deploy: braindecode/braindecode@7263a4c

dev/.buildinfo

@@ -1,4 +1,4 @@
 # Sphinx build info version 1
 # This file hashes the configuration used when building these files. When it is not found, a full rebuild will be done.
-config: b2469e7cb779c069e8d7ccba8affea43
+config: e088aa471188a8316236d3f2d62304d5
 tags: 645f666f9bcd5a90fca523b33c5a78b7

dev/_downloads/0a8b8bc2f1b933515b7b4101626dd179/plot_bcic_iv_2a_moabb_trial.py

@@ -39,7 +39,7 @@
 from braindecode.datasets import MOABBDataset
 
 subject_id = 3
-dataset = MOABBDataset(dataset_name="BNCI2014001", subject_ids=[subject_id])
+dataset = MOABBDataset(dataset_name="BNCI2014_001", subject_ids=[subject_id])
 
 
 ######################################################################
@@ -131,12 +131,12 @@
 ######################################################################
 # We can easily split the dataset using additional info stored in the
 # description attribute, in this case ``session`` column. We select
-# ``session_T`` for training and ``session_E`` for validation.
+# ``T`` for training and ``test`` for validation.
 #
 
 splitted = windows_dataset.split('session')
-train_set = splitted['session_T']
-valid_set = splitted['session_E']
+train_set = splitted['0train']  # Session train
+valid_set = splitted['1test']  # Session evaluation
 
 
 ######################################################################

dev/_downloads/0f763ae384277e558103757157e170fb/plot_data_augmentation_search.py

@@ -123,8 +123,8 @@
 
 
 splitted = windows_dataset.split('session')
-train_set = splitted['session_T']
-eval_set = splitted['session_E']
+train_set = splitted['0train']  # Session train
+eval_set = splitted['1test']  # Session evaluation
 
 ######################################################################
 # Defining a list of transforms

dev/_downloads/1d879df548fa18be8c23d9ca0dc008d4/plot_data_augmentation.ipynb

@@ -87,7 +87,7 @@
       },
       "outputs": [],
       "source": [
-        "splitted = windows_dataset.split('session')\ntrain_set = splitted['session_T']\nvalid_set = splitted['session_E']"
+        "splitted = windows_dataset.split('session')\ntrain_set = splitted['0train']  # Session train\nvalid_set = splitted['1test']  # Session evaluation"
       ]
     },
     {

dev/_downloads/2466f8ec5c733d0bd65e187b45d875cc/plot_data_augmentation_search.ipynb

@@ -105,7 +105,7 @@
       },
       "outputs": [],
       "source": [
-        "splitted = windows_dataset.split('session')\ntrain_set = splitted['session_T']\neval_set = splitted['session_E']"
+        "splitted = windows_dataset.split('session')\ntrain_set = splitted['0train']  # Session train\neval_set = splitted['1test']  # Session evaluation"
       ]
     },
     {

dev/_downloads/263464a28477cf8decb861ae6e2e9be7/plot_how_train_test_and_tune.ipynb

@@ -93,7 +93,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "We can easily split the dataset BCIC IV 2a dataset using additional\ninfo stored in the description attribute, in this case the ``session``\ncolumn. We select ``session_T`` for training and ``session_E`` for testing.\nFor other datasets, you might have to choose another column and/or column.\n\n<div class=\"alert alert-info\"><h4>Note</h4><p>No matter which of the three schemes you use, this initial\n   two-fold split into train_set and test_set always remains the same.\n   Remember that you are not allowed to use the test_set during any\n   stage of training or tuning.</p></div>\n\n\n"
+        "We can easily split the dataset BCIC IV 2a dataset using additional\ninfo stored in the description attribute, in this case the ``session``\ncolumn. We select ``0train`` for training and ``0test`` for testing.\nFor other datasets, you might have to choose another column and/or column.\n\n<div class=\"alert alert-info\"><h4>Note</h4><p>No matter which of the three schemes you use, this initial\n   two-fold split into train_set and test_set always remains the same.\n   Remember that you are not allowed to use the test_set during any\n   stage of training or tuning.</p></div>\n\n\n"
       ]
     },
     {
@@ -104,7 +104,7 @@
       },
       "outputs": [],
       "source": [
-        "splitted = windows_dataset.split(\"session\")\ntrain_set = splitted[\"session_T\"]\ntest_set = splitted[\"session_E\"]"
+        "splitted = windows_dataset.split(\"session\")\ntrain_set = splitted['0train']  # Session train\ntest_set = splitted['1test']  # Session evaluation"
       ]
     },
     {
@@ -233,7 +233,7 @@
       },
       "outputs": [],
       "source": [
-        "def plot_train_valid_test(ax, all_dataset, train_subset, val_subset, test_set):\n    \"\"\"Create a sample plot for training, validation, testing.\"\"\"\n\n    bd_cmap = [\"#3A6190\", \"#683E00\", \"#2196F3\", \"#DDF2FF\",]\n\n    n_train, n_val, n_test = len(train_subset), len(val_subset), len(test_set)\n    ax.barh(\"Original\\ndataset\", len(all_dataset), left=0, height=0.5, color=bd_cmap[0])\n    ax.barh(\"Train-Test-Valid\\nsplit\", n_train, left=0, height=0.5, color=bd_cmap[1])\n    ax.barh(\"Train-Test-Valid\\nsplit\", n_val, left=n_train, height=0.5, color=bd_cmap[2])\n    ax.barh(\"Train-Test-Valid\\nsplit\", n_test, left=n_train + n_val, height=0.5, color=bd_cmap[3])\n\n    ax.invert_yaxis()\n    ax.set(xlabel=\"Number of samples.\", title=\"Train-Test-Valid split\")\n    ax.legend([\"Original set\", \"Training set\", \"Validation set\", \"Testing set\"],\n              loc=\"lower center\", ncols=2, bbox_to_anchor=(0.5, 0.4))\n    ax.set_xlim([-int(0.1 * len(all_dataset)), int(1.1 * len(all_dataset))])\n    return ax\n\n\nfig, ax = plt.subplots(figsize=(12, 5))\nplot_train_valid_test(ax=ax, all_dataset=windows_dataset,\n                      train_subset=train_subset, val_subset=val_subset, test_set=test_set,)"
+        "def plot_train_valid_test(ax, all_dataset, train_subset, val_subset, test_set):\n    \"\"\"Create a sample plot for training, validation, testing.\"\"\"\n\n    bd_cmap = [\"#3A6190\", \"#683E00\", \"#2196F3\", \"#DDF2FF\", ]\n\n    n_train, n_val, n_test = len(train_subset), len(val_subset), len(test_set)\n    ax.barh(\"Original\\ndataset\", len(all_dataset), left=0, height=0.5, color=bd_cmap[0])\n    ax.barh(\"Train-Test-Valid\\nsplit\", n_train, left=0, height=0.5, color=bd_cmap[1])\n    ax.barh(\"Train-Test-Valid\\nsplit\", n_val, left=n_train, height=0.5, color=bd_cmap[2])\n    ax.barh(\"Train-Test-Valid\\nsplit\", n_test, left=n_train + n_val, height=0.5, color=bd_cmap[3])\n\n    ax.invert_yaxis()\n    ax.set(xlabel=\"Number of samples.\", title=\"Train-Test-Valid split\")\n    ax.legend([\"Original set\", \"Training set\", \"Validation set\", \"Testing set\"],\n              loc=\"lower center\", ncols=2, bbox_to_anchor=(0.5, 0.4))\n    ax.set_xlim([-int(0.1 * len(all_dataset)), int(1.1 * len(all_dataset))])\n    return ax\n\n\nfig, ax = plt.subplots(figsize=(12, 5))\nplot_train_valid_test(ax=ax, all_dataset=windows_dataset,\n                      train_subset=train_subset, val_subset=val_subset, test_set=test_set,)"
       ]
     },
     {
@@ -276,7 +276,7 @@
       },
       "outputs": [],
       "source": [
-        "def plot_k_fold(ax, cv, all_dataset, X_train, y_train, test_set):\n    \"\"\"Create a sample plot for training, validation, testing.\"\"\"\n\n    bd_cmap = [\"#3A6190\", \"#683E00\", \"#2196F3\", \"#DDF2FF\",]\n\n    ax.barh(\"Original\\nDataset\", len(all_dataset), left=0, height=0.5, color=bd_cmap[0])\n\n    # Generate the training/validation/testing data fraction visualizations for each CV split\n    for ii, (tr_idx, val_idx) in enumerate(cv.split(X=X_train, y=y_train)):\n        n_train, n_val, n_test = len(tr_idx), len(val_idx), len(test_set)\n        n_train2 = n_train + n_val - max(val_idx) - 1\n        ax.barh(\"cv\" + str(ii + 1), min(val_idx), left=0, height=0.5, color=bd_cmap[1])\n        ax.barh(\"cv\" + str(ii + 1), n_val, left=min(val_idx), height=0.5, color=bd_cmap[2])\n        ax.barh(\"cv\" + str(ii + 1), n_train2, left=max(val_idx) + 1, height=0.5, color=bd_cmap[1])\n        ax.barh(\"cv\" + str(ii + 1), n_test, left=n_train + n_val, height=0.5, color=bd_cmap[3])\n\n    ax.invert_yaxis()\n    ax.set_xlim([-int(0.1 * len(all_dataset)), int(1.1 * len(all_dataset))])\n    ax.set(xlabel=\"Number of samples.\", title=\"KFold Train-Test-Valid split\")\n    ax.legend([Patch(color=bd_cmap[i]) for i in range(4)],\n              [\"Original set\", \"Training set\", \"Validation set\", \"Testing set\"],\n              loc=\"lower center\", ncols=2)\n    ax.text(-0.07, 0.45, 'Train-Valid-Test split', rotation=90,\n            verticalalignment='center', horizontalalignment='left', transform=ax.transAxes)\n    return ax\n\n\nfig, ax = plt.subplots(figsize=(15, 7))\nplot_k_fold(ax, cv=train_val_split, all_dataset=windows_dataset,\n            X_train=X_train, y_train=y_train, test_set=test_set,)"
+        "def plot_k_fold(ax, cv, all_dataset, X_train, y_train, test_set):\n    \"\"\"Create a sample plot for training, validation, testing.\"\"\"\n\n    bd_cmap = [\"#3A6190\", \"#683E00\", \"#2196F3\", \"#DDF2FF\", ]\n\n    ax.barh(\"Original\\nDataset\", len(all_dataset), left=0, height=0.5, color=bd_cmap[0])\n\n    # Generate the training/validation/testing data fraction visualizations for each CV split\n    for ii, (tr_idx, val_idx) in enumerate(cv.split(X=X_train, y=y_train)):\n        n_train, n_val, n_test = len(tr_idx), len(val_idx), len(test_set)\n        n_train2 = n_train + n_val - max(val_idx) - 1\n        ax.barh(\"cv\" + str(ii + 1), min(val_idx), left=0, height=0.5, color=bd_cmap[1])\n        ax.barh(\"cv\" + str(ii + 1), n_val, left=min(val_idx), height=0.5, color=bd_cmap[2])\n        ax.barh(\"cv\" + str(ii + 1), n_train2, left=max(val_idx) + 1, height=0.5, color=bd_cmap[1])\n        ax.barh(\"cv\" + str(ii + 1), n_test, left=n_train + n_val, height=0.5, color=bd_cmap[3])\n\n    ax.invert_yaxis()\n    ax.set_xlim([-int(0.1 * len(all_dataset)), int(1.1 * len(all_dataset))])\n    ax.set(xlabel=\"Number of samples.\", title=\"KFold Train-Test-Valid split\")\n    ax.legend([Patch(color=bd_cmap[i]) for i in range(4)],\n              [\"Original set\", \"Training set\", \"Validation set\", \"Testing set\"],\n              loc=\"lower center\", ncols=2)\n    ax.text(-0.07, 0.45, 'Train-Valid-Test split', rotation=90,\n            verticalalignment='center', horizontalalignment='left', transform=ax.transAxes)\n    return ax\n\n\nfig, ax = plt.subplots(figsize=(15, 7))\nplot_k_fold(ax, cv=train_val_split, all_dataset=windows_dataset,\n            X_train=X_train, y_train=y_train, test_set=test_set,)"
       ]
     },
     {

dev/_downloads/3862cafd7f0d815e434319ffe525afc8/plot_bcic_iv_2a_moabb_cropped.ipynb

@@ -29,7 +29,7 @@
       },
       "outputs": [],
       "source": [
-        "from braindecode.datasets import MOABBDataset\n\nsubject_id = 3\ndataset = MOABBDataset(dataset_name=\"BNCI2014001\", subject_ids=[subject_id])\n\nfrom numpy import multiply\n\nfrom braindecode.preprocessing import (\n    Preprocessor,\n    exponential_moving_standardize,\n    preprocess,\n)\n\nlow_cut_hz = 4.  # low cut frequency for filtering\nhigh_cut_hz = 38.  # high cut frequency for filtering\n# Parameters for exponential moving standardization\nfactor_new = 1e-3\ninit_block_size = 1000\n# Factor to convert from V to uV\nfactor = 1e6\n\npreprocessors = [\n    Preprocessor('pick_types', eeg=True, meg=False, stim=False),  # Keep EEG sensors\n    Preprocessor(lambda data: multiply(data, factor)),  # Convert from V to uV\n    Preprocessor('filter', l_freq=low_cut_hz, h_freq=high_cut_hz),  # Bandpass filter\n    Preprocessor(exponential_moving_standardize,  # Exponential moving standardization\n                 factor_new=factor_new, init_block_size=init_block_size)\n]\n\n# Transform the data\npreprocess(dataset, preprocessors, n_jobs=-1)"
+        "from braindecode.datasets import MOABBDataset\n\nsubject_id = 3\ndataset = MOABBDataset(dataset_name=\"BNCI2014_001\", subject_ids=[subject_id])\n\nfrom numpy import multiply\n\nfrom braindecode.preprocessing import (\n    Preprocessor,\n    exponential_moving_standardize,\n    preprocess,\n)\n\nlow_cut_hz = 4.  # low cut frequency for filtering\nhigh_cut_hz = 38.  # high cut frequency for filtering\n# Parameters for exponential moving standardization\nfactor_new = 1e-3\ninit_block_size = 1000\n# Factor to convert from V to uV\nfactor = 1e6\n\npreprocessors = [\n    Preprocessor('pick_types', eeg=True, meg=False, stim=False),\n    # Keep EEG sensors\n    Preprocessor(lambda data: multiply(data, factor)),  # Convert from V to uV\n    Preprocessor('filter', l_freq=low_cut_hz, h_freq=high_cut_hz),\n    # Bandpass filter\n    Preprocessor(exponential_moving_standardize,\n                 # Exponential moving standardization\n                 factor_new=factor_new, init_block_size=init_block_size)\n]\n\n# Transform the data\npreprocess(dataset, preprocessors, n_jobs=-1)"
       ]
     },
     {
@@ -137,7 +137,7 @@
       },
       "outputs": [],
       "source": [
-        "splitted = windows_dataset.split('session')\ntrain_set = splitted['session_T']\nvalid_set = splitted['session_E']"
+        "splitted = windows_dataset.split('session')\ntrain_set = splitted['0train']  # Session train\nvalid_set = splitted['1test']  # Session evaluation"
       ]
     },
     {
@@ -155,7 +155,7 @@
       },
       "outputs": [],
       "source": [
-        "from skorch.callbacks import LRScheduler\nfrom skorch.helper import predefined_split\n\nfrom braindecode import EEGClassifier\nfrom braindecode.training import CroppedLoss\n\n# These values we found good for shallow network:\nlr = 0.0625 * 0.01\nweight_decay = 0\n\n# For deep4 they should be:\n# lr = 1 * 0.01\n# weight_decay = 0.5 * 0.001\n\nbatch_size = 64\nn_epochs = 2\n\nclf = EEGClassifier(\n    model,\n    cropped=True,\n    criterion=CroppedLoss,\n    criterion__loss_function=torch.nn.functional.nll_loss,\n    optimizer=torch.optim.AdamW,\n    train_split=predefined_split(valid_set),\n    optimizer__lr=lr,\n    optimizer__weight_decay=weight_decay,\n    iterator_train__shuffle=True,\n    batch_size=batch_size,\n    callbacks=[\n        \"accuracy\", (\"lr_scheduler\", LRScheduler('CosineAnnealingLR', T_max=n_epochs - 1)),\n    ],\n    device=device,\n    classes=classes,\n)\n# Model training for a specified number of epochs. `y` is None as it is already supplied\n# in the dataset.\n_ = clf.fit(train_set, y=None, epochs=n_epochs)"
+        "from skorch.callbacks import LRScheduler\nfrom skorch.helper import predefined_split\n\nfrom braindecode import EEGClassifier\nfrom braindecode.training import CroppedLoss\n\n# These values we found good for shallow network:\nlr = 0.0625 * 0.01\nweight_decay = 0\n\n# For deep4 they should be:\n# lr = 1 * 0.01\n# weight_decay = 0.5 * 0.001\n\nbatch_size = 64\nn_epochs = 2\n\nclf = EEGClassifier(\n    model,\n    cropped=True,\n    criterion=CroppedLoss,\n    criterion__loss_function=torch.nn.functional.nll_loss,\n    optimizer=torch.optim.AdamW,\n    train_split=predefined_split(valid_set),\n    optimizer__lr=lr,\n    optimizer__weight_decay=weight_decay,\n    iterator_train__shuffle=True,\n    batch_size=batch_size,\n    callbacks=[\n        \"accuracy\",\n        (\"lr_scheduler\", LRScheduler('CosineAnnealingLR', T_max=n_epochs - 1)),\n    ],\n    device=device,\n    classes=classes,\n)\n# Model training for a specified number of epochs. `y` is None as it is already supplied\n# in the dataset.\n_ = clf.fit(train_set, y=None, epochs=n_epochs)"
       ]
     },
     {
@@ -173,7 +173,7 @@
       },
       "outputs": [],
       "source": [
-        "import matplotlib.pyplot as plt\nimport pandas as pd\nfrom matplotlib.lines import Line2D\n\n# Extract loss and accuracy values for plotting from history object\nresults_columns = ['train_loss', 'valid_loss', 'train_accuracy', 'valid_accuracy']\ndf = pd.DataFrame(clf.history[:, results_columns], columns=results_columns,\n                  index=clf.history[:, 'epoch'])\n\n# get percent of misclass for better visual comparison to loss\ndf = df.assign(train_misclass=100 - 100 * df.train_accuracy,\n               valid_misclass=100 - 100 * df.valid_accuracy)\n\nfig, ax1 = plt.subplots(figsize=(8, 3))\ndf.loc[:, ['train_loss', 'valid_loss']].plot(\n    ax=ax1, style=['-', ':'], marker='o', color='tab:blue', legend=False, fontsize=14)\n\nax1.tick_params(axis='y', labelcolor='tab:blue', labelsize=14)\nax1.set_ylabel(\"Loss\", color='tab:blue', fontsize=14)\n\nax2 = ax1.twinx()  # instantiate a second axes that shares the same x-axis\n\ndf.loc[:, ['train_misclass', 'valid_misclass']].plot(\n    ax=ax2, style=['-', ':'], marker='o', color='tab:red', legend=False)\nax2.tick_params(axis='y', labelcolor='tab:red', labelsize=14)\nax2.set_ylabel(\"Misclassification Rate [%]\", color='tab:red', fontsize=14)\nax2.set_ylim(ax2.get_ylim()[0], 85)  # make some room for legend\nax1.set_xlabel(\"Epoch\", fontsize=14)\n\n# where some data has already been plotted to ax\nhandles = []\nhandles.append(Line2D([0], [0], color='black', linewidth=1, linestyle='-', label='Train'))\nhandles.append(Line2D([0], [0], color='black', linewidth=1, linestyle=':', label='Valid'))\nplt.legend(handles, [h.get_label() for h in handles], fontsize=14)\nplt.tight_layout()"
+        "import matplotlib.pyplot as plt\nimport pandas as pd\nfrom matplotlib.lines import Line2D\n\n# Extract loss and accuracy values for plotting from history object\nresults_columns = ['train_loss', 'valid_loss', 'train_accuracy',\n                   'valid_accuracy']\ndf = pd.DataFrame(clf.history[:, results_columns], columns=results_columns,\n                  index=clf.history[:, 'epoch'])\n\n# get percent of misclass for better visual comparison to loss\ndf = df.assign(train_misclass=100 - 100 * df.train_accuracy,\n               valid_misclass=100 - 100 * df.valid_accuracy)\n\nfig, ax1 = plt.subplots(figsize=(8, 3))\ndf.loc[:, ['train_loss', 'valid_loss']].plot(\n    ax=ax1, style=['-', ':'], marker='o', color='tab:blue', legend=False,\n    fontsize=14)\n\nax1.tick_params(axis='y', labelcolor='tab:blue', labelsize=14)\nax1.set_ylabel(\"Loss\", color='tab:blue', fontsize=14)\n\nax2 = ax1.twinx()  # instantiate a second axes that shares the same x-axis\n\ndf.loc[:, ['train_misclass', 'valid_misclass']].plot(\n    ax=ax2, style=['-', ':'], marker='o', color='tab:red', legend=False)\nax2.tick_params(axis='y', labelcolor='tab:red', labelsize=14)\nax2.set_ylabel(\"Misclassification Rate [%]\", color='tab:red', fontsize=14)\nax2.set_ylim(ax2.get_ylim()[0], 85)  # make some room for legend\nax1.set_xlabel(\"Epoch\", fontsize=14)\n\n# where some data has already been plotted to ax\nhandles = []\nhandles.append(\n    Line2D([0], [0], color='black', linewidth=1, linestyle='-', label='Train'))\nhandles.append(\n    Line2D([0], [0], color='black', linewidth=1, linestyle=':', label='Valid'))\nplt.legend(handles, [h.get_label() for h in handles], fontsize=14)\nplt.tight_layout()"
       ]
     },
     {

dev/_downloads/5adeee55d3085fdf6cf644f6c7d12084/plot_hyperparameter_tuning_with_scikit-learn.py

@@ -172,12 +172,12 @@
 ######################################################################
 # We can easily split the dataset using additional info stored in the
 # description attribute, in this case ``session`` column. We select
-# ``session_T`` for training and ``session_E`` for evaluation.
+# ``0train`` for training and ``1test`` for evaluation.
 #
 
 splitted = windows_dataset.split('session')
-train_set = splitted['session_T']
-eval_set = splitted['session_E']
+train_set = splitted['0train']  # Session train
+eval_set = splitted['1test']  # Session evaluation
 
 ######################################################################
 # Create model

dev/_downloads/66bb851a261b0b302f4dcb9e2f361aa6/plot_train_in_pure_pytorch_and_pytorch_lightning.py

@@ -73,7 +73,7 @@
 from braindecode.datasets import MOABBDataset
 
 subject_id = 3
-dataset = MOABBDataset(dataset_name="BNCI2014001", subject_ids=[subject_id])
+dataset = MOABBDataset(dataset_name="BNCI2014_001", subject_ids=[subject_id])
 
 ######################################################################
 # Preprocessing, the offline transformation of the raw dataset
@@ -185,7 +185,7 @@
 ######################################################################
 # We can easily split the dataset using additional info stored in the
 # description attribute, in this case the ``session`` column. We
-# select ``session_T`` for training and ``session_E`` for testing.
+# select ``Train`` for training and ``test`` for testing.
 # For other datasets, you might have to choose another column.
 #
 # .. note::
@@ -196,8 +196,8 @@
 #
 
 splitted = windows_dataset.split("session")
-train_set = splitted["session_T"]
-test_set = splitted["session_E"]
+train_set = splitted['0train']  # Session train
+test_set = splitted['1test']  # Session evaluation
 
 ######################################################################
 # Option 1: Pure PyTorch training loop