分析了LCNN的約束項(xiàng)的物理意義，認(rèn)為約束項(xiàng)λ是有監(jiān)督學(xué)習(xí)的加速度，使得整個(gè)算法無論是學(xué)習(xí)矩陣還是獨(dú)立成分的求解效率都可達(dá)到O(n)。針對不同的λ和源信號、觀測信號對的不同特性，提出了4種快速LCNN算法，分析了靜態(tài)圖像獨(dú)立成分分析模型，建立了單像素內(nèi)的獨(dú)立模型，并總結(jié)了其優(yōu)勢。
關(guān) 鍵 詞 Helmholtz自由能; 獨(dú)立成分分析(ICA); Lagrange約束神經(jīng)網(wǎng)絡(luò); 單像素內(nèi)獨(dú)立成分模型

Abstract The traditional independent component analysis is based on statistics mean of all aposteriori data and dismissed geometry. The classical lagrange constraint neural network (LCNN) employs Helmholtz free energy to unify supervised and unsupervised learning, and uses aprior and multi-sensing to solve independent components in one pixel, whose geometrical grain reached single pixel. The operations among pixels can be completely run parallelly. However, the constraints of classical LCNN bring ill-conditional matrix. In this paper, the real meaning of the constraints λ is discussed, four fast LCNN algorithms are proposed, the independent component (IC) models of still image is analyzed, and a new sub-pixel IC model is presented.
Key words Helmholtz free energy; independent component; lagrange constraint neural network; sub-pixel independent component model