HW 07: PRINCIPLE COMPONENTS ANALYSIS AND LINEAR DISCRIMINANT ANALYSIS
- Replicate the Gaussian data shown
here
and establish that the support regions match the lecture notes.
- For the last case involving an unconstrained covariance matrix,
generate a two-class data problem using means of [5, 0] and [-5,0].
Use a covariance matrix with diagonal element values of [1, 0.5]
and off-diagonal element values of 0.5.
Classify this data using class-independent principle components
analysis. Explain the result.
- Repeat the previous part using linear discriminant analysis.
- Repeat the previous two previous parts using class-dependent
transforms.
- Verify your results using the
pattern recognition applet.