff, Martin and
|
| 80 |
+
Gemein, Lukas and
|
| 81 |
+
Rommel, Cedric and
|
| 82 |
+
Banville, Hubert and
|
| 83 |
+
Sliwowski, Maciej and
|
| 84 |
+
Wilson, Daniel and
|
| 85 |
+
Brandt, Simon and
|
| 86 |
+
Gnassounou, Théo and
|
| 87 |
+
Paillard, Joseph and
|
| 88 |
+
{Junqueira Lopes}, Bruna and
|
| 89 |
+
Sedlar, Sara and
|
| 90 |
+
Moreau, Thomas and
|
| 91 |
+
Chevallier, Sylvain and
|
| 92 |
+
Gramfort, Alexandre and
|
| 93 |
+
Schirrmeister, Robin Tibor},
|
| 94 |
+
title = {Braindecode: toolbox for decoding raw electrophysiological brain data
|
| 95 |
+
with deep learning models},
|
| 96 |
+
url = {https://github.com/braindecode/braindecode},
|
| 97 |
+
doi = {10.5281/zenodo.17699192},
|
| 98 |
+
publisher = {Zenodo},
|
| 99 |
+
license = {BSD-3-Clause},
|
| 100 |
+
}
|
| 101 |
+
```
|