oh

• whitening data why ${\Sigma }^{-1/2}$
• Decorrelating. right mulitply by V why rotation
• what’s axes aligned
• isocontours when $\Sigma$ is PSD
• pool covariance. mean or weighted with prior probabilities.
• lda and qda parameters d + 1 vs d(d+3)/ 2 + 1
• design matrix confusion $x_1^T$

$$\left[\begin{array}{cc}{x}_1& x_2\\ x_3& x_4\\ x_5& x_6\end{array}\right]$$

row= samples columns = features / dimensions

GDA