I love this video, it's much better than the mechanical approach to Variable Elimination. Now I understand WHY it works, and it is so much easier to remember!
I dont understand pushing the Summation, I know when constant we can push but here I cant identify which is constant. And moreover I believe that they are the distribution. Containg 2,4 values not only single value. Correct me if im wrong.
@Akash We have assumed that Earthquake and Burglary are two independent events and they influence Alarm. Hence P(B) and P(E) are multiplied as they are. But alarm ringing depends on E and B hence P(A|,E.B). Since alarm influences John calling or Mary calling, we have P(j|A) and P(m|A). Hence P(B) P(E) P(A|E,B) P(m|A) P(j|A)
@@kpb6 I have one more doubt , what if we have some more parents to earthquake and burglary and some more child nodes to john and mary ?? Do all these things should be considered as hidden variables for computations?
Good lecture, but he expects students to guess what he's about to say often without it yet being clear what he's looking for, which seems common in teaching probability.
I love this video, it's much better than the mechanical approach to Variable Elimination. Now I understand WHY it works, and it is so much easier to remember!
i wish you the very best,
with all the love
- a struggling student
very helpful explanation! i wish my professor explained it this well :)
Excellent explanation professor!
Thanks a lot IIT D
This is what should be captioned as watch till end
awesome lecture
I dont understand pushing the Summation, I know when constant we can push but here I cant identify which is constant. And moreover I believe that they are the distribution. Containg 2,4 values not only single value. Correct me if im wrong.
Love the intro
From 14:00 How was sum over E written as f1(A,B) and sum over B written as f2(A)?
I don't understand how that full joint distribution summing over hidden variable came. EE Dept.
@Akash We have assumed that Earthquake and Burglary are two independent events and they influence Alarm. Hence P(B) and P(E) are multiplied as they are. But alarm ringing depends on E and B hence P(A|,E.B). Since alarm influences John calling or Mary calling, we have P(j|A) and P(m|A). Hence P(B) P(E) P(A|E,B) P(m|A) P(j|A)
@@kpb6 I have one more doubt , what if we have some more parents to earthquake and burglary and some more child nodes to john and mary ?? Do all these things should be considered as hidden variables for computations?
@@manjunathakapilsharma Precisely.
For instance:
if [Earthquake] had 2 causal nodes (i.e. parents), then we would do P(E | parents) etc.
If u can mention the particular timestamp, we might be able to exactly clarify the doubt.
tq so much 😇@@kpb6
21:48 i think its because p(bug...)= 0.001 and p(earth...) = 0.002
Good lecture, but he expects students to guess what he's about to say often without it yet being clear what he's looking for, which seems common in teaching probability.