机构地区: 华南理工大学电子与信息学院
出 处: 《武汉大学学报(理学版)》 2003年第1期117-120,共4页
摘 要: 采用神经网络方法 ,在对于混合矩阵的秩知道一些先验信息的条件下 ,对于检测信号少于盲信号的问题 ,从理论上给出了提取盲信号的充分与必要条件 ,根据这个条件 ,从检测信号中能够提取出某些盲信号 By using neural network method, under the condition of that some information about rank of the mixture matrix is known, we present a condition of extraction. By using the condition, one can extract some blind signals from the observed signals, when the number of observed signals less than the number of blind signals. A theoretical necessary and sufficient condition of extractability is given. The structure of the neural network and the method of extraction are proposed in detail.