GIN

class torcheeg.models.pyg.GIN(in_channels: int = 4, hid_channels: int = 64, num_classes: int = 3)

A simple but effective graph isomorphism network (GIN) structure from the book of Zhang et al. For more details, please refer to the following information.

• Book: Zhang X, Yao L. Deep Learning for EEG-Based Brain-Computer Interfaces: Representations, Algorithms and Applications[M]. 2021.

• URL: https://www.worldscientific.com/worldscibooks/10.1142/q0282#t=aboutBook

• Related Project: https://github.com/xiangzhang1015/Deep-Learning-for-BCI/blob/master/pythonscripts/4-3_GIN.py

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

from torcheeg.datasets import DEAPDataset
from torcheeg import transforms
from torcheeg.transforms.pyg import ToG
from torcheeg.datasets.constants import SEED_STANDARD_ADJACENCY_MATRIX
from torcheeg.models import GIN
from torch_geometric.data import DataLoader

dataset = DEAPDataset(root_path='./data_preprocessed_python',
                      offline_transform=transforms.BandDifferentialEntropy(),
                      online_transform=ToG(SEED_STANDARD_ADJACENCY_MATRIX),
                      label_transform=transforms.Compose([
                          transforms.Select('valence'),
                          transforms.Binary(5.0),
                      ]))

model = GIN(in_channels=4, hid_channels=64, num_classes=2)

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

Parameters:

• in_channels (int) – The feature dimension of each electrode. (default: 4)

• hid_channels (int) – The number of hidden nodes in the GRU layers and the fully connected layer. (default: 64)

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

forward(data: Batch) → Tensor

Parameters:

data (torch_geometric.data.Batch) – EEG signal representation, the ideal input shape of data.x is [n, 62, 4]. Here, n corresponds to the batch size, 62 corresponds to the number of electrodes, and 4 corresponds to in_channels.

Returns:

the predicted probability that the samples belong to the classes.

Return type:

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