A new algorithm to improve the generalization capability of feedforward neural network through network inversion

The combination of input vector tuning with traditional weight tuning of back-propagation algorithm resulted in a new algorithm on the basis of network inversion (IBP). In the process neural network learning and training were estimated from an angle of network inversion along with the objective to effectively improve the learning performance of feed-forward neural network. Quite a few simulation experiments served to make comparison between IBP algorithm, the BP algorithm with momentum term, and a newly published algorithm using weight updating method to speeds up convergence. The experimental results show that this new algorithm has the dual merits of quick training speed and good generalization capability. It proves to be a very effective learning method.