Non-asymptoticAnalysisofBiasedAdaptive StochasticApproximation

While these algorithms havebeen extensively studied, both theoretically and practically, see, e.g., [10], many questions remain open. In particular, most results are based onthecase where theestimatord V isunbiased.