[english][all] (Ъ`NG媩åH^媩PBsI)
Slides
pGڭ̰]bwƶAC@ƳOWߪAb]UAC@ƪPDFiH²ƦƦbC@PDFnCyܻAڭ̥iHXC@ƦbC@ӺשҹPDFAMNP@ƪƭPDFisANiHoo@ƪPDFCڭ̪]iHϥμƾǦlӪܦpUG
p(X|C) = P(X1|C)P(X2|C) ... P(Xd|C) 䤤 X = [X1, X2, ..., Xd] O@ӯSxVqA C N@ӯSwOCoӰ]ݨӦGLjA@ڥ@ɪƦGLk]AѦ]Ҳͪ¨]naive Bayes classifierA² NBC^oO۷ΩʡAѮį``鵹䥦ѾC
b갵WAڭ̳q`]@ƩҹPDFOvKר禡AbpUANBCBJiHpUG
- ]C@OƧO d vKרơ]Gaussian probability density function^Ҳ͡GG
gi(x, m, S) = (2p)-d/2*det(S)-0.5*exp[-(x-m)TS-1(x-m)/2] 䤤 m OvKרƪVq]Mean vector^AS hO@ܲx}]Covariance matrix^Aڭ̥iHھ MLEAͳ̨ΪVq m M@ܲx} SC- YݭnAiHC@ӰvKרƭW@v wiC
- bڶiɡAwi*gi(x, m, S) VjAh x ݩO i iʴNVC
bڶiBɡAڭ̳q`hp wi*gi(x, m, S) AӬOp log(wi*gi(x, m, S)) = log(wi) + log(gi(x, m, S))AHK}pƮɥioͪغذD]pTפBpӮɡ^Alog(gi(x, m, S)) pUG
log[p(ci)g(x, mi, Si)] = log(p(ci)) - (d*log(2p) + log|Si|)/2 - (x-mi)TSi-1(x-mi)/2 The decision boundary between class i and j is represented by the following trajectory:p(ci)g(x, mi, Si) = p(cj)g(x, mj, Sj). Taking the logrithm of both sides, we havelog(p(ci)) - (d*log(2p) + log|Si|)/2 - (x-mi)TSi-1(x-mi)/2 = log(p(cj)) - (d*log(2p) + log|S|j)/2 - (x-mj)TSj-1(x-mj)/2 After simplification, we have the decision boundary as the following equation:(x-mj)TSj-1(x-mj) - (x-mi)TSi-1(x-mi) = log{[|S|i p2(ci)]/[|S|j p2(cj)]} where the right-hand side is a constant. Since both (x-mj)TSj-1(x-mj) and (x-mi)TSi-1(x-mi) are quadratic, the above equation represents a decision boundary of the quadratic form in the d-dimensional feature space.ҦpApGϥ NBC ӹ IRIS ƪĤTβĥ|iAiϥΤUCdҵ{G
WϨqXIAHΤ~I]ee^CSOݭn`NOAbWz{XAڭ̥Ψ classWeightAoO@ӦVqAΨӫwC@OvAq`ذkG
- pGnz]Ш`^AviH]wOC@OƭӼơC]pCOƭӼơAi dsClassSize.m ӹFC^
- pGCOƥXuvۮtjAڭ̥iNC@Ov]w 1C
ڭ̥iHeXCOΨCӺת@PDFơAHΨơAШUCdҡG
ڭ̤]iHNCOPDFƥHTe{AõeX䵥uAШUCdҡG
ھڳoǰKרơAڭ̴NiHeXCOɡApUG
Data Clustering and Pattern Recognition (ƤsP˦{)