Develop an incremental, gradient descent weight change rule for a single sigmoid unit in which the error function is eld-fi, where d is the desired output, f=1/(1-e), s=X W, X is an (n + 1)-dimensional input vector, and W is the (n + 1)-dimensional weight vector being changed.