backpropagation through time (BPTT) is the standard algorithm for training
recurrent neural networks (RNNs), which requires separate simulation phases for
the forward and backward passes for inference and learnin
介绍了 Sparse n-step Approximation (SnAp) 来优化 Real Time Recurrent Learning (RTRL) 的影响矩阵,在保持计算成本低的同时提高了网络的学习速度,特别是在稀疏网络时,n=2 时的 SnAp 能够优于 backpropagation in terms of learning speed when updates are done online。