Using the Committee Machine Method to Forecasting on the FOREX

In the paper we consider a particular case of a neural network - a committee method. The described approach was first introduced in the paper by Ablow, C. M. and Kaylor, D. J [1]. Then further development of the committee machine methods have been carried out by Yekaterinburg pattern recognition school in the Institute of Mathematics and Mechanics, Russian Academy of Sciences. Nowadays theory of committee machine structures is based on results of Mazurov, Vl. D. and Hachay M. Yu. [2],[3]. In the presentation we describe the generalized p-committee construction and its application to the simplest case with two possible solutions: the course of one currency to another either grows, or falls. The problem of forecasting EUR/USD exchange rate on FOREX is studied by committee machine methods. Objective is to determine a direction of the course in 10 min. interval range. Another issues related to committee methods are considered in the paper [4].