A fuzzy-controlled delta-bar-delta learning rule

In classic backpropagation nets, as introduced by Rumelhart et al. (1986), the weights are modified according to the method of steepest descent. The goal of this weight modification is to minimise the error in net-outputs for a given training set. Basing upon Jacobs' work (1988), we point out drawbacks of steepest descent and suggest improvements on it. These yield a backpropagation net, which adjusts its weights according to a parallel coordinate descent method, whose parameters are being fuzzy-controlled.<
>