braindecode.models.TIDNet

class braindecode.models.TIDNet(in_chans, n_classes, input_window_samples, s_growth=24, t_filters=32, drop_prob=0.4, pooling=15, temp_layers=2, spat_layers=2, temp_span=0.05, bottleneck=3, summary=- 1)

Thinker Invariance DenseNet model from Kostas et al 2020.

See [TIDNet] for details.

Parameters

n_classesint

Number of classes.

in_chansint

Number of EEG channels.

input_window_samplesint

Number of samples.

s_growthint

DenseNet-style growth factor (added filters per DenseFilter)

t_filtersint

Number of temporal filters.

drop_probfloat

Dropout probability

poolingint

Max temporal pooling (width and stride)

temp_layersint

Number of temporal layers

spat_layersint

Number of DenseFilters

temp_spanfloat

Percentage of input_window_samples that defines the temporal filter length: temp_len = ceil(temp_span * input_window_samples) e.g A value of 0.05 for temp_span with 1500 input_window_samples will yield a temporal filter of length 75.

bottleneckint

Bottleneck factor within Densefilter

summaryint

Output size of AdaptiveAvgPool1D layer. If set to -1, value will be calculated automatically (input_window_samples // pooling).

Notes

Code adapted from: https://github.com/SPOClab-ca/ThinkerInvariance/

References

TIDNet

Kostas, D. & Rudzicz, F. Thinker invariance: enabling deep neural networks for BCI across more people. J. Neural Eng. 17, 056008 (2020). doi: 10.1088/1741-2552/abb7a7.

Methods

forward(x)

Forward pass.

Parameters

x: torch.Tensor

Batch of EEG windows of shape (batch_size, n_channels, n_times).