Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h)...
Transcript of Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h)...
![Page 1: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/1.jpg)
Approximate Inference in Bayes NetsSampling based methods
Mausam
(Based on slides by Jack Breese and Daphne Koller)
1
![Page 2: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/2.jpg)
Intuition
• Suppose I have a coin whose p(heads) is unknown
• How could I estimate it?
• When will I get the correct probability?
• Bayes Net inference is not a learning problem– But similar intuitions apply– In particular, generate samples from a Bayes net– But the samples should be unbiased!
2
![Page 3: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/3.jpg)
Sampling
• Samples should be representative of the world
• Samples: P(people > 60 yrs age in Delhi)
– Computer Science class
– Call on landline
– Call on cellphone
– Check facebook…
– Count at election booth
3
![Page 4: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/4.jpg)
Bayes Nets is a generative model• We can easily generate samples from the
distribution represented by the Bayes net
– Generate one variable at a time in topological order
4Use the samples to compute marginal probabilities, say P(c)
![Page 5: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/5.jpg)
P(B|C)
5
![Page 6: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/6.jpg)
P(B|C)
6
![Page 7: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/7.jpg)
P(B|C)
7
![Page 8: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/8.jpg)
P(B|C)
8
![Page 9: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/9.jpg)
P(B|C)
9
![Page 10: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/10.jpg)
P(B|C)
10
![Page 11: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/11.jpg)
P(B|C)
11
![Page 12: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/12.jpg)
P(B|C)
12
![Page 13: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/13.jpg)
P(B|C)
13
![Page 14: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/14.jpg)
P(B|C)
14
![Page 15: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/15.jpg)
P(B|C)
15
![Page 16: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/16.jpg)
P(B|C)
16
![Page 17: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/17.jpg)
P(B|C)
17
![Page 18: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/18.jpg)
P(B|C)
18
![Page 19: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/19.jpg)
P(B|C)
19
![Page 20: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/20.jpg)
P(B|C)
20
![Page 21: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/21.jpg)
Rejection Sampling
• Sample from the prior
– reject if do not match the evidence
• Returns consistent posterior estimates
• Hopelessly expensive if P(e) is small
– P(e) drops off exponentially with no. of evidence vars
21
![Page 22: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/22.jpg)
Likelihood Weighting
• Idea
– each sample agrees with evidence
– pays some price for the agreement (weight)
• Algorithm
– fix evidence variables
– sample only non-evidence variables
– weight each sample by the likelihood of evidence
22
![Page 23: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/23.jpg)
P(B|C)
23
![Page 24: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/24.jpg)
P(B|C)
24
![Page 25: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/25.jpg)
P(B|C)
25
![Page 26: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/26.jpg)
P(B|C)
26
![Page 27: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/27.jpg)
P(B|C)
27
![Page 28: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/28.jpg)
P(B|C)
28
![Page 29: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/29.jpg)
P(B|C)
29
![Page 30: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/30.jpg)
P(B|C)
30
![Page 31: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/31.jpg)
P(B|C)
31
![Page 32: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/32.jpg)
P(B|C)
32
![Page 33: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/33.jpg)
Likelihood Weighting
• Sampling probability: S(z,e) =
– Neither prior nor posterior
• Wt for a sample <z,e>:
• Weighted Sampling probability S(z,e)w(z,e)
=
= P(z,e)
• returns consistent estimates
• performance degrades w/ many evidence vars
– but a few samples have nearly all the total weight
– late occuring evidence vars do not guide sample generation
i
))Parents(Z|P(z ii
i
ii )Parents(E|P(e e) w(z,
i
ii )Parents(E|P(ei
))Parents(Z|P(z ii
33
![Page 34: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/34.jpg)
34
MCMC with Gibbs Sampling• Fix the values of observed variables
• Set the values of all non-observed variables randomly
• Perform a random walk through the space of complete variable assignments. On each move:
1. Pick a variable X
2. Calculate Pr(X=true | all other variables)
3. Set X to true with that probability
• Repeat many times. Frequency with which any variable X is true is it’s posterior probability.
• Converges to true posterior when frequencies stop changing significantly
– stationary distribution, mixing
![Page 35: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/35.jpg)
35
Markov Blanket Sampling
• How to calculate Pr(X=true | all other variables) ?
• Recall: a variable is independent of all others given it’s Markov Blanket
– parents
– children
– other parents of children
• So problem becomes calculating Pr(X=true | MB(X))
– We solve this sub-problem exactly
– Fortunately, it is easy to solve
( )
( ) ( | ( )) ( | ( ))Y Children X
P X P X Parents X P Y Parents Y
![Page 36: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/36.jpg)
36
Example
( )
( ) ( | ( )) ( | ( ))Y Children X
P X P X Parents X P Y Parents Y
( )
( , , , )( | , , )
( , , )
( , , )
( ) ( )( | ) ( | , )
( , , )
( | )
( | ) ( ) ( | , )
( | , )
P X A B CP X A B C
P A B C
P A B
P A P X A P C P B
C
P A P CP X A P B X C
P A B C
P X
X
A P B X C
C
A
X
B
C
![Page 37: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/37.jpg)
Example
37
Smoking
Heartdisease
Lungdisease
Shortnessof breath
P(h)
s 0.6
~s 0.1
P(g)
s 0.8
~s 0.1
H G P(b)
h g 0.9
h ~g 0.8
~h g 0.7
~h ~g 0.1
P(s)
0.2
![Page 38: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/38.jpg)
Example
• Evidence: s, b
38
Smoking
Heartdisease
Lungdisease
Shortnessof breath
P(h)
s 0.6
~s 0.1
P(g)
s 0.8
~s 0.1
H G P(b)
h g 0.9
h ~g 0.8
~h g 0.7
~h ~g 0.1
P(s)
0.2
![Page 39: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/39.jpg)
Example
• Evidence: s, b
• Randomly set: h, g
39
Smoking
Heartdisease
Lungdisease
Shortnessof breath
P(h)
s 0.6
~s 0.1
P(g)
s 0.8
~s 0.1
H G P(b)
h g 0.9
h ~g 0.8
~h g 0.7
~h ~g 0.1
P(s)
0.2
![Page 40: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/40.jpg)
Example
• Evidence: s, b
• Randomly set: h, g
• Sample H using P(H|s,g,b)
40
Smoking
Heartdisease
Lungdisease
Shortnessof breath
P(h)
s 0.6
~s 0.1
P(g)
s 0.8
~s 0.1
H G P(b)
h g 0.9
h ~g 0.8
~h g 0.7
~h ~g 0.1
P(s)
0.2
![Page 41: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/41.jpg)
Example
• Evidence: s, b
• Randomly set: h, g
• Sample H using P(H|s,g,b)
• Suppose result is ~h
41
Smoking
Heartdisease
Lungdisease
Shortnessof breath
P(h)
s 0.6
~s 0.1
P(g)
s 0.8
~s 0.1
H G P(b)
h g 0.9
h ~g 0.8
~h g 0.7
~h ~g 0.1
P(s)
0.2
![Page 42: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/42.jpg)
Example
• Evidence: s, b
• Randomly set: h, g
• Sample H using P(H|s,g,b)
Suppose result is ~h
Sample G using P(G|s,~h,b)
42
Smoking
Heartdisease
Lungdisease
Shortnessof breath
P(h)
s 0.6
~s 0.1
P(g)
s 0.8
~s 0.1
H G P(b)
h g 0.9
h ~g 0.8
~h g 0.7
~h ~g 0.1
P(s)
0.2
![Page 43: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/43.jpg)
Example
• Evidence: s, b
• Randomly set: h, g
• Sample H using P(H|s,g,b)
Suppose result is ~h
Sample G using P(G|s,~h,b)
Suppose result is g
43
Smoking
Heartdisease
Lungdisease
Shortnessof breath
P(h)
s 0.6
~s 0.1
P(g)
s 0.8
~s 0.1
H G P(b)
h g 0.9
h ~g 0.8
~h g 0.7
~h ~g 0.1
P(s)
0.2
![Page 44: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/44.jpg)
Example
• Evidence: s, b
• Randomly set: h, g
• Sample H using P(H|s,g,b)
Suppose result is ~h
Sample G using P(G|s,~h,b)
Suppose result is g
Sample G using P(G|s,~h,b)44
Smoking
Heartdisease
Lungdisease
Shortnessof breath
P(h)
s 0.6
~s 0.1
P(g)
s 0.8
~s 0.1
H G P(b)
h g 0.9
h ~g 0.8
~h g 0.7
~h ~g 0.1
P(s)
0.2
![Page 45: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/45.jpg)
Example
• Evidence: s, b
• Randomly set: h, g
• Sample H using P(H|s,g,b)
Suppose result is ~h
Sample G using P(G|s,~h,b)
Suppose result is g
Sample G using P(G|s,~h,b)
Suppose result is ~g 45
Smoking
Heartdisease
Lungdisease
Shortnessof breath
P(h)
s 0.6
~s 0.1
P(g)
s 0.8
~s 0.1
H G P(b)
h g 0.9
h ~g 0.8
~h g 0.7
~h ~g 0.1
P(s)
0.2
![Page 46: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/46.jpg)
Gibbs MCMC Summary
• Advantages:
– No samples are discarded
– No problem with samples of low weight
– Can be implemented very efficiently• 10K samples @ second
• Disadvantages:
– Can get stuck if relationship between two variables is deterministic
– Many variations have been devised to make MCMC more robust
46
P(X|E) =number of samples with X=x
total number of samples
![Page 47: Approximate Inference in Bayes Netsmausam/courses/col333/autumn2019/lectures/1… · of breath P(h) s 0.6 ~s 0.1 P(g) s 0.8 ~s 0.1 H G P(b) h g 0.9 h ~g 0.8 ~h g 0.7 ~h ~g 0.1 P(s)](https://reader034.fdocuments.net/reader034/viewer/2022050612/5fb33bcc8528e8408a6613e3/html5/thumbnails/47.jpg)
Other inference methods
• Exact inference
– Junction tree
• Approximate inference
– Belief Propagation
– Variational Methods
– Metropolis-Hastings
47