2. Let a (0,1). Define the a-cost-sensitive risk of a classifier h to be Re(h) = Exy [(1 a)1{Y=1,h(x)=0} + ol{Y-0,4(X)=1}]. 5 This assigns different weights to "false positives" and "false negatives. Also define the associated Bayes risk to be R* := inff R. (f), where the infimum is over all classifiers. Determine the Bayes classifier for this risk, and prove an analogue of Theorem 1. You can check your result using the fact that 2R,(f) = R() when a = 1. 2. Let a (0,1). Define the a-cost-sensitive risk of a classifier h to be Re(h) = Exy [(1 a)1{Y=1,h(x)=0} + ol{Y-0,4(X)=1}]. 5 This assigns different weights to "false positives" and "false negatives. Also define the associated Bayes risk to be R* := inff R. (f), where the infimum is over all classifiers. Determine the Bayes classifier for this risk, and prove an analogue of Theorem 1. You can check your result using the fact that 2R,(f) = R() when a = 1