資源簡介
ICA程序,利用Infomax算法實現(xiàn),比較實用,建議跳過最優(yōu)成分?jǐn)?shù)估計,直接手動輸入。

代碼片段和文件信息
function?[SAUllinfo]=icaML(XKpardebug_draw)
%?icaML?????:?ICA?by?ML?(Infomax)?with?square?mixing?matrix?and?no?noise.
%
%?function?[SAUllinfo]=icaML(X[K][par])?Independent?component?analysis?(ICA)?using
%???????????????????????????????????????maximum?likelihood?square?mixing?matrix?and?
%???????????????????????????????????????no?noise?[1]?(Infomax).?Source?prior?is?assumed?
%???????????????????????????????????????to?be?p(s)=1/pi*exp(-ln(cosh(s))).?For?optimization
%???????????????????????????????????????the?BFGS?algorithm?is?used?[2].?See?code?for?
%???????????????????????????????????????references.
%???????????????????????????????????????
%???????????????????????????????????????X??:?Mixed?signals
%???????????????????????????????????????K??:?Number?of?source?components.
%???????????????????????????????????????????For?K=0?(default)?number?of?sources?are?equal?to?number?of
%???????????????????????????????????????????observations.
%???????????????????????????????????????????For?K?
%???????????????????????????????????????????dimension.
%???????????????????????????????????????par:??Vector?with?4?elements:
%???????????????????????????????????????????(1)??:??Expected?length?of?initial?step
%???????????????????????????????????????????Stopping?criteria:
%???????????????????????????????????????????(2)??:??Gradient??||g||_inf?<=?par(2)?
%???????????????????????????????????????????(3)??:??Parameter?changes?||dW||_2??<=?par(3)*(par(3)?+?||W||_2)
%???????????????????????????????????????????(4)??:??Maximum?number?of?iterations?
%???????????????????????????????????????????Any?illegal?element?in??opts??is?replaced?by?its
%???????????????????????????????????????????default?value??[1??1e-4*||g(x0)||_inf??1e-8??100]
%???????????????????????????????????????debug_draw?:?Draw?debug?information
%???????????????????????????????????????
%???????????????????????????????????????S??:?Estimated?source?signals?with?variance
%????????????????????????????????????????????scaled?to?one.
%???????????????????????????????????????A??:?Estimated?mixing?matrix
%???????????????????????????????????????U??:?Principal?directions?of?preprocessing?PCA.?
%????????????????????????????????????????????If?K?(the?number?of?sources)?is?equal?to?the?number
%????????????????????????????????????????????of?observations?then?no?PCA?is?performed?and?U=eye(K).??
%???????????????????????????????????????ll?:?Log?likelihood?for?estimated?sources
%???????????????????????????????????????info?:??Performance?information?vector?with?6?elements:
%???????????????????????????????????????????(1:3)??:?final?values?of?[ll??||g||_inf??||dx||_2]?
%???????????????????????????????????????????(4:5)??:?no.?of?iteration?steps?and?evaluations?of?(llg)
%???????????????????????????????????????????(6)????:?1?means?stopped?by?small?gradient
%????????????????????????????????????????????????????2?mea
?屬性????????????大小?????日期????時間???名稱
-----------?---------??----------?-----??----
?????文件???????15496??2003-09-09?13:37??MLcorrected\icaML.m
?????文件????????3347??2006-06-22?14:15??MLcorrected\icaML_bic.m
?????文件?????????845??2002-12-18?14:56??MLcorrected\sc
評論
共有 條評論