braindecode 1.2.0.dev169062562__py3-none-any.whl → 1.2.0.dev176281713__py3-none-any.whl

braindecode/models/eegconformer.py

@@ -12,33 +12,129 @@ from braindecode.modules import FeedForwardBlock, MultiHeadAttention
 
 
 class EEGConformer(EEGModuleMixin, nn.Module):
-    """EEG Conformer from Song et al. (2022) from [song2022]_.
+    """EEG Conformer from Song et al. (2022) [song2022]_.
 
-     .. figure:: https://raw.githubusercontent.com/eeyhsong/EEG-Conformer/refs/heads/main/visualization/Fig1.png
+    :bdg-success:`Convolution` :bdg-info:`Small Attention`
+
+    .. figure:: https://raw.githubusercontent.com/eeyhsong/EEG-Conformer/refs/heads/main/visualization/Fig1.png
         :align: center
         :alt: EEGConformer Architecture
+        :width: 600px
+
+
+    .. rubric:: Architectural Overview
+
+    EEG-Conformer is a *convolution-first* model augmented with a *lightweight transformer
+    encoder*. The end-to-end flow is:
+
+    - (i) :class:`_PatchEmbedding` converts the continuous EEG into a compact sequence of tokens via a
+      :class:`ShallowFBCSPNet` temporal–spatial conv stem and temporal pooling;
+    - (ii) :class:`_TransformerEncoder` applies small multi-head self-attention to integrate
+      longer-range temporal context across tokens;
+    - (iii) :class:`_ClassificationHead` aggregates the sequence and performs a linear readout.
+      This preserves the strong inductive biases of shallow CNN filter banks while adding
+      just enough attention to capture dependencies beyond the pooling horizon [song2022]_.
+
+    .. rubric:: Macro Components
+
+    - :class:`_PatchEmbedding` **(Shallow conv stem → tokens)**
+
+        - *Operations.*
+        - A temporal convolution (`:class:torch.nn.Conv2d`) ``(1 x L_t)`` forms a data-driven "filter bank";
+        - A spatial convolution (`:class:torch.nn.Conv2d`) (n_chans x 1)`` projects across electrodes,
+          collapsing the channel axis into a virtual channel.
+        - **Normalization function** :class:`torch.nn.BatchNorm`
+        - **Activation function** :class:`torch.nn.ELU`
+        - **Average Pooling** :class:`torch.nn.AvgPool` along time (kernel ``(1, P)`` with stride ``(1, S)``)
+        -  final ``1x1`` :class:`torch.nn.Linear` projection.
+
+    The result is rearranged to a token sequence ``(B, S_tokens, D)``, where ``D = n_filters_time``.
+
+    *Interpretability/robustness.* Temporal kernels can be inspected as FIR filters;
+    the spatial conv yields channel projections analogous to :class:`ShallowFBCSPNet`’s learned
+    spatial filters. Temporal pooling stabilizes statistics and reduces sequence length.
+
+    - :class:`_TransformerEncoder` **(context over temporal tokens)**
+
+        - *Operations.*
+        - A stack of ``att_depth`` encoder blocks. :class:`_TransformerEncoderBlock`
+        - Each block applies LayerNorm :class:`torch.nn.LayerNorm`
+        - Multi-Head Self-Attention (``att_heads``) with dropout + residual :class:`MultiHeadAttention` (:class:`torch.nn.Dropout`)
+        - LayerNorm :class:`torch.nn.LayerNorm`
+        - 2-layer feed-forward (≈4x expansion, :class:`torch.nn.GELU`) with dropout + residual.
+
+    Shapes remain ``(B, S_tokens, D)`` throughout.
 
-    Convolutional Transformer for EEG decoding.
+    *Role.* Small attention focuses on interactions among *temporal patches* (not channels),
+    extending effective receptive fields at modest cost.
 
-    The paper and original code with more details about the methodological
-    choices are available at the [song2022]_ and [ConformerCode]_.
+    - :class:`ClassificationHead` **(aggregation + readout)**
 
-    This neural network architecture receives a traditional braindecode input.
-    The input shape should be three-dimensional matrix representing the EEG
-    signals.
+        - *Operations*.
+        - Flatten, :class:`torch.nn.Flatten` the sequence ``(B, S_tokens·D)`` -
+        - MLP (:class:`torch.nn.Linear` → activation (default: :class:`torch.nn.ELU`) → :class:`torch.nn.Dropout` → :class:`torch.nn.Linear`)
+        - final Linear to classes.
 
-         `(batch_size, n_channels, n_timesteps)`.
+    With ``return_features=True``, features before the last Linear can be exported for
+    linear probing or downstream tasks.
 
-    The EEG Conformer architecture is composed of three modules:
-        - PatchEmbedding
-        - TransformerEncoder
-        - ClassificationHead
+    .. rubric:: Convolutional Details
+
+    - **Temporal (where time-domain patterns are learned).**
+        The initial ``(1 x L_t)`` conv per channel acts as a *learned filter bank* for oscillatory
+        bands and transients. Subsequent **AvgPool** along time performs local integration,
+        converting activations into “patches” (tokens). Pool length/stride control the
+        token rate and set the lower bound on temporal context within each token.
+
+    - **Spatial (how electrodes are processed).**
+        A single conv with kernel ``(n_chans x 1)`` spans the full montage to learn spatial
+        projections for each temporal feature map, collapsing the channel axis into a
+        virtual channel before tokenization. This mirrors the shallow spatial step in
+        :class:`ShallowFBCSPNet` (temporal filters → spatial projection → temporal condensation).
+
+    - **Spectral (how frequency content is captured).**
+        No explicit Fourier/wavelet stage is used. Spectral selectivity emerges implicitly
+        from the learned temporal kernels; pooling further smooths high-frequency noise.
+        The effective spectral resolution is thus governed by ``L_t`` and the pooling
+        configuration.
+
+    .. rubric:: Attention / Sequential Modules
+
+    - **Type.** Standard multi-head self-attention (MHA) with ``att_heads`` heads over the token sequence.
+    - **Shapes.** Input/Output: ``(B, S_tokens, D)``; attention operates along the ``S_tokens`` axis.
+    - **Role.** Re-weights and integrates evidence across pooled windows, capturing dependencies
+      longer than any single token while leaving channel relationships to the convolutional stem.
+      The design is intentionally *small*—attention refines rather than replaces convolutional feature extraction.
+
+    .. rubric:: Additional Mechanisms
+
+    - **Parallel with ShallowFBCSPNet.** Both begin with a learned temporal filter bank,
+        spatial projection across electrodes, and early temporal condensation.
+        :class:`ShallowFBCSPNet` then computes band-power (via squaring/log-variance), whereas
+        EEG-Conformer applies BN/ELU and **continues with attention** over tokens to
+        refine temporal context before classification.
+
+    - **Tokenization knob.** ``pool_time_length`` and especially ``pool_time_stride`` set
+        the number of tokens ``S_tokens``. Smaller strides → more tokens and higher attention
+        capacity (but higher compute); larger strides → fewer tokens and stronger inductive bias.
+
+    - **Embedding dimension = filters.** ``n_filters_time`` serves double duty as both the
+        number of temporal filters in the stem and the transformer’s embedding size ``D``,
+        simplifying dimensional alignment.
+
+    .. rubric:: Usage and Configuration
+
+    - **Instantiation.** Choose ``n_filters_time`` (embedding size ``D``) and
+        ``filter_time_length`` to match the rhythms of interest. Tune
+        ``pool_time_length/stride`` to trade temporal resolution for sequence length.
+        Keep ``att_depth`` modest (e.g., 4–6) and set ``att_heads`` to divide ``D``.
+        ``final_fc_length="auto"`` infers the flattened size from PatchEmbedding.
 
     Notes
     -----
     The authors recommend using data augmentation before using Conformer,
     e.g. segmentation and recombination,
-    Please refer to the original paper and code for more details.
+    Please refer to the original paper and code for more details [ConformerCode]_.
 
     The model was initially tuned on 4 seconds of 250 Hz data.
     Please adjust the scale of the temporal convolutional layer,
@@ -47,7 +143,10 @@ class EEGConformer(EEGModuleMixin, nn.Module):
     .. versionadded:: 0.8
 
     We aggregate the parameters based on the parts of the models, or
-    when the parameters were used first, e.g. n_filters_time.
+    when the parameters were used first, e.g. ``n_filters_time``.
+
+    .. versionadded:: 1.1
+
 
     Parameters
     ----------

braindecode/models/eegnet.py

@@ -20,13 +20,59 @@ from braindecode.modules import (
 
 
 class EEGNetv4(EEGModuleMixin, nn.Sequential):
-    """EEGNet v4 model from Lawhern et al. (2018) [EEGNet4]_.
+    """EEGNet v4 model from Lawhern et al. (2018) [Lawhern2018]_.
+
+    :bdg-success:`Convolution` :bdg-secondary:`Depthwise–Separable`
 
     .. figure:: https://content.cld.iop.org/journals/1741-2552/15/5/056013/revision2/jneaace8cf01_hr.jpg
        :align: center
-       :alt: EEGNet4 Architecture
+       :alt: EEGNetv4 Architecture
+       :width: 600px
+
+    .. rubric:: Architectural Overview
+
+    EEGNetv4 is a compact convolutional network designed for EEG decoding with a pipeline that mirrors classical EEG processing:
+    - (i) learn temporal frequency-selective filters,
+    - (ii) learn spatial filters for those frequencies, and
+    - (iii) condense features with depthwise–separable convolutions before a lightweight classifier.
+
+    The architecture is deliberately small (temporal convolutional and spatial patterns) [Lawhern2018]_.
+
+    .. rubric:: Macro Components
+
+    - **Temporal convolution**
+      Temporal convolution applied per channel; learns ``F1`` kernels that act as data-driven band-pass filters.
+    - **Depthwise Spatial Filtering.**
+      Depthwise convolution spanning the channel dimension with ``groups = F1``,
+      yielding ``D`` spatial filters for each temporal filter (no cross-filter mixing).
+    - **Norm–Nonlinearity–Pooling (+ dropout).**
+      Batch normalization → ELU → temporal pooling, with dropout.
+    - **Depthwise–Separable Convolution Block.**
+      (a) depthwise temporal conv to refine temporal structure;
+      (b) pointwise 1x1 conv to mix feature maps into ``F2`` combinations.
+    - **Classifier Head.**
+      Lightweight 1x1 conv or dense layer (often with max-norm constraint).
 
-    See details in [EEGNet4]_.
+    .. rubric:: Convolutional Details
+
+    - **Temporal.** The initial temporal convs serve as a *learned filter bank*:
+      long 1-D kernels (implemented as 2-D with singleton spatial extent) emphasize oscillatory bands and transients.
+      Because this stage is linear prior to BN/ELU, kernels can be analyzed as FIR filters to reveal each feature’s spectrum [Lawhern2018]_.
+
+    - **Spatial.** The depthwise spatial conv spans the full channel axis (kernel height = #electrodes; temporal size = 1).
+      With ``groups = F1``, each temporal filter learns its own set of ``D`` spatial projections—akin to CSP, learned end-to-end and
+      typically regularized with max-norm.
+
+    - **Spectral.** No explicit Fourier/wavelet transform is used. Frequency structure
+      is captured implicitly by the temporal filter bank; later depthwise temporal kernels act as short-time integrators/refiners.
+
+    .. rubric:: Additional Comments
+
+    - **Filter-bank structure:** Parallel temporal kernels (``F1``) emulate classical filter banks; pairing them with frequency-specific spatial filters
+      yields features mappable to rhythms and topographies.
+    - **Depthwise & separable convs:** Parameter-efficient decomposition (depthwise + pointwise) retains power while limiting overfitting
+      [Chollet2017]_ and keeps temporal vs. mixing steps interpretable.
+    - **Regularization:** Batch norm, dropout, pooling, and optional max-norm on spatial kernels aid stability on small EEG datasets.
 
     Parameters
     ----------
@@ -68,10 +114,13 @@ class EEGNetv4(EEGModuleMixin, nn.Sequential):
 
     References
     ----------
-    .. [EEGNet4] Lawhern, V. J., Solon, A. J., Waytowich, N. R., Gordon, S. M.,
+    .. [Lawhern2018] Lawhern, V. J., Solon, A. J., Waytowich, N. R., Gordon, S. M.,
         Hung, C. P., & Lance, B. J. (2018). EEGNet: a compact convolutional
         neural network for EEG-based brain–computer interfaces. Journal of
         neural engineering, 15(5), 056013.
+    .. [Chollet2017] Chollet, F., *Xception: Deep Learning with Depthwise Separable
+        Convolutions*, CVPR, 2017.
+
     """
 
     def __init__(

braindecode/models/eegnex.py

@@ -16,9 +16,124 @@ from braindecode.modules import Conv2dWithConstraint, LinearWithConstraint
 class EEGNeX(EEGModuleMixin, nn.Module):
     """EEGNeX model from Chen et al. (2024) [eegnex]_.
 
+    :bdg-success:`Convolution`
+
     .. figure:: https://braindecode.org/dev/_static/model/eegnex.jpg
         :align: center
         :alt: EEGNeX Architecture
+        :width: 620px
+
+    .. rubric:: Architectural Overview
+
+    EEGNeX is a **purely convolutional** architecture that refines the EEGNet-style stem
+    and deepens the temporal stack with **dilated temporal convolutions**. The end-to-end
+    flow is:
+
+    - (i) **Block-1/2**: two temporal convolutions ``(1 x L)`` with BN refine a
+      learned FIR-like *temporal filter bank* (no pooling yet);
+    - (ii) **Block-3**: depthwise **spatial** convolution across electrodes
+      ``(n_chans x 1)`` with max-norm constraint, followed by ELU → AvgPool (time) → Dropout;
+    - (iii) **Block-4/5**: two additional **temporal** convolutions with increasing **dilation**
+      to expand the receptive field; the last block applies ELU → AvgPool → Dropout → Flatten;
+    - (iv) **Classifier**: a max-norm–constrained linear layer.
+
+    The published work positions EEGNeX as a compact, conv-only alternative that consistently
+    outperforms prior baselines across MOABB-style benchmarks, with the popular
+    “EEGNeX-8,32” shorthand denoting *8 temporal filters* and *kernel length 32*.
+
+
+    .. rubric:: Macro Components
+
+    - **Block-1 / Block-2 — Temporal filter (learned).**
+
+       - *Operations.*
+       - :class:`torch.nn.Conv2d` with kernels ``(1, L)``
+       - :class:`torch.nn.BatchNorm2d` (no nonlinearity until Block-3, mirroring a linear FIR analysis stage).
+         These layers set up frequency-selective detectors before spatial mixing.
+
+    - *Interpretability.* Kernels can be inspected as FIR filters; two stacked temporal
+      convs allow longer effective kernels without parameter blow-up.
+
+    - **Block-3 — Spatial projection + condensation.**
+
+        - *Operations.*
+        - :class:`braindecode.modules.Conv2dWithConstraint` with kernel``(n_chans, 1)``
+          and ``groups = filter_2`` (depthwise across filters)
+        - :class:`torch.nn.BatchNorm2d`
+        - :class:`torch.nn.ELU`
+        - :class:`torch.nn.AvgPool2d` (time)
+        - :class:`torch.nn.Dropout`.
+
+    *Role.* Learns per-filter spatial patterns over the **full montage** while temporal
+      pooling stabilizes and compresses features; max-norm encourages well-behaved spatial
+      weights similar to EEGNet practice.
+
+    - **Block-4 / Block-5 — Dilated temporal integration.**
+
+        - *Operations.*
+        - :class:`torch.nn.Conv2d` with kernels ``(1, k)`` and **dilations**
+        (e.g., 2 then 4);
+        - :class:`torch.nn.BatchNorm2d`
+        - :class:`torch.nn.ELU`
+        - :class:`torch.nn.AvgPool2d` (time)
+        - :class:`torch.nn.Dropout`
+        - :class:`torch.nn.Flatten`.
+
+    *Role.* Expands the temporal receptive field efficiently to capture rhythms and
+      long-range context after condensation.
+
+    - **Final Classifier — Max-norm linear.**
+
+        - *Operations.*
+        - :class:`braindecode.modules.LinearWithConstraint` maps the flattened
+          vector to the target classes; the max-norm constraint regularizes the readout.
+
+
+    .. rubric:: Convolutional Details
+
+    - **Temporal (where time-domain patterns are learned).**
+      Blocks 1-2 learn the primary filter bank (oscillations/transients), while Blocks 4-5
+      use **dilation** to integrate over longer horizons without extra pooling. The final
+      AvgPool in Block-5 sets the output token rate and helps noise suppression.
+
+    - **Spatial (how electrodes are processed).**
+      A *single* depthwise spatial conv (Block-3) spans the entire electrode set
+      (kernel ``(n_chans, 1)``), producing per-temporal-filter topographies; no cross-filter
+      mixing occurs at this stage, aiding interpretability.
+
+    - **Spectral (how frequency content is captured).**
+      Frequency selectivity emerges from the learned temporal kernels; dilation broadens effective
+      bandwidth coverage by composing multiple scales.
+
+    .. rubric:: Additional Mechanisms
+
+    - **EEGNeX-8,32 naming.** “8,32” indicates *8 temporal filters* and *kernel length 32*,
+      reflecting the paper's ablation path from EEGNet-8,2 toward thicker temporal kernels
+      and a deeper conv stack.
+    - **Max-norm constraints.** Spatial (Block-3) and final linear layers use max-norm
+      regularization—standard in EEG CNNs—to reduce overfitting and encourage stable spatial
+      patterns.
+
+    .. rubric:: Usage and Configuration
+
+    - **Kernel schedule.** Start with the canonical **EEGNeX-8,32** (``filter_1=8``,
+      ``kernel_block_1_2=32``) and keep **Block-3** depth multiplier modest (e.g., 2) to match
+      the paper's “pure conv” profile.
+    - **Pooling vs. dilation.** Use pooling in Blocks 3 and 5 to control compute and variance;
+      increase dilations (Blocks 4-5) to widen temporal context when windows are short.
+    - **Regularization.** Combine dropout (Blocks 3 & 5) with max-norm on spatial and
+      classifier layers; prefer ELU activations for stable training on small EEG datasets.
+
+
+    Notes
+    -----
+    - The braindecode implementation follows the paper's conv-only design with five blocks
+      and reproduces the depthwise spatial step and dilated temporal stack. See the class
+      reference for exact kernel sizes, dilations, and pooling defaults. You can check the
+      original implementation at [EEGNexCode]_.
+
+    .. versionadded:: 1.1
+
 
     Parameters
     ----------
@@ -45,12 +160,6 @@ class EEGNeX(EEGModuleMixin, nn.Module):
     avg_pool_block5 : tuple[int, int], optional
         Pooling size for block 5. Default is (1, 8).
 
-    Notes
-    -----
-    This implementation is not guaranteed to be correct, has not been checked
-    by original authors, only reimplemented from the paper description and
-    source code in tensorflow [EEGNexCode]_.
-
     References
     ----------
     .. [eegnex] Chen, X., Teng, X., Chen, H., Pan, Y., & Geyer, P. (2024).

braindecode/models/summary.csv

@@ -1,41 +1,40 @@
-Model,Paradigm,Type,Freq(Hz),Hyperparameters,#Parameters,get_#Parameters
-ATCNet,General,Classification,250,"n_chans, n_outputs, n_times",113732,"ATCNet(n_chans=22, n_outputs=4, n_times=1000)"
-AttentionBaseNet,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times",3692,"AttentionBaseNet(n_chans=22, n_outputs=4, n_times=1000)"
-BDTCN,Normal/Abnormal,Classification,100,"n_chans, n_outputs, n_times",456502,"BDTCN(n_chans=21, n_outputs=2, n_times=6000, n_blocks=5, n_filters=55, kernel_size=16)"
-BIOT,"Sleep Staging, Epilepsy",Classification,200,"n_chans, n_outputs",3183879,"BIOT(n_chans=2,  n_outputs=5,  n_times=6000)"
-ContraWR,Sleep Staging,"Classification, Embedding",125,"n_chans, n_outputs, sfreq",1160165,"ContraWR(n_chans=2, n_outputs=5, n_times=3750, emb_size=256, sfreq=125)"
-CTNet,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times",26900,"CTNet(n_chans=22, n_outputs=4, n_times=1000, n_filters_time=8, kernel_size=16, heads=2, emb_size=16)"
-Deep4Net,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times",282879,"Deep4Net(n_chans=22, n_outputs=4, n_times=1000)"
-DeepSleepNet,Sleep Staging,Classification,256,"n_chans, n_outputs",24744837,"DeepSleepNet(n_chans=1, n_outputs=5, n_times=7680, sfreq=256)"
-EEGConformer,General,Classification,250,"n_chans, n_outputs, n_times",789572,"EEGConformer(n_chans=22, n_outputs=4, n_times=1000)."
-EEGInceptionERP,"ERP, SSVEP",Classification,128,"n_chans, n_outputs",14926,"EEGInceptionERP(n_chans=8, n_outputs=2, n_times=128, sfreq=128)"
-EEGInceptionMI,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times",558028,"EEGInceptionMI(n_chans=22, n_outputs=4, n_times=1000, n_convs=5, n_filters=12)"
-EEGITNet,Motor Imagery,Classification,125,"n_chans, n_outputs, n_times",5212,"EEGITNet(n_chans=22,  n_outputs=4,  n_times=500)"
-EEGNetv1,General,Classification,128,"n_chans, n_outputs, n_times",3052,"EEGNetv1(n_chans=22,  n_outputs=4,  n_times=512)"
-EEGNetv4,General,Classification,128,"n_chans, n_outputs, n_times",2484,"EEGNetv4(n_chans=22,  n_outputs=4,  n_times=512)"
-EEGNeX,Motor Imagery,Classification,125,"n_chans, n_outputs, n_times",55940,"EEGNeX(n_chans=22,  n_outputs=4,  n_times=500)"
-EEGMiner,Emotion Recognition,Classification,128,"n_chans, n_outputs, n_times, sfreq",7572,"EEGMiner(n_chans=62, n_outputs=2, n_times=2560, sfreq=128)"
-EEGResNet,General,Classification,250,"n_chans, n_outputs, n_times",247484,"EEGResNet(n_chans=22,  n_outputs=4,  n_times=1000)"
-EEGSimpleConv,Motor Imagery,Classification,80,"n_chans, n_outputs, sfreq",730404,"EEGSimpleConv(n_chans=22,  n_outputs=4,  n_times=320, sfreq=80)"
-EEGTCNet,Motor Imagery,Classification,250,"n_chans, n_outputs",4516,"EEGTCNet(n_chans=22,  n_outputs=4,  n_times=1000, kern_length=32)"
-Labram,General,"Classification, Embedding",200,"n_chans, n_outputs, n_times",5866180,"Labram(n_chans=22, n_outputs=4, n_times=1000, sfreq=250)"
-MSVTNet,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times",75494," MSVTNet(n_chans=22,  n_outputs=4,  n_times=1000)"
-SCCNet,Motor Imagery,Classification,125,"n_chans, n_outputs, n_times, sfreq",12070,"SCCNet(n_chans=22,  n_outputs=4,  n_times=1000, sfreq=125)"
-SignalJEPA,"Motor Imagery, ERP, SSVEP",Embedding,128,"n_times, chs_info",3456882,"SignalJEPA(n_times=512, chs_info=Lee2019_MI().get_data(subjects=[1])[1]['0']['1train'].info[""chs""][:62])"
-SignalJEPA_Contextual,"Motor Imagery, ERP, SSVEP",Classification,128,"n_outputs, n_times, chs_info",3459184,"SignalJEPA_Contextual(n_outputs=2, input_window_seconds=4.19, sfreq=128, chs_info=Lee2019_MI().get_data(subjects=[1])[1]['0']['1train'].info[""chs""][:62])"
-SignalJEPA_PostLocal,"Motor Imagery, ERP, SSVEP",Classification,128,"n_chans, n_outputs, n_times",16142,"SignalJEPA_PostLocal(n_chans=62, n_outputs=2, input_window_seconds=4.19, sfreq=128)"
-SignalJEPA_PreLocal,"Motor Imagery, ERP, SSVEP",Classification,128,"n_outputs, n_times, chs_info",16142,"SignalJEPA_PreLocal(n_chans=62, n_outputs=2, input_window_seconds=4.19, sfreq=128)"
-SincShallowNet,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times, sfreq",21892,"SincShallowNet(n_chans=22,  n_outputs=4,  n_times=1000, sfreq=250)"
-ShallowFBCSPNet,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times",46084,"ShallowFBCSPNet(n_chans=22,  n_outputs=4,  n_times=1000, sfreq=250)"
-SleepStagerBlanco2020,Sleep Staging,Classification,100,"n_chans, n_outputs, n_times",2845,"SleepStagerBlanco2020(n_chans=2, n_outputs=5, n_times=3000, sfreq=100)"
-SleepStagerChambon2018,Sleep Staging,Classification,128,"n_chans, n_outputs, n_times, sfreq",5835,"SleepStagerChambon2018(n_chans=2, n_outputs=5, n_times=3840, sfreq=128)"
-SleepStagerEldele2021,Sleep Staging,Classification,100,"n_chans, n_outputs, n_times, sfreq",719925,"SleepStagerEldele2021(n_chans=2, n_outputs=5, n_times=3000, sfreq=100)"
-SPARCNet,Epilepsy,Classification,200,"n_chans, n_outputs, n_times",1141921,"SPARCNet(n_chans=16, n_outputs=6, n_times=2000, sfreq=200)"
-SyncNet,"Emotion Recognition, Alcoholism",Classification,256,"n_chans, n_outputs, n_times",554,"SyncNet(n_chans=62, n_outputs=3, n_times=5120, sfreq=256)"
-TSceptionV1,Emotion Recognition,Classification,256,"n_chans, n_outputs, n_times, sfreq",2187206,"TSceptionV1(n_chans=62, n_outputs=3, n_times=5120, sfreq=256)"
-TIDNet,General,Classification,250,"n_chans, n_outputs, n_times",240404,"TIDNet(n_chans=22,  n_outputs=4,  n_times=1000)"
-USleep,Sleep Staging,Classification,128,"n_chans, n_outputs, n_times, sfreq",2482011,"USleep(n_chans=2, n_outputs=5, n_times=3000, sfreq=100)"
-FBCNet,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times, sfreq",11812,"FCNet(n_chans=22, n_outputs=4, n_times=1000, sfreq=250)"
-FBMSNet,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times, sfreq",16231,"FBMSNet(n_chans=22, n_outputs=4, n_times=1000, sfreq=250)"
-FBLightConvNet,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times, sfreq",6596,"FBLightConvNet(n_chans=22, n_outputs=4, n_times=1000, sfreq=250)"
-IFNet,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times, sfreq",9860,"IFNet(n_chans=22, n_outputs=4, n_times=1000, sfreq=250)"
+Model,Paradigm,Type,Freq(Hz),Hyperparameters,#Parameters,get_#Parameters,Categorization
+ATCNet,General,Classification,250,"n_chans, n_outputs, n_times",113732,"ATCNet(n_chans=22, n_outputs=4, n_times=1000)","Convolution,Recurrent,Small Attention"
+AttentionBaseNet,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times",3692,"AttentionBaseNet(n_chans=22, n_outputs=4, n_times=1000)","Convolution,Small Attention"
+BDTCN,Normal Abnormal,Classification,100,"n_chans, n_outputs, n_times",456502,"BDTCN(n_chans=21, n_outputs=2, n_times=6000, n_blocks=5, n_filters=55, kernel_size=16)","Convolution,Recurrent"
+BIOT,"Sleep Staging, Epilepsy",Classification,200,"n_chans, n_outputs",3183879,"BIOT(n_chans=2,  n_outputs=5,  n_times=6000)","Large Language Model"
+ContraWR,Sleep Staging,"Classification, Embedding",125,"n_chans, n_outputs, sfreq",1160165,"ContraWR(n_chans=2, n_outputs=5, n_times=3750, emb_size=256, sfreq=125)",Convolution
+CTNet,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times",26900,"CTNet(n_chans=22, n_outputs=4, n_times=1000, n_filters_time=8, kernel_size=16, heads=2, emb_size=16)","Convolution,Small Attention"
+Deep4Net,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times",282879,"Deep4Net(n_chans=22, n_outputs=4, n_times=1000)","Convolution"
+DeepSleepNet,Sleep Staging,Classification,256,"n_chans, n_outputs",24744837,"DeepSleepNet(n_chans=1, n_outputs=5, n_times=7680, sfreq=256)","Convolution"
+EEGConformer,General,Classification,250,"n_chans, n_outputs, n_times",789572,"EEGConformer(n_chans=22, n_outputs=4, n_times=1000)","Convolution,Small Attention"
+EEGInceptionERP,"ERP, SSVEP",Classification,128,"n_chans, n_outputs",14926,"EEGInceptionERP(n_chans=8, n_outputs=2, n_times=128, sfreq=128)","Convolution"
+EEGInceptionMI,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times",558028,"EEGInceptionMI(n_chans=22, n_outputs=4, n_times=1000, n_convs=5, n_filters=12)","Convolution"
+EEGITNet,Motor Imagery,Classification,125,"n_chans, n_outputs, n_times",5212,"EEGITNet(n_chans=22,  n_outputs=4,  n_times=500)","Convolution,Recurrent"
+EEGNetv4,General,Classification,128,"n_chans, n_outputs, n_times",2484,"EEGNetv4(n_chans=22,  n_outputs=4,  n_times=512)","Convolution"
+EEGNeX,Motor Imagery,Classification,125,"n_chans, n_outputs, n_times",55940,"EEGNeX(n_chans=22,  n_outputs=4,  n_times=500)","Convolution"
+EEGMiner,Emotion Recognition,Classification,128,"n_chans, n_outputs, n_times, sfreq",7572,"EEGMiner(n_chans=62, n_outputs=2, n_times=2560, sfreq=128)","Convolution"
+EEGResNet,General,Classification,250,"n_chans, n_outputs, n_times",247484,"EEGResNet(n_chans=22,  n_outputs=4,  n_times=1000)","Interpretability"
+EEGSimpleConv,Motor Imagery,Classification,80,"n_chans, n_outputs, sfreq",730404,"EEGSimpleConv(n_chans=22,  n_outputs=4,  n_times=320, sfreq=80)","Convolution"
+EEGTCNet,Motor Imagery,Classification,250,"n_chans, n_outputs",4516,"EEGTCNet(n_chans=22,  n_outputs=4,  n_times=1000, kern_length=32)","Convolution,Recurrent"
+Labram,General,"Classification, Embedding",200,"n_chans, n_outputs, n_times",5866180,"Labram(n_chans=22, n_outputs=4, n_times=1000, sfreq=250)","Convolution,Large Language Model"
+MSVTNet,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times",75494," MSVTNet(n_chans=22,  n_outputs=4,  n_times=1000)","Convolution,Recurrent,Small Attention"
+SCCNet,Motor Imagery,Classification,125,"n_chans, n_outputs, n_times, sfreq",12070,"SCCNet(n_chans=22,  n_outputs=4,  n_times=1000, sfreq=125)","Convolution"
+SignalJEPA,"Motor Imagery, ERP, SSVEP",Embedding,128,"n_times, chs_info",3456882,"SignalJEPA(n_times=512, chs_info=Lee2019_MI().get_data(subjects=[1])[1]['0']['1train'].info[""chs""][:62])","Convolution,Channel,Large Language Model"
+SignalJEPA_Contextual,"Motor Imagery, ERP, SSVEP",Classification,128,"n_outputs, n_times, chs_info",3459184,"SignalJEPA_Contextual(n_outputs=2, input_window_seconds=4.19, sfreq=128, chs_info=Lee2019_MI().get_data(subjects=[1])[1]['0']['1train'].info[""chs""][:62])","Convolution,Channel,Large Language Model"
+SignalJEPA_PostLocal,"Motor Imagery, ERP, SSVEP",Classification,128,"n_chans, n_outputs, n_times",16142,"SignalJEPA_PostLocal(n_chans=62, n_outputs=2, input_window_seconds=4.19, sfreq=128)","Convolution,Channel,Large Language Model"
+SignalJEPA_PreLocal,"Motor Imagery, ERP, SSVEP",Classification,128,"n_outputs, n_times, chs_info",16142,"SignalJEPA_PreLocal(n_chans=62, n_outputs=2, input_window_seconds=4.19, sfreq=128)","Convolution,Channel,Large Language Model"
+SincShallowNet,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times, sfreq",21892,"SincShallowNet(n_chans=22,  n_outputs=4,  n_times=1000, sfreq=250)","Convolution,Interpretability"
+ShallowFBCSPNet,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times",46084,"ShallowFBCSPNet(n_chans=22,  n_outputs=4,  n_times=1000, sfreq=250)","Convolution"
+SleepStagerBlanco2020,Sleep Staging,Classification,100,"n_chans, n_outputs, n_times",2845,"SleepStagerBlanco2020(n_chans=2, n_outputs=5, n_times=3000, sfreq=100)","Convolution"
+SleepStagerChambon2018,Sleep Staging,Classification,128,"n_chans, n_outputs, n_times, sfreq",5835,"SleepStagerChambon2018(n_chans=2, n_outputs=5, n_times=3840, sfreq=128)","Convolution"
+SleepStagerEldele2021,Sleep Staging,Classification,100,"n_chans, n_outputs, n_times, sfreq",719925,"SleepStagerEldele2021(n_chans=2, n_outputs=5, n_times=3000, sfreq=100)","Convolution, Small Attention"
+SPARCNet,Epilepsy,Classification,200,"n_chans, n_outputs, n_times",1141921,"SPARCNet(n_chans=16, n_outputs=6, n_times=2000, sfreq=200)","Convolution"
+SyncNet,"Emotion Recognition, Alcoholism",Classification,256,"n_chans, n_outputs, n_times",554,"SyncNet(n_chans=62, n_outputs=3, n_times=5120, sfreq=256)","Interpretability"
+TSceptionV1,Emotion Recognition,Classification,256,"n_chans, n_outputs, n_times, sfreq",2187206,"TSceptionV1(n_chans=62, n_outputs=3, n_times=5120, sfreq=256)","Convolution"
+TIDNet,General,Classification,250,"n_chans, n_outputs, n_times",240404,"TIDNet(n_chans=22,  n_outputs=4,  n_times=1000)","Convolution"
+USleep,Sleep Staging,Classification,128,"n_chans, n_outputs, n_times, sfreq",2482011,"USleep(n_chans=2, n_outputs=5, n_times=3000, sfreq=100)","Convolution"
+FBCNet,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times, sfreq",11812,"FCNet(n_chans=22, n_outputs=4, n_times=1000, sfreq=250)","Convolution,FilterBank"
+FBMSNet,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times, sfreq",16231,"FBMSNet(n_chans=22, n_outputs=4, n_times=1000, sfreq=250)","Convolution,FilterBank"
+FBLightConvNet,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times, sfreq",6596,"FBLightConvNet(n_chans=22, n_outputs=4, n_times=1000, sfreq=250)","Convolution,FilterBank"
+IFNet,Motor Imagery,Classification,250,"n_chans, n_outputs, n_times, sfreq",9860,"IFNet(n_chans=22, n_outputs=4, n_times=1000, sfreq=250)","Convolution,FilterBank"

braindecode/models/util.py

@@ -55,7 +55,6 @@ models_mandatory_parameters = [
     ("EEGInceptionERP", ["n_chans", "n_outputs", "n_times", "sfreq"], None),
     ("EEGInceptionMI", ["n_chans", "n_outputs", "n_times", "sfreq"], None),
     ("EEGITNet", ["n_chans", "n_outputs", "n_times"], None),
-    ("EEGNetv1", ["n_chans", "n_outputs", "n_times"], None),
     ("EEGNetv4", ["n_chans", "n_outputs", "n_times"], None),
     ("EEGResNet", ["n_chans", "n_outputs", "n_times"], None),
     ("ShallowFBCSPNet", ["n_chans", "n_outputs", "n_times"], None),

braindecode/modules/attention.py

@@ -157,7 +157,8 @@ class FCA(nn.Module):
     ):
         super(FCA, self).__init__()
         mapper_y = [freq_idx]
-        assert in_channels % len(mapper_y) == 0
+        if in_channels % len(mapper_y) != 0:
+            raise ValueError("in_channels must be divisible by number of DCT filters")
 
         self.weight = nn.Parameter(
             self.get_dct_filter(seq_len, mapper_y, in_channels), requires_grad=False
@@ -295,7 +296,8 @@ class ECA(nn.Module):
     def __init__(self, in_channels: int, kernel_size: int):
         super(ECA, self).__init__()
         self.gap = nn.AdaptiveAvgPool2d(1)
-        assert kernel_size % 2 == 1, "kernel size must be odd for same padding"
+        if kernel_size % 2 != 1:
+            raise ValueError("kernel size must be odd for same padding")
         self.conv = nn.Conv1d(
             1, 1, kernel_size=kernel_size, padding=kernel_size // 2, bias=False
         )
@@ -530,7 +532,8 @@ class CBAM(nn.Module):
             nn.ReLU(),
             nn.Conv2d(in_channels // reduction_rate, in_channels, 1, bias=False),
         )
-        assert kernel_size % 2 == 1, "kernel size must be odd for same padding"
+        if kernel_size % 2 != 1:
+            raise ValueError("kernel size must be odd for same padding")
         self.conv = nn.Conv2d(2, 1, (1, kernel_size), padding=(0, kernel_size // 2))
 
     def forward(self, x):

braindecode/modules/convolution.py

@@ -136,7 +136,8 @@ class CombinedConv(nn.Module):
         # Calculate bias terms
         if self.bias_time:
             time_bias = self.conv_time.bias
-            assert time_bias is not None
+            if time_bias is None:
+                raise RuntimeError("conv_time.bias is None despite bias_time=True")
             calculated_bias = (
                 self.conv_spat.weight.squeeze()
                 .sum(-1)
@@ -145,7 +146,8 @@ class CombinedConv(nn.Module):
             )
         if self.bias_spat:
             spat_bias = self.conv_spat.bias
-            assert spat_bias is not None
+            if spat_bias is None:
+                raise RuntimeError("conv_spat.bias is None despite bias_spat=True")
             if calculated_bias is None:
                 calculated_bias = spat_bias
             else:
@@ -190,11 +192,12 @@ class CausalConv1d(nn.Conv1d):
         dilation=1,
         **kwargs,
     ):
-        assert "padding" not in kwargs, (
-            "The padding parameter is controlled internally by "
-            f"{type(self).__name__} class. You should not try to override this"
-            " parameter."
-        )
+        if "padding" in kwargs:
+            raise ValueError(
+                "The padding parameter is controlled internally by "
+                f"{type(self).__name__} class. You should not try to override this"
+                " parameter."
+            )
 
         super().__init__(
             in_channels=in_channels,

braindecode/modules/filter.py

@@ -452,12 +452,14 @@ class GeneralizedGaussianFilter(nn.Module):
         self.inverse_fourier = inverse_fourier
         self.affine_group_delay = affine_group_delay
         self.clamp_f_mean = clamp_f_mean
-        assert out_channels % in_channels == 0, (
-            "out_channels has to be multiple of in_channels"
-        )
-        assert len(f_mean) * in_channels == out_channels
-        assert len(bandwidth) * in_channels == out_channels
-        assert len(shape) * in_channels == out_channels
+        if out_channels % in_channels != 0:
+            raise ValueError("out_channels has to be multiple of in_channels")
+        if len(f_mean) * in_channels != out_channels:
+            raise ValueError("len(f_mean) * in_channels must equal out_channels")
+        if len(bandwidth) * in_channels != out_channels:
+            raise ValueError("len(bandwidth) * in_channels must equal out_channels")
+        if len(shape) * in_channels != out_channels:
+            raise ValueError("len(shape) * in_channels must equal out_channels")
 
         # Range from 0 to half sample rate, normalized
         self.n_range = nn.Parameter(

braindecode/training/scoring.py

@@ -11,7 +11,6 @@ from contextlib import contextmanager
 
 import numpy as np
 import torch
-from mne.utils.check import check_version
 from skorch.callbacks.scoring import EpochScoring
 from skorch.dataset import unpack_data
 from skorch.utils import to_numpy
@@ -370,13 +369,8 @@ class PostEpochTrainScoring(EpochScoring):
             y_preds = []
             y_test = []
             for batch in iterator:
-                batch_X, batch_y = unpack_data(batch)
-                # TODO: remove after skorch 0.10 release
-                if not check_version("skorch", min_version="0.10.1"):
-                    yp = net.evaluation_step(batch_X, training=False)
-                # X, y unpacking has been pushed downstream in skorch 0.10
-                else:
-                    yp = net.evaluation_step(batch, training=False)
+                _, batch_y = unpack_data(batch)
+                yp = net.evaluation_step(batch, training=False)
                 yp = yp.to(device="cpu")
                 y_test.append(self.target_extractor(batch_y))
                 y_preds.append(yp)

braindecode/version.py

@@ -1 +1 @@
-__version__ = "1.2.0.dev169062562"
+__version__ = "1.2.0.dev176281713"

braindecode/visualization/gradients.py

@@ -6,8 +6,13 @@ import numpy as np
 import torch
 from skorch.utils import to_numpy, to_tensor
 
+from braindecode.util import set_random_seeds
 
-def compute_amplitude_gradients(model, dataset, batch_size):
+
+def compute_amplitude_gradients(model, dataset, batch_size, seed=20240205):
+    """Compute amplitude gradients after seeding for reproducibility."""
+    cuda = next(model.parameters()).is_cuda
+    set_random_seeds(seed=seed, cuda=cuda)
     loader = torch.utils.data.DataLoader(
         dataset, batch_size=batch_size, drop_last=False, shuffle=False
     )