Declaration Tax Audit Using Chaos Theory (Neural Networks)

Legal person's Declaration Tax audit or corporate income tax returns for audit computer to help classify the exact percentage difference of profit before tax compared with the pre express income tax return is expected. The number of floor tax affairs is able to make with priorities based on risk, they should consider. The aim of the research, applied research and the nature of the research is event or the last study. The cognitive theory adopts a model that attempted to measure the degree of chaos and turbulence variables are investigated. According to research findings from the ten companies were listening the Tehran securities and stock exchange, during the period 1989-2010 showed that the results of research hypotheses. Variables Using Neural Network projections were made forecasting before. Next, for a variable earnings before income tax predicting for the manner of neurons artificial nose to network with five input neurons (the two major variables: the rate of inflation and rate of Economic growth and 3 small variables incomes, the total of costs and earning before deduction of tax reviewed company) was. In siding and out siding the error forecasting a sample of firms shows that forecasting to Artificial neural network with one input neuron has better result than expected with five input neurons. In the study between artificial neural network methods with two classical methods of linear simple regression, and regression to explanation, a comparison was done. The results showed that the hypothesis was accepted. And artificial neural network method for predicting is efficient and accurate tool