Caesar AI Atlas

Backpropagation Through Time

Caesar AI Atlas Definition

Backpropagation through time is a gradient-based technique for training recurrent neural networks by unfolding a sequence across time steps. It computes how errors at later time steps depend on earlier states and uses those gradients to update shared model parameters.

Other Definitions

Backpropagation Through Time Source

A gradient-based technique for training certain types of recurrent neural networks, such as Elman networks. The algorithm was independently derived by numerous researchers.

Related Terms