Hetero Associative Memory Hopfield NN Model, Learning Algorithms and Performance
异联想记忆Hopfield神经网络的模型、算法及性能

This paper analyzes the performance of associative memory neural network. On the basis of bidirectional associative memory theory, a hetero associative memory Hopfield NN model is built to be used for pattern recognition. For the shortcoming of lack of storage capacity of Hopfield NN this paper improves traditional learning algorithm, proposes pseudo converse and generalized converse learning algorithm, and therefore the capacity of Hopfield NN to store training samples is increased. The simulation and results show that the study of this paper solves the key of Hopfield NN being applied in hetero associative memory pattern recognition.