الفهرس | يوجد فقط 14 صفحة متاحة للعرض العام |
المستخلص Stochastic approximation is a procedure for finding the root of an equation, or the solution of a system of equations, when the values of the respective functions can only be observed (measured) with experimental errors, at recursively determined points. The procedure IS nonparametric with respect to the type of the function as well as with respect to the distribution of the experimental error (although an information about this distribution, if available, can be made use of). The procedure for finding the root is called the Robbins-Monro procedure. Stochastic approximation and related topics, as recursive parameter estimation, or up and down experimentation, belong to the so-called on-line methods, where the solution of the problem takes place in real time, in the course of the process of observations. Typically, observations follow each other after fixed time-intervals; the point of the next observation is corrected according to the result of the preceding one. |