[MRG] Add EEG classification task#6
Open
tgnassou wants to merge 26 commits into
Open
Conversation
tgnassou
changed the title
Author
|
@tomMoral Ready to be merge in our side |
Contributor
|
It might make sense to do one review from within the EGG subgroup? |
Author
|
Do we need to be additional things ? |
Contributor
I think a rebase on/merge of |
There was a problem hiding this comment.
Pull request overview
This PR introduces an EEG classification track to the TSFM benchmark by adding three EEG datasets and three EEG-focused solvers (a baseline EEGNet classifier and two foundation-model + linear-probe solvers).
Changes:
- Add EEG datasets: BNCI2014-001, Sleep PhysioNet, and HGD (motor imagery).
- Add EEG solvers: EEGNet (braindecode), REVE + linear probe, and CBraMod + linear probe.
- Plumb dataset metadata needed by foundation models (e.g., sampling frequency and channel names).
Reviewed changes
Copilot reviewed 6 out of 6 changed files in this pull request and generated 10 comments.
Show a summary per file
| File | Description |
|---|---|
solvers/reve.py |
Adds REVE foundation-model encoder with resampling + linear probe adapter. |
solvers/eegnet.py |
Adds a braindecode EEGNetv4 training/prediction solver. |
solvers/cbramod.py |
Adds CBraMod foundation-model encoder with resampling/truncation + linear probe adapter. |
datasets/sleepphysionet.py |
Adds Sleep PhysioNet windowed sleep-stage classification dataset loader. |
datasets/hgd.py |
Adds HGD motor-imagery classification dataset loader and preprocessing. |
datasets/bnci2014_001.py |
Adds BNCI2014-001 motor-imagery classification dataset loader. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
+78
to
+82
| # windows_dataset.description contains a 'split' column ('train'/'test') | ||
| # Each window is indexed into its base dataset; we replicate that mapping. | ||
| descriptions = windows_dataset.datasets # list of WindowsDataset | ||
|
|
||
| for ds in descriptions: |
Comment on lines
+11
to
+13
| import numpy as np | ||
| import torch | ||
| from benchopt import BaseSolver |
| } | ||
|
|
||
|
|
||
| class EEGNetAdapter: |
Comment on lines
+3
to
+6
| Wraps the sleep recordings from the Sleep Physionet. | ||
| Each recording is split into a training portion (first 10 %) and a test | ||
| portion. | ||
| Labels are from 0 to 4, corresponding to the sleep stages W, N1, N2, N3, REM. |
|
|
||
|
|
||
| class Dataset(BaseDataset): | ||
| """Sleep classification dataset (TSB-UAD). |
Comment on lines
+85
to
+87
| factor : float | ||
| Factor to multiply the raw signals by (e.g. to convert from V to uV | ||
| train_ratio : float |
| @@ -0,0 +1,126 @@ | |||
| import numpy as np | |||
| y_test = np.concatenate( | ||
| [y_all[i] for i in range(len(y_all)) if i not in ids_train] | ||
| ) | ||
| np.unique(y_train), np.unique(y_test) # sanity check |
Comment on lines
+100
to
+115
| X_t = torch.tensor(X, dtype=torch.float32, device=self._device) | ||
| y_t = torch.tensor(y, dtype=torch.long, device=self._device) | ||
|
|
||
| dataset = torch.utils.data.TensorDataset(X_t, y_t) | ||
| loader = torch.utils.data.DataLoader( | ||
| dataset, batch_size=self.batch_size, shuffle=True | ||
| ) | ||
|
|
||
| self._network.train() | ||
| for _ in range(self.n_epochs): | ||
| for xb, yb in loader: | ||
| self._optimizer.zero_grad() | ||
| logits = self._network(xb) | ||
| loss = self._criterion(logits, yb) | ||
| loss.backward() | ||
| self._optimizer.step() |
Comment on lines
+152
to
+161
| X = self._prepare_inputs(np.asarray(X, dtype=np.float32)) | ||
| X_t = torch.tensor(X, dtype=torch.float32, device=self._device) | ||
|
|
||
| self._network.eval() | ||
| preds = [] | ||
| with torch.no_grad(): | ||
| for i in range(0, len(X_t), self._batch_size): | ||
| logits = self._network(X_t[i:i + self._batch_size]) | ||
| preds.append(logits.argmax(dim=1).cpu().numpy()) | ||
| return np.concatenate(preds) |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
DATASETS:
MODELS: