Acceleration by prediction for error bac
β
Arihiro Kanda; Satoshi Fujita; Tadashi Ae
π
Article
π
1994
π
John Wiley and Sons
π
English
β 687 KB
## Abstract This paper proposes a speed improvement of the error back propagation algorithm, which is employed widely in the multilayered neural network, by introducing the prediction. The idea is to realize a larger acceleration by introducing the differential factor for the moment terms in the er