Revision of the observed data based on neural networks

It is a very important task for graduation of the observed data in composing the mortality table. Currently, there are different limitations influencing the methods of graduation. In this paper, the revision of data is completed by means of neural networks. This paper mainly uses a three -to -eight -to -one's model of neural network for graduation of the observed data. As it is illustrated in the paper, the method is right, and can be generalized to the graduation of observed data of other schemes.