BP(λ): Online Learning via Synthetic Gradients

Joseph Pemberton, Rui Ponte Costa

Training recurrent neural networks typically relies on backpropagation through time (BPTT). BPTT depends on forward and backward passes to be completed, rendering the network locke…