Local-sparse connection multilayer networks
暂无分享,去创建一个
This paper describes a method for reducing the connection weight number in neural networks. Hopfield networks and perception networks are usually fully connected networks. The weight number increases dramatically with neuron number. We construct a kind of local-sparse connection network and give a simulated result to decide the hidden node number.
[1] Rodney M. Goodman,et al. Recurrent correlation associative memories , 1991, IEEE Trans. Neural Networks.