Update Braindecode dependency to 0.8 (#542)

* Update braindecode to 0.8 * Simplify plot_braindecode.py * Update whats_new.rst * Update load_model.py * Remove unused code in examples * Update braindecode pipelines
NeuroTechX · Feb 1, 2024 · fcfae76 · fcfae76
1 parent 1b03827
commit fcfae76
Show file tree

Hide file tree

Showing 8 changed files with 68 additions and 222 deletions.
diff --git a/docs/source/whats_new.rst b/docs/source/whats_new.rst
@@ -21,6 +21,7 @@ Enhancements
 - Adding cache option to the evaluation (:gh:`517` by `Bruno Aristimunha`_)
 - Option to interpolate channel in paradigms' `match_all` method (:gh:`480` by `Gregoire Cattan`_)
 - Adding leave k-Subjects out evaluations (:gh:`470` by `Bruno Aristimunha`_)
+- Update Braindecode dependency to 0.8 (:gh:`542` by `Pierre Guetschel`_)
 
 Bugs
 ~~~~

diff --git a/examples/load_model.py b/examples/load_model.py
@@ -16,13 +16,12 @@
 from braindecode import EEGClassifier
 from braindecode.models import EEGInception
 from scikeras.wrappers import KerasClassifier
-from sklearn.pipeline import Pipeline
+from sklearn.pipeline import Pipeline, make_pipeline
 from skorch.callbacks import EarlyStopping, EpochScoring
 from skorch.dataset import ValidSplit
 
 from moabb import set_log_level
 from moabb.pipelines.features import StandardScaler_Epoch
-from moabb.pipelines.utils_pytorch import BraindecodeDatasetLoader, InputShapeSetterEEG
 from moabb.utils import setup_seed
 
 
@@ -70,9 +69,6 @@
 ###############################################################################
 # Loading the PyTorch model
 
-# Set EEG Inception model
-model = EEGInception(in_channels=22, n_classes=2)
-
 # Hyperparameter
 LEARNING_RATE = 0.0001
 WEIGHT_DECAY = 0
@@ -84,8 +80,7 @@
 
 # Define a Skorch classifier
 clf = EEGClassifier(
-    module=model,
-    criterion=torch.nn.CrossEntropyLoss,
+    module=EEGInception,
     optimizer=torch.optim.Adam,
     optimizer__lr=LEARNING_RATE,
     batch_size=BATCH_SIZE,
@@ -99,9 +94,6 @@
         EpochScoring(
             scoring="accuracy", on_train=False, name="valid_acc", lower_is_better=False
         ),
-        InputShapeSetterEEG(
-            params_list=["in_channels", "input_window_samples", "n_classes"],
-        ),
     ],
     verbose=VERBOSE,  # Not printing the results for each epoch
 )
@@ -114,8 +106,5 @@
 
 clf.load_params(f_params=f_params, f_optimizer=f_optimizer, f_history=f_history)
 
-# Create the dataset
-create_dataset = BraindecodeDatasetLoader(drop_last_window=False)
-
 # Create the pipelines
-pipes_pytorch = Pipeline([("Braindecode_dataset", create_dataset), ("EEGInception", clf)])
+pipes_pytorch = make_pipeline(clf)
diff --git a/examples/pipelines_braindecode/braindecode_EEGInception.py b/examples/pipelines_braindecode/braindecode_EEGInception.py
@@ -6,7 +6,6 @@
 from skorch.dataset import ValidSplit
 
 from moabb.pipelines.features import Resampler_Epoch
-from moabb.pipelines.utils_pytorch import BraindecodeDatasetLoader, InputShapeSetterEEG
 
 
 # Set up GPU if it is there
@@ -22,16 +21,9 @@
 EPOCH = 10
 PATIENCE = 3
 
-# Create the dataset
-create_dataset = BraindecodeDatasetLoader()
-
-# Set random Model
-model = EEGInception(in_channels=1, n_classes=2, input_window_samples=100)
-
 # Define a Skorch classifier
 clf = EEGClassifier(
-    module=model,
-    criterion=torch.nn.CrossEntropyLoss,
+    module=EEGInception,
     optimizer=torch.optim.Adam,
     optimizer__lr=LEARNING_RATE,
     batch_size=BATCH_SIZE,
@@ -46,9 +38,6 @@
         EpochScoring(
             scoring="accuracy", on_train=False, name="valid_acc", lower_is_better=False
         ),
-        InputShapeSetterEEG(
-            params_list=["in_channels", "input_window_samples", "n_classes"],
-        ),
     ],
     verbose=VERBOSE,  # Not printing the results for each epoch
 )
@@ -57,7 +46,6 @@
 pipes = Pipeline(
     [
         ("resample", Resampler_Epoch(128)),
-        ("braindecode_dataset", create_dataset),
         ("EEGInception", clf),
     ]
 )

diff --git a/examples/pipelines_braindecode/braindecode_EEGNetv4.py b/examples/pipelines_braindecode/braindecode_EEGNetv4.py
@@ -6,7 +6,6 @@
 from skorch.dataset import ValidSplit
 
 from moabb.pipelines.features import Resampler_Epoch
-from moabb.pipelines.utils_pytorch import BraindecodeDatasetLoader, InputShapeSetterEEG
 
 
 # Set up GPU if it is there
@@ -22,16 +21,9 @@
 EPOCH = 10
 PATIENCE = 3
 
-# Create the dataset
-create_dataset = BraindecodeDatasetLoader()
-
-# Set random Model
-model = EEGNetv4(in_chans=1, n_classes=2, input_window_samples=100)
-
 # Define a Skorch classifier
 clf = EEGClassifier(
-    module=model,
-    criterion=torch.nn.CrossEntropyLoss,
+    module=EEGNetv4,
     optimizer=torch.optim.Adam,
     optimizer__lr=LEARNING_RATE,
     batch_size=BATCH_SIZE,
@@ -46,9 +38,6 @@
         EpochScoring(
             scoring="accuracy", on_train=False, name="valid_acc", lower_is_better=False
         ),
-        InputShapeSetterEEG(
-            params_list=["in_chans", "input_window_samples", "n_classes"],
-        ),
     ],
     verbose=VERBOSE,  # Not printing the results for each epoch
 )
@@ -57,7 +46,6 @@
 pipes = Pipeline(
     [
         ("resample", Resampler_Epoch(128)),
-        ("braindecode_dataset", create_dataset),
         ("EEGNetv4", clf),
     ]
 )

diff --git a/examples/pipelines_braindecode/braindecode_ShallowFBCSPNet.py b/examples/pipelines_braindecode/braindecode_ShallowFBCSPNet.py
@@ -6,7 +6,6 @@
 from skorch.dataset import ValidSplit
 
 from moabb.pipelines.features import Resampler_Epoch
-from moabb.pipelines.utils_pytorch import BraindecodeDatasetLoader, InputShapeSetterEEG
 
 
 # Set up GPU if it is there
@@ -22,18 +21,10 @@
 EPOCH = 5
 PATIENCE = 3
 
-# Create the dataset
-create_dataset = BraindecodeDatasetLoader()
-
-# Set random Model
-model = ShallowFBCSPNet(
-    in_chans=1, n_classes=2, input_window_samples=100, final_conv_length="auto"
-)
-
 # Define a Skorch classifier
 clf = EEGClassifier(
-    module=model,
-    criterion=torch.nn.CrossEntropyLoss,
+    module=ShallowFBCSPNet,
+    module__final_conv_length="auto",
     optimizer=torch.optim.Adam,
     optimizer__lr=LEARNING_RATE,
     batch_size=BATCH_SIZE,
@@ -48,9 +39,6 @@
         EpochScoring(
             scoring="accuracy", on_train=False, name="valid_acc", lower_is_better=False
         ),
-        InputShapeSetterEEG(
-            params_list=["in_chans", "input_window_samples", "n_classes"],
-        ),
     ],
     verbose=VERBOSE,  # Not printing the results for each epoch
 )
@@ -59,7 +47,6 @@
 pipes = Pipeline(
     [
         ("resample", Resampler_Epoch(250)),
-        ("braindecode_dataset", create_dataset),
         ("ShallowFBCSPNet", clf),
     ]
 )

diff --git a/examples/plot_braindecode.py b/examples/plot_braindecode.py
@@ -16,14 +16,13 @@
 import torch
 from braindecode import EEGClassifier
 from braindecode.models import EEGNetv4
-from sklearn.pipeline import Pipeline
+from sklearn.pipeline import make_pipeline
 from skorch.callbacks import EarlyStopping, EpochScoring
 from skorch.dataset import ValidSplit
 
 from moabb.datasets import BNCI2014_001
 from moabb.evaluations import CrossSessionEvaluation
 from moabb.paradigms import MotorImagery
-from moabb.pipelines.utils_pytorch import BraindecodeDatasetLoader, InputShapeSetterEEG
 from moabb.utils import setup_seed
 
 
@@ -81,8 +80,6 @@
 )
 subjects = [1]
 X, _, _ = paradigm.get_data(dataset=dataset, subjects=subjects)
-# Define Transformer of Dataset compatible with Brain Decode
-create_dataset = BraindecodeDatasetLoader()
 
 ##############################################################################
 # Create Pipelines
@@ -94,16 +91,9 @@
 # callbacks InputShapeSetterEEG, where we have to specify the correct name of the parameter.
 # Here, we will use the EEGNet v4 model [1]_ .
 
-model = EEGNetv4(in_chans=1, n_classes=1, input_window_samples=100)
-
-# Send model to GPU
-if cuda:
-    model.cuda()
-
 # Define a Skorch classifier
 clf = EEGClassifier(
-    module=model,
-    criterion=torch.nn.CrossEntropyLoss,
+    module=EEGNetv4,
     optimizer=torch.optim.Adam,
     optimizer__lr=LEARNING_RATE,
     batch_size=BATCH_SIZE,
@@ -118,16 +108,13 @@
         EpochScoring(
             scoring="accuracy", on_train=False, name="valid_acc", lower_is_better=False
         ),
-        InputShapeSetterEEG(
-            params_list=["in_chans", "input_window_samples", "n_classes"],
-        ),
     ],
     verbose=1,  # Not printing the results for each epoch
 )
 
 # Create the pipelines
 pipes = {}
-pipes["EEGNetV4"] = Pipeline([("Braindecode_dataset", create_dataset), ("Net", clf)])
+pipes["EEGNetV4"] = make_pipeline(clf)
 
 ##############################################################################
 # Evaluation