Lecture 9
Jan. 25-2006
Computational Inference
STAT 440 / 840, CM 461
Lecture: Ali Ghodsi
1
Scribes: Youn Kim
From Last Class
Parameterization of the joint probability on a undirected tree is:
1
p(x) =
(xi )
(xi , xj )
Z
iV
(1)
(i,j)E
Parameterization of
Lecture 19
Feb 17,2006
Computational Inference
STAT 440 / 840, CM 461
Lecture: Ali Ghodsi
Scribes: Zhao-Hui Li
in each step, a state will be selected according to p(z).
given a state, a data vector is drawn from p(x|z).
the value of each state is independ
Lecture 18
Feb. 15-2006
Computational Inference
STAT 440 / 840, CM 461
Lecture: Ali Ghodsi
Scribes: Jia Xi Luo
Mixture of Guassian
Given P (x|) = N (x; 1 , 1 ) + (1 )N (x; 2 , 2 )
Data = cfw_x1 , x2 .xn and x1 , x2 .xn are iid. with P (x|)
Find = cfw_, 1
Lecture 14
Feb. 6-2006
Computational Inference
STAT 440 / 840, CM 461
Lecture: Ali Ghodsi
Scribes: Yuan Xu
Exp. Learning A Bit Probability x = (x1 , x2 , , xn ) are iid, where
x1 , x2 , , xn are observations of x.
P (X|) P (x)
Model
P (xi = 1) =
P (xi =
Lecture 12
Feb 1-2006
Computational Inference
STAT 440 / 840, CM 461
Lecture: Ali Ghodsi
1
Scribes: Hye-Sung Baek
Statistical Concepts
1.1
(Contd) Bayesian and Frequentist statistics
In the previous class weve introduced Maximum Likelihood.
Maximum Likeli
Lecture 4
Jan. 11-2006
Computational Inference
STAT 440 / 840, CM 461
Lecture: Ali Ghodsi
1
Scribes: Alexandra Laamme
Canonical Graphs
Recall that to derive the Bayes Ball Algorithm we have been studying the conditional independence properties of three ca
Lecture 11
Jan. 30-2006
Computational Inference
STAT 440 / 840, CM 461
Lecture: Ali Ghodsi
1
Scribes: Juan Li
Example from last class
John is not a professional trader. However he trades in the copper market. Copper stock increase if
demand for copper is
Lecture 6
Jan. 16-2006
Computational Inference
STAT 440 / 840, CM 461
Lecture: Ali Ghodsi
1
Scribes: Stefan Pintilie
From Last Class
In chapter two there were two kinds of graphical models that were used to represent dependencies between variables. One gr
Lecture 7
Jan. 18-2006
Computational Inference
STAT 440 / 840, CM 461
Lecture: Ali Ghodsi
1
Scribes: Sukwoo Kim
Evaluation
Xi is an evidence node whose observed value is xi . To show that Xi is xed at the value xi ,
we dene an evidence potential (xi , xi
Computational Inference
STAT440/840, CM461
Assignment 3
Winter 2006
Department of Statistics and Actuarial Science
University of Waterloo
Due: Wednesday March 29, at the start of class
Worth: 20% of nal grade
Instructor: Ali Ghodsi
MS 6081G x7316, aghodsi
Lecture 3
Jan. 9-2006
Computational Inference
STAT 440 / 840, CM 461
Lecture: Ali Ghodsi
Scribes: Fei Yuan
n
From last lecture, we got this result: p(xv ) =
P (xi |xi ).
i=1
Now we take a look at the joint probability of an six-node example:
Figure 1: Dir