Multi-scale radial basis function neural network

专利名称：Multi-scale radial basis function neural

network

发明人：Georgios Mountrakis申请号：US11441954申请日：20060526公开号：US07577626B1公开日：20090818

专利附图：

摘要：A network architecture of radial basis function neural network system utilizes ablocking layer () to exclude successfully mapped neighborhoods from later nodeinfluence. A signal is inserted into the system at input nodes (I, I, . . . In), which then

promulgates to a non-linear layer (). The non-linear layer () comprises a number of non-linear activation function nodes (). After passing through the non-linear layer (), the signalpasses through the blocking layer () that is comprised of either binary signal blockingnodes, or inverted symmetrical Sigmoidal signal blocking nodes () that act in a binaryfashion. Finally, the signal is weighted by a weighting function (), summed at a summer ()and outputted at (O).

申请人：Georgios Mountrakis

地址：1101 Cumberland Ave. Syracuse NY 13210 US

国籍：US

代理机构：MacMillan, Sobanski & Todd, LLC

更多信息请下载全文后查看