الگوريتم هاي جديد بر اساس اتوماتاي يادگير براي يافتن پارامترهاي ويولت

Learning Automata Based Algorithms For Determination Of Wavelet Parameters

Wevelet are able to approximate functions with different precisions, which makes them appropriate for modeling nonlinear systems. There are number of algorithms such as algorithms based on neural nets (wavenet) and L2 learning algorithm, reported in the literatures for determination of wavelet parameters. In this paper several algorithms based on learning automata are reported. The proposed methods use the searching ability of learning automata to find the wavelet parameters. These methods unlike L2 learning algorithms, which requires high computations (inverting a large matrix) is very fast, and unlike the algorithms based on neural nets have a dynamic structure, that is, the number of base functions need not be known in advance and are determined during the parameters estimation process. The methods are tested on several problems and good results are obtained.