PURPOSE: To provide the analysis method for a neural network after learning, which fits for the tuning of the neural network and the analysis of data, on the analysis processing method of the neural network.
CONSTITUTION: A learning processor 2 executes the learning processing of a neuron computer 1 so that prescribed output data can be obtained for input data. Then, an analysis device 3 inputs, for instance, the average value of input data to the learnt neuron computer 1 obtains corresponding first output data, obtains the variance of input data, inputs a value obtained by changing the average value by the variance to the learnt neuron computer 1 obtains corresponding second output data and measures the change of the second output data from the first output data.