Details of Lecture 21, ECE662 Spring 2010
(see lecture notes)
In Lecture 21, we continued our discussion of the Perceptron criterion function. We discussed the batch method and the online method for minimizing this criterion. We stated that both methods are proved to converge if the data is linearly separable. In order to address the case where the data in not linearly separable, we proposed the use of a linear least-squares procedure, which is linked to Fisher's linear discriminant. We then introduced Fisher's linear discriminant.
The lecture notes have been typed by a student and can be found here
Previous: Lecture 20 Next: Lecture 22
