The Community for Technology Leaders
Green Image
Abstract?A definition of the tolerance for the weights of adaptive threshold elements is given which is more relevant to the characteristics of real adaptive components. A learning procedure is then proposed, which employs a sequence of different and increasing values of dead zone. This procedure achieves the determination of the set of maximum tolerance weights and its soundness is shown by systematic tests performed by simulation on an IBM 7094 digital computer on all the binary functions of 5 variables.
Index terms?Adaptive systems, adaptive threshold element, learning procedures, learning time, optimum tolerance learning.
S.P. Colussi, G.V. Pallottino, "An Approach to Optimum Tolerance Adaptive Threshold Elements", IEEE Transactions on Computers, vol. 17, no. , pp. 684-691, July 1968, doi:10.1109/TC.1968.227418
97 ms
(Ver 3.1 (10032016))