Optimal parameter estimation algorithms in identification problems

Identification algorithms, optimal and optimal on a class in terms of asymptotic Tate of convergence are described. Algorithms optimal on a class are found to be related with robust algorithms. The possibility is discussed of accelerating the obtaining of estimates at initial steps by incorporating a priori information on the optimal solution. Such accelerating algorithms are shown to regularize the estimates. Papers on development of algorithms, optimal and optimal on a class, are surveyed.