LinearDiscrimination线性判别式课程.pptVIP

* * ir.dlut.edu.cn * * ir.dlut.edu.cn 1 2 3 4 * * ir.dlut.edu.cn * * ir.dlut.edu.cn * * ir.dlut.edu.cn * * ir.dlut.edu.cn * * ir.dlut.edu.cn * * ir.dlut.edu.cn * * ir.dlut.edu.cn * * ir.dlut.edu.cn * * ir.dlut.edu.cn * * ir.dlut.edu.cn 1 3 2 * * ir.dlut.edu.cn * * ir.dlut.edu.cn * * ir.dlut.edu.cn Linear Discrimination outline 10.7 Logistic Discrimination 10.7.1 Two Classes 10.7.2 Multiple Classes 10.8 Discrimination by Regression Gradient Descent wt wt+1 η E (wt) E (wt+1) Two Classes We do not model the class-conditional densities, p(x/Ci), but rather their ratio. Two classes and assum