一种双向映射神经网络模型

A Neural Network Model with Bidirectional Mapping

在传统BP神经网络的基础上,提出了一种双向映射网络,证明了该网络的存在性,并验证了该网络的可实现性.该双向映射网络为某些问题的研究开辟了一条新思路.

A bidirectional mapping network is proposed based on the conventional BP neural network. The structure of the network proposed is similar to that of the BP network. The BP network can realize mapping from Rn to Rm while the bidirectional mapping network can realize two-way mapping between Rn and Rm. For mapping among many samples, one-to-one mapping between them should be observed. The existence and feasibility of the network proposed are verified.