10708 Probabilistic Graphical Models
Homework 1 Solutions
1
March 11, 2015
Directed Graphical Models (Pengtao)
1.1
Factorization to Imap
Given that P factorizes according to G, we have that:
n
P (Xi  Pa(Xi ),
P (X1 , . . . , Xn ) =
(1)
i=1
where X1 , .
18551
Group 18:
License Plate Recognition
Final Report
Pierre Ponce  Stanley S. Wang David L. Wang
Ponce

sswang

wang4
The purpose of this report is to explain the implementation of our
project, License Plate Recognition System.
This report will beg
Automatic Feedback Detection and
Elimination Using TI 67xx DSP Processors
18551, Digital Signal Processing and
Communications Design Project
Final Report
7 May 2000
Group 17
Alison Greenwald
Ros Neplokh
Will Wong
1 Problem Description
1.1 What Feedback I
Parameter Learning in MN
Outline
CRF
Learning CRF for 2d image segmenta:on
IPF parameter sharing revisited
Loglinear Markov network
(most common representa:on)
Feature is some func:on [D] for some subset of vari
10708 Probabilistic Graphical Models
Homework 2 Solutions
1
April 2, 2015
Expectation Maximization
1.1
Generalized EM
Let q(z) be an arbitrary distribution of z. The incomplete data loglikelihood:
(; x) = log p(x; ) = log
p(x, z; )
(1)
z
= log
z
Eq log
10708 Probabilistic Graphical Models
Homework 3
Due Apr 13, in class
Rules:
1. Homework is due on the due date in the class on April 13. Please see course website for policy on late
submission.
A
2. We recommend that you typeset your homework using appro
10708 Probabilistic Graphical Models
Homework 3 Solutions
1
April 16, 2015
Exact Inference
TODO
2
Variational Inference
TODO
3
Markov Chain Monte Carlo
3.1
Sampling Basics
1. We have U Unif (0, 1) and X F 1 (U ). Evaluate the cdf of X at t:
P (X t) = P F
10708 Probabilistic Graphical Models
Homework 2
Due Mar 2, in class
Rules:
1. Homework is due on the due date in the class on March 2. Please see course website for policy on late
submission.
A
2. We recommend that you typeset your homework using appropr
10708 Probabilistic Graphical Models
Homework 4
Due Apr 29, in class
Rules:
1. Homework is due on the due date in the class on April 29. Please see course website for policy on late
submission.
A
2. We recommend that you typeset your homework using appro
10708: Probabilistic Graphical Models, Spring 2015
1 : Introduction to GM and Directed GMs: Bayesian Networks
Lecturer: Eric P. Xing
1
Scribes: Wenbo Liu, Venkata Krishna Pillutla
Overview
This lecture introduces the notion of Probabilistic Graphical Mod
10708: Probabilistic Graphical Models, Spring 2015
3: Representation of Undirected GM
Lecturer: Eric P. Xing
1
Scribes: Karima Ma, Manu Reddy
Graphical Model Review
In the rst couple of lectures, we talked about Bayesian Networks (also called Directed Gr
10708: Probabilistic Graphical Models, Spring 2015
2 : Directed GMs: Bayesian Networks
Lecturer: Eric P. Xing
1
Scribes: Yi Cheng, Cong Lu
Notation
Here the notations used in this course are dened:
Random variables and values: Random variables are denote
Home Description People
LecturesRecitations
Homework Project
Previous
Probabilistic
GraphicalModels
10708, Spring 2015
EricXing
School of Computer Science, Carnegie Mellon
University
CourseDescription
Manyoftheproblemsinartificialintelligence,statistics,
10708: Probabilistic Graphical Models, Spring 2015
4 : Parameter Estimation in Fully Observed BNs
Lecturer: Eric P. Xing
1
Scribes: How Jing and Xiaoqiu Huang
Learning Graphical Models
The goal of learning graphical models is to discover the best Bayesia
10708 Probabilistic Graphical Models
Homework 1
Due Feb 13, 12:00 noon
Rules:
1. Homework is due on the due date at 12:00 noon. Please hand over your homework to Mallory Deptola
(GHC 8001). Please see course website for policy on late submission.
A
2. We