You are to have your face digitized for a face database. Pick either a smiling or frowning expression. Next use Matlab and the method described in Sec4.5 of the text to pick the largest eigenvalues and their corresponding eigenvectors from the data.
In the space of eigenvectors, average the data from female and male faces to get female and male prototypes. Test your data by the leaving-one-out method. Leave one sample out and them see if it is classified correctly by classifying it as the nearest prototype.
Do the same for smiling versus frowning faces.
Face images are available Here.
Simulate the "deer-moose-wolves" problem using both a linear and nonlinear model. Investigate the behavior as a function of distance from the principal equilibrium point.
-------------------------------
ALSO: Do problem 7- chapter 4 by hand showing all steps.