In this problem, we revisit the adaptive control of the learning rate discussed in Section 4.10. The issue of interest is to demonstrate that the asymptotic behavior of the learn ingrate parameter η (x) in Eq. (4.85) does not converge to zero as the number of iterations increases to infinity.
