Formulas for Quadratic Discriminant Analysis

Dave Jones wrote:
"I've developed some routines in Matlab to implement linear
discriminant analysis and would like to add a quadratic version. I'd
appreciate any pointers to tracking down the proper formulas,
preferably in standard matrix notation."

Greg Heath responded:
"A lazy approach is to augment the n-dimensional input vectors with the
n squares and n*(n-1)/2 cross products to end up with an m-dimensional
input vector with m = n*(n+3)/2."


Is this (expanded inputs to LDA) exactly the same as QDA? I assumed so
until I read this:

http://www.stat.psu.edu/~jiali/course/stat597e/notes2/lda.pdf

Comments, anyone?


-Will Dwinnell
http://will.dwinnell.com

.