论文部分内容阅读
Ｌｉｎｓｋｅｒ提出，生物系统对信息的处理可能遵循信息传输最大化准则，并导出了一类神经网络自组织算法——Ｉｎｆｏｍａｘ算法。本文详细讨论了在输入－输出为非线性映射且无输入噪声条件下Ｉｎｆｏｍａｘ算法的性能，证明了在一定条件下Ｉｎｆｏｍａｘ算法与ＩＣＡ算法等价。
Linsker proposed that the processing of information by biological systems may follow the maximization of information transmission and derive a type of neural network self-organizing algorithm - Infomax algorithm. In this paper, the performance of Infomax algorithm under non-linear input-output mapping and no input noise is discussed in detail. It is proved that Infomax algorithm is equivalent to ICA algorithm under certain conditions.