Law of the iterated logarithm for a constant-gain linear stochastic gradient algorithm
J. A. Joslin, A.J. Heunis · Proceedings of the 40th IEEE Conference on Decision and Control (Cat. No.01CH37228) · 2002
We characterize the finite-horizon limiting properties of a constant-gain linear stochastic gradient algorithm, as the adaptation gain tends to zero, in the form of a functional law of the iterated logarithm.