摘要
针对全互连结构是大多数神经联想记忆模型采用的连接方式,在提高模型的性能时必然会以提高模型的复杂度为代价的特点,通过借鉴小世界网络模型对互连结构进行了改进,提出了一种反馈型自联想记忆神经网络模型,应用计算机仿真分析表明,该模型达到了预期效果.
Up to now, most of the associative memory models are the neural networks with full connectivity, and the complexity of computation increases with the performance improvement of the models. On the basis of improvement in connectivity with SWA, a model of recurrent associative memory network is put forward. The simulation results show that the model achieves the expected effect.
出处
《重庆工学院学报》
2007年第23期82-85,共4页
Journal of Chongqing Institute of Technology
关键词
自联想记忆
小世界网络
神经网络
associative memory
small-world network
neural network
