【摘要】 自从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（Ｉｎｆｏｍａｘ）以来，Ｉｎｆｏｍａｘ算法已被成功地应用于很多领域。作者在“非线性Ｉｎｆｏｒｍａｘ自组织算法的盲源分离机理”一文中研究了输入－输出为非线性映射、无输入噪声时Ｉｎｆｏｍａｘ算法的性能，本文在以前工作的基础上，详细讨论了在输入－输出为线性映射、输入信号和噪声为高斯分布条件下Ｉｎｆｏｍａｘ算法的性能，并与传统的ＰＣＡ算法作了比较分析。更多还原

【Abstract】 Since it is suggested in Linsker that a network which develops to maximize the mutual information between its output and the signal portion of its input may provide a model of biological neural systems, Infomax algorithm is successfully applied in many fields. In “Blind source separation mechanism of the non linear informax self organizing algorithm” paper, the authors investigate the Infomax algorithm in detail, focusing on the case of nonlinear transfer functions and the absence of input noise. As a supplement of “Blind source separation mechanism of the non linear informax self organizing algorithm” paper, in this paper, the authors analyze the Infomax algorithm for the case in which the input output mapping is linear and the input signal and noise are multivariate Gaussian, the equivalence of Infomax algorithm and PCA algorithm in some cases is also showed.更多还原