A sequence of binary statistical experiments generated by a sample of random variables with persistent linear regression is studied. A stochastic approximation for a sequence of statistical experiments is constructed in terms of an autoregressive process with normal noise. For a sequence of exponential statistical experiments, a stochastic approximation is constructed, as well, with the help of an exponential normal autoregressive process.
27 February 2016 in Journal of Mathematical Sciences
A sequence of binary stochastic experiments with persistent nonlinear regression is considered. The regression is defined as a product of a linear directed force and nonlinear term changing the directed force in a neighborhood of boundary points. A stochastic approximation for the sequence of stationary experiments is constructed with the help of an autoregressive process with normal perturbation. A stochastic approximation of the sequence of exponential stochastic experiments is also constructed with the help of an exponential autoregressive process.
30 August 2016 in Statistics, Optimization & Information Computing