DEPENDENT NOISE FOR STOCHASTIC ALGORITHMS
Paul Doukhan
Odile Brandière
Abstract: We introduce different ways of being dependent for the input noise of stochastic
algorithms. We are aimed to prove that such innovations allow to use the ODE (ordinary
differential equation) method. Illustrations to the linear regression frame and to the law of
large numbers for triangular arrays of weighted dependent random variables are also given.
2000 AMS Mathematics Subject Classification: Primary 62L20; Secondary
62J05.
Key words and phrases: Stochastic approximation, ordinary differential equations,
dependent noise, linear regression.