braindecode.visualization.saliency#
- braindecode.visualization.saliency(model, x, target)[source]#
Vanilla saliency:
|d y[target] / d x|.- Parameters:
model (torch.nn.Module)
x (torch.Tensor of shape
(batch, n_chans, n_times))target (torch.Tensor of shape
(batch,)) – Class indices.
- Return type:
torch.Tensor of the same shape as
x.
