Expert Systems of Real Time as Key Tendency of Artificial Intelligence in Tax Administration

Abstract

The problem of neural network mathematical modeling of economic objects and systems, including the objects of the tax control and taxation, attracts scientists. In practice, the economic systems operate under conditions of uncertainty, which makes the results of strict mathematical calculation ineffective for solving the task. Despite numerous developments in the field of neural network modeling for stochastic objects with very noisy data, in particular, tax control facilities, the methods and principles of adequate and sufficiently precise neural network modeling have not been fully developed. Our experience shows that the “frontal” building of effective neural network modeling on the basis of standard neuropackage is impossible without the development of the fundamental concepts and the use of preliminary data processing.