Linear Discriminant Analysis (LDA) for dimensionality reduction seeks projection vectors w that do which of the following?

Question

Answer

Find projections that preserve pairwise distances between all data points.

Answer

Minimize the within-class scatter S_W only, ignoring between-class distances.

Answer

Maximize the total variance of projected data regardless of class labels.

Answer

Maximize the Fisher criterion J(w) = (w-transpose * S_B * w) / (w-transpose * S_W * w), where S_B is between-class scatter and S_W is within-class scatter.

Crowdly

Linear Discriminant Analysis (LDA) for dimensionality reduction seeks projection...

Want instant access to all verified answers on moodle.spit.ac.in?