FBCNet

class torcheeg.models.FBCNet(num_electrodes: int = 20, chunk_size: int = 1000, in_channels: int = 9, num_S: int = 32, num_classes: int = 2, temporal: str = 'LogVarLayer', stride_factor: int = 4, weight_norm: bool = True)

An Efficient Multi-view Convolutional Neural Network for Brain-Computer Interface. For more details, please refer to the following information.

• Paper: Mane R, Chew E, Chua K, et al. FBCNet: A multi-view convolutional neural network for brain-computer interface[J]. arXiv preprint arXiv:2104.01233, 2021.

• URL: https://arxiv.org/abs/2104.01233

• Related Project: https://github.com/ravikiran-mane/FBCNet

Below is a recommended suite for use in emotion recognition tasks:

from torcheeg.datasets import DEAPDataset
from torcheeg import transforms
from torcheeg.models import FBCNet
from torch.utils.data import DataLoader

dataset = DEAPDataset(root_path='./data_preprocessed_python',
                      chunk_size=512,
                      num_baseline=1,
                      baseline_chunk_size=512,
                      offline_transform=transforms.BandSignal(),
                      online_transform=transforms.ToTensor(),
                      label_transform=transforms.Compose([
                          transforms.Select('valence'),
                          transforms.Binary(5.0),
                      ]))

model = FBCNet(num_classes=2,
               num_electrodes=32,
               chunk_size=512,
               in_channels=4,
               num_S=32)

x, y = next(iter(DataLoader(dataset, batch_size=64)))
model(x)

Parameters:

• num_electrodes (int) – The number of electrodes. (default: 28)

• chunk_size (int) – Number of data points included in each EEG chunk. (default: 1000)

• in_channels (int) – The number of channels of the signal corresponding to each electrode. If the original signal is used as input, in_channels is set to 1; if the original signal is split into multiple sub-bands, in_channels is set to the number of bands. (default: 9)

• num_S (int) – The number of spatial convolution block. (default: 32)

• num_classes (int) – The number of classes to predict. (default: 2)

• temporal (str) – The temporal layer used, with options including VarLayer, StdLayer, LogVarLayer, MeanLayer, and MaxLayer, used to compute statistics using different techniques in the temporal dimension. (default: LogVarLayer)

• stride_factor (int) – The stride factor. (default: 4)

• weight_norm (bool) – Whether to use weight renormalization technique in Conv2dWithConstraint. (default: True)

forward(x: Tensor) → Tensor

Parameters:

x (torch.Tensor) – EEG signal representation, the ideal input shape is [n, in_channel, num_electrodes, chunk_size]. Here, n corresponds to the batch size

Returns:

the predicted probability that the samples belong to the classes.

Return type:

torch.Tensor[number of sample, number of classes]