Extending Taguchi's Approach to Adaptive Stochastic Control: some Theoretical Results