A HME neural network knowledge-increasable model

Abstract

The HME network divides a task into small tasks by the principle of divide and conquer to improve the performance of a single network. This approach often brings simple, elegant and efficient algorithms. By studying the dual manifold architecture for mixtures of neural networks and analyzing the probability of knowledge-increasable model based on information geometry, the paper proposes a new method to achieve the multi-HME model that has knowledge-increasable and structure-extendible ability. The method helps to provide an explanation of the transformation mechanism of the human recognition system and understand the theory of the global architecture of the neural network.