Recurrent Backpropagation

While the backpropagation algorithm originally presented by McClelland and Rummelhart was only valid for feedforward networks (that is, networks without recurrent connections), Pineda (1987) and Almeida (1987) realized that backpropagation is a special case of a more general gradient descent algorithm. This more general error computation, called recurrent backpropagation, provides equations that allow error computation even in the case of a network with feedback loops.

Back to the Table of Content

Back to the Previous subject

To the Next Subject