PyG GCNConv Layer Shape Mismatch Error: Reshaping Input Tensor for Compatibility

The error occurs in the forward pass of the GCNConv layer in the GCN model. The input tensor x has a shape of (120, 40) and the weight matrix has a shape of (1600, 8), which cannot be multiplied together.

To fix this error, you need to adjust the shape of the input tensor x to match the shape expected by the GCNConv layer. This can be done by reshaping the tensor to have a shape of (batch_size * num_nodes, num_node_features) before passing it to the GCNConv layer.

Here's the updated code for the forward method in the GCN model:

class GCN(torch.nn.Module):
    def __init__(self, num_node_features, num_classes):
        super(GCN, self).__init__()
        self.conv1 = GCNConv(num_node_features, 8)
        self.conv2 = GCNConv(8, 16)
        self.conv3 = GCNConv(16, num_classes)
        self.cnn = CNN()  # 添加CNN模型

    def forward(self, data):
        x, edge_index = data.x, data.edge_index

        # 执行卷积神经网络前向传播
        features = []
        for i in range(x.shape[0]):
            features.append(self.cnn(x[i].unsqueeze(0)))
        features = torch.cat(features, dim=0).view(x.shape[0], -1)

        # 替换PyG数据集的节点特征x
        data.x = features

        # Reshape the input tensor x
        x = x.view(-1, x.shape[-1])

        x = self.conv1(x, edge_index)
        x = F.relu(x)
        x = self.conv2(x, edge_index)
        x = F.relu(x)
        x = F.dropout(x, training=self.training)
        x = self.conv3(x, edge_index)

        return x

With this update, the error should be resolved.