11/17 Notes¶
HW 3 due December 1st
Project 3 due December 16th
Quiz 3 sometime
Can take quiz 4 if you want, takes top 3 quizzes
PCA¶
Face images can be denoted as a point in high dimentional space (d x 1)
Each eigenvector is known as a PC
Eigenvectors pivot in center of image points
face image = f
f = weight_1 x eigenvector_1 + weight_2 x eigenvector_2 + ... + weight_d x eigenvector_d
See notes from tuesday
LDA¶
Each subject has a label, PCA does not do this
In LDA you want the subjects seperated
The only difference between PCA and LDA is how we compute Covarience matrix (calculate 2 of them in LDA)
Sw and SB are the 2 Cov matricies