0:15:17: Q1. Do you know lagrange multiplier technique? ANS:
0:34:39: Q2 Why H(K) >= H(K|C)? ANS: conditioning reduces entropy
0:37:25: Q3 Why H(M|C)=H(M)? ANS: M and C are independent
0:47:50: Q4 Should we consider an individual as vampire if he has a shadow? ANS: No
1:06:37: Q5. Given the size order of the conditional entropy (H(V|S) < H(V|G) < H(V|C) < H(V|A)), which feature should we use to make a decision? ANS: Shadow
1:11:22: Q6. what do you expect H(V|S=Y) to be for the given dataset with 8 indviduals? ANS: 0
1:13:52: Q7. what would H(V|S=Y) be if we have 2999 non-vampire and 1 vampire for those with shadow? ANS: 0.0043
1:39:59: Q8. What is a reasonable approximate of P(S=Y) for this dataset with 8 individuals? ANS: 3/8
1:41:45: Q9 What is H(V|C) given this 8-sample dataset? ANS: 0.69
1:50:07: Q10 what is H(V|S=?)? ANS: 1
1:53:07: Q11 What is H(V|C,S=?)? ANS: .5
2:01:37: Q12, what is H(V|A,S=?)? ANS: 1
2:41:43: Q13, what do you expect max_p(x) H(Y) would be? ANS: 1