Bayesian Classifier for Normally Distributed classes

Reference

Click here for Lecture PPT: TAMU

Introduction

Bayes classifier for Normally distributed classes.

Case 1: If the same covariance for each class: Linear Classifier

Case 2: If different covariance: Quadratic Classifier

When the likelihood is expressed as a normal distribution:

Linear Classifier: Linear Discriminant Analysis (LDA)

The boundary function can be expressed as a linear function $g_i(x) = \bf{w_i^Tx } + b_i$

Quadratic Classifier

Example