Recurrent Layer

This is a special layer that allows the activation calculation to take into account the weights of previous activations to the same node. This process of feeding previous inputs is called Unrolling