Abstract:
An overview of optimal and nearly optimal sequential hypothesis testing and changepoint (disorder) detection methods is given. We consider a variety of scenarios and problem settings: simple and composite hypotheses; two- and multi-decision problems; iid and general non-iid models; Bayesian and non-Bayesian (minimax and other) optimality criteria. Classical results as well as more complicated sequential testing and disorder detection problems are discussed. New challenges and open problems are also addressed.