To train the convolutional network, we must perform both forward- and backward-propagation, as shown in the following example:
public virtual void Train(Volume<T> x, Volume<T> y)
{
Forward(x);
Backward(y);
}
The following screenshot illustrates our training in progress: