专利内容由知识产权出版社提供
专利名称：Artificial neural network method and
architecture adaptive signal filtering
发明人：Mehmet E. Ulug申请号：US08/290672申请日：19940815公开号：US05467428A公开日：19951114
摘要：An architecture and data processing method for a neural network that canapproximate any mapping function between the input and output vectors without the useof hidden layers. The data processing is done at the sibling nodes (second row). It isbased on the orthogonal expansion of the functions that map the input vector to theoutput vector. Because the nodes of the second row are simply data processing stations,they remain passive during training. As a result the system is basically a single- layerlinear network with a filter at its entrance. Because of this it is free from the problems oflocal minima. The invention also includes a method that reduces the sum of the square oferrors over all the output nodes to zero (0.000000) in fewer than ten cycles. This is doneby initialization of the synaptic links with the coefficients of the orthogonal expansion.This feature makes it possible to design a computer chip which can perform the trainingprocess in real time. Similarly, the ability to train in real time allows the system to retrainitself and improve its performance while executing its normal testing functions. Becausethe second synaptic link values represent the frequency spectrum of the signal appearingon a given output node, by training the ONN with all N sibling nodes and using only someof them in testing, we can create a low pass, a high pass or a band pass filter.
申请人：ULUG; MEHMET E.
代理机构：Malin, Haley, DiMaggio & Crosby
更多信息请下载全文后查看
