Neural network in which inputs from a current layer are fed into a previous layer.

ETSI GR ENI 004 ANN that is connected between nodes form a directed graph across a temporal sequence.

Note: RNNs contain cyclic connections that make them effective in storing history and state. This is done by using feedback loops in the processing of an output.

ISO/IEC DIS 22989 neural network (3.3.7) in which outputs from both the previous layer and the previous processing step are fed into the current layer