✦ LIBER ✦

Back-propagation algorithm which varies the number of hidden units

✍ Scribed by Yoshio Hirose; Koichi Yamashita; Shimpei Hijiya

Publisher: Elsevier Science
Year: 1991
Tongue: English
Weight: 495 KB
Volume: 4
Category: Article
ISSN: 0893-6080
DOI: 10.1016/0893-6080(91)90032-z

No coin nor oath required. For personal study only.

✦ Synopsis

gTds report presents a back-propagation algorithm t/tat varies the number of hidden units. 77ti,s" algorithm is expected to escape local minima and makes it no longer necessary to decide the number of hidden ttnits. We tested this algorithm on two examples. One was exclusive-OR learning and the other was 8 x 8 dot alphanumeric fimt learning. In both examples, the probability ~ff'beeoming trapped in local minima was reduced. b)irthermore, in alphanumeric font learning, the network converged two to three times ]aster than con t,entional hack -propagation.