Probabilistic Models
Exam
Year | Semester | Date | Period | Language | In charge |
---|---|---|---|---|---|
2014 | spring | 14.01-20.02. | 3-3 | English | Brandon Malone |
Lectures
Time | Room | Lecturer | Date |
---|---|---|---|
Tue 16-18 | D122 | Brandon Malone | 14.01.2014-20.02.2014 |
Thu 16-18 | D122 | Brandon Malone | 14.01.2014-20.02.2014 |
Exercise groups
Time | Room | Instructor | Date | Observe |
---|---|---|---|---|
Wed 16-18 | C222 | Quan Nguyen | 20.01.2014—21.02.2014 |
General
Please see the course syllabus.
The course learning objectives are available here.
A (tentative) detailed description of the lectures is available here.
A (tentative) set of objectives for each of the exercise sets is available here.
Enable the HTML5 player in YouTube here: https://www.youtube.com/html5
Literature and material
Materials for Class
Tuesday, January 16: Introduction to Probabilistic Models: Slides
Thursday, January 18: Refresher on Probability Theory: Slides
- A Tutorial on Probability Theory. Particularly Sections 1 - 7.
- (Optional) Combinations and Permutations: https://www.youtube.com/embed/C0p7kHOfMq8
- Introduction to Probability: http://www.youtube.com/embed/-8eSOmTPUbk
- Conditional Probability 1: http://www.youtube.com/embed/cwADSMeiIoE
- (Optional) Conditional Probability 2: http://www.youtube.com/embed/XYCCWrON7gQ
- Independent Events: http://www.youtube.com/embed/JV3_xdgC6yQ
- Bayes' Rule 1: http://www.youtube.com/embed/E2pOJwSwWDk
- (Optional) Bayes' Rule 2: http://www.youtube.com/embed/12fAdvRxqHI
- Joint Distributions: https://class.coursera.org/pgm/lecture/preview (Week 1, Distributions)
Note: Kevin deLaplante (http://www.youtube.com/user/PhilosophyFreak/videos) has many more introductory videos that may be helpful.
Tuesday, January 21: Bayesian Networks: Slides, Class work 1 solutions, d-Separation handout
- David Barber: Bayesian Reasoning and Machine Learning. Sections 3.1, 3.2, 3.3 (You will have to follow links to download the pdf.)
- (Optional) Graphical Models. Particularly the first 40 slides.
- (Optional) Independence and Conditional Independence: https://class.coursera.org/pgm/lecture/preview (Week 1, Conditional Independence)
- (Optional, but may be necessary to understand the next video) Factors: https://class.coursera.org/pgm/lecture/preview (Week 1, Factors)
- Introduction to Bayesian Networks: https://class.coursera.org/pgm/lecture/preview (Week 1, Semantics and Factorization)
- Intuitions of Reasoning with Bayesian Networks: https://class.coursera.org/pgm/lecture/preview (Week 1, Reasoning Patterns)
- d-separation 1: https://class.coursera.org/pgm/lecture/preview (Week 1, Flow of Probabilistic Influence)
- d-separation 2: https://class.coursera.org/pgm/lecture/preview (Week 1, Independencies in Bayesian Networks)
Thursday, January 23: Special Models (Naive Bayes and Hidden Markov Models): Naive Bayes Classifiers Slides, Document Classification Handout, (Partial) Document Classification Solutions
Note: In this class, we will review d-separation and discuss naive Bayes classifiers. We will discuss hidden Markov models and the forward-backward algorithm next Tuesday, January 28.
- Naive Bayes Classifiers
- Naive Bayes: https://class.coursera.org/pgm/lecture/preview (Week 1, Naive Bayes)
- Bayesian Classifiers: http://youtube.com/embed/ivBSZZyaRHY (You can ignore the bit about continuous variables at the end.)
- Naive Bayes Classifier: http://youtube.com/embed/54wfthrhwLQ (Again, you can ignore the part about continuous variables.)
Tuesday, January 28: Special Models (Hidden Markov Models): Hidden Markov Models Slides, Gene Finding Handout, Gene Finding Handout Solutions
- Hidden Markov Models
- Hidden Markov Models, Introduction: http://youtube.com/embed/TPRoLreU9lA
- Hidden Markov Models, Detailed Example: http://youtube.com/embed/M_IIW0VYMEA
- Forward-Backward Algorithm, Introduction: http://youtube.com/embed/7zDARfKVm7s
- Forward Algorithm: http://youtube.com/embed/M7afek1nEKM
- (Optional) Forward Algorithm, Detailed Discussion: http://youtube.com/embed/MPmrFu4jFk4
- Backward Algorithm: http://youtube.com/embed/jwYuki9GgJo
- (Optional) Template Models: https://class.coursera.org/pgm/lecture/preview (Week 1, Overview of Template Models)
- (Optional) Hidden Markov Models: https://class.coursera.org/pgm/lecture/preview (Week 1, Temporal Models - HMMs)
Thursday, January 30: Inference by Factors Elimination: Factor Elimination Slides
Note: Much of the material posted for today will also be relevant when discussing the junction tree algorithm on Tuesday, February 4.
- Bishop: Pattern Recognition and Machine Learning, Chapter 8. Sections 8.4.1 - 8.4.4
- (Optional) David Barber: Bayesian Reasoning and Machine Learning. Section 5.1 (You will have to follow links to download the pdf if you have not already.)
- Factors: https://class.coursera.org/pgm/lecture/preview (Week 1, Factors) (You may have already watched this one)
- Message Passing: https://class.coursera.org/pgm/lecture/preview (Week 3, Belief Propagation)
- Cluster Graphs: https://class.coursera.org/pgm/lecture/preview (Week 3, Properties of Cluster Graphs)
- (Optional) Belief Propagation in Trees: https://class.coursera.org/pgm/lecture/preview (Week 3, Clique Tree Algorithm - Computation)
- (Optional) Cluster Trees
- (Optional) HMM Algorithms as Sum-Product
Tuesday, February 4: Inference in General Bayesian Networks (Jointree Algorithm, also called the junction tree algorithm): Jointree Slides, Asia Network Handout, Asia Network Handout Solutions
- Junction Tree Algorithm
- (Optional) The Junction Tree Algorithm
- (Optional) Huang and Darwiche. Inference in belief networks: A procedural guide, International Journal of Approximate Reasoning, 1996, 15, 225 - 263.
- (Optional) The Junction Tree Algorithm, Presentation
Thursday, February 6: Multinomial Parameter Estimation: Parameter Estimation with Complete Data Slides, Parameter Estimation Handout, (Partial) Parameter Estimation Handout Solutions
- Maximum likelihood estimates: https://class.coursera.org/pgm/lecture/preview (Week 7, Maximum Likelihood Estimation)
- Maximum likelihood estimates for Bayesian networks: https://class.coursera.org/pgm/lecture/preview (Week 7, Maximum Likelihood Estimation for Bayesian Networks)
- Bayesian estimates: https://class.coursera.org/pgm/lecture/preview (Week 7, Bayesian Estimation)
- Bayesian estimates for Bayesian networks: https://class.coursera.org/pgm/lecture/preview (Week 7, Bayesian Estimation for Bayesian Networks)
Tuesday, February 11: Scoring Functions for Structure Learning: Scoring Functions Slides, Scoring Functions Handout, Scoring Functions Handout Solutions
- Likelihood scores: https://class.coursera.org/pgm/lecture/preview (Week 8, Likelihood Scores)
- Bayesian scores: https://class.coursera.org/pgm/lecture/preview (Week 8, Bayesian Scores)
- Optimal structure learning (Section 2)
- (Optional) BIC and asymptotic consistency: https://class.coursera.org/pgm/lecture/preview (Week 8, BIC and Asymptotic Consistency)
Thursday, February 13: Algorithms for Score-based Structure Learning: Structure Learning Slides, Structure Learning Handout, Structure Learning Handout Solutions
- Greedy hill climbing: https://class.coursera.org/pgm/lecture/preview (Week 8, Learning General Graphs: Heuristic Search)
- Optimal structure learning (Section 3)
- (Optional) Efficient algorithms: https://class.coursera.org/pgm/lecture/preview (Week 8, Learning General Graphs: Search and Decomposability)
Tuesday, February 18: Expectation-Maximization and Poisson Mixture Models: Poisson Mixture Model slides, Poisson Mixture Models Handout, Poisson Mixture Models Handout Solutions
- Expectation-maximization: https://class.coursera.org/pgm/lecture/preview (Week 9, Expectation-Maximization - Intro)
- Deriving the Poisson distribution, part 1: http://www.youtube.com/embed/3z-M6sbGIZ0
- Deriving the Poisson distribution, part 2: http://www.youtube.com/embed/Jkr4FSrNEVY
- Update rules for Poisson mixture models
- (Optional) Gaussian mixture models: http://www.youtube.com/embed/Rkl30Fr2S38
- (Optional) Introduction to Gaussian mixture models
- (Optional) The Poisson distribution: http://www.youtube.com/embed/jmqZG6roVqU
- (Optional) Introduction to Gaussian mixture models (Sections 1, 2 and 3)
- (Optional) Applications of EM: https://class.coursera.org/pgm/lecture/preview (Week 9, Latent Variables)
Thursday, February 20: Topic Models, A Recent Success Story: Topic Models Slides
- Blei, "Probabilistic topic models," Communications of the ACM, vol. 55, no.4, April 2012
- (Optional) Topic models: http://videolectures.net/mlss09uk_blei_tm/ (Parts 1 and 2)
- (Optional) Steyvers and Griffiths, "Probabilistic topic models," in Latent Semantic Analysis: A Road to Meaning, eds. T. Landauer et al.
Lab Exercises
Exercise grades (Updated with Set 5 and Total Points)
Wednesday, January 22: Foundations of Probability, 3 points
- Exercise set 1
- The LaTex file
- Solutions
Wednesday, January 29: Bayesian Networks and Naive Bayes Classifiers, 5 points
- Exercise set 2
- The LaTex file
- Solutions, NBC.R
Wednesday, February 5: Hidden Markov Models and Inference by Factor Elimination
- Exercise set 3
- The LaTex file
- Solutions, HMM.R, ET.R
Wednesday, February 12: Inference with Jointrees and Parameter Estimation with Complete Data
- Exercise set 4
- The LaTex file
- Solutions, jointree.R
Wedensday, February 19: Scoring Functions and Structure Learning
- Exercise set 5
- The LaTex file
- Solutions, BN.R
LaTex References
The Wikibook on LaTex is quite thorough and well organized. In particular, the Mathematics section may be helpful.
Stack Overflow and the relevant section in Stack Exchange are also good resources, although it is often easiest to search using Google.
Additional Material
- Material of the year 2013 course
- Material of the year 2012 course
- Material of the year 2011 course
- Material of the year 2010 course.
- David Barber: Bayesian Reasoning and Machine Learning, parts I and II.
- Richard E. Neapolitan: Learning Bayesian Networks.
- Stuart Russell and Peter Norvig, Artificial Intelligence: A Modern Approach. Prentice Hall, 2003. Chapters 13, 14, (15, optional), 20.1-20.3
- Judea Pearl: Probabilistic Reasoning in Intelligent Systems. Morgan Kaufmann, 1988.
- Daphne Koller and Nir Friedman: Probabilistic graphical models. MIT Press, 2009.
- Daphne Koller's online course on Probabilstic Graphical Models (recorded lectures can be found here).
- Myllymäki & Tirri, Bayes-verkkojen mahdollisuudet.
- B-Course: Bayesian software and on-line tutorial.
- Bishop: Pattern Recognition and Machine Learning, Chapter 8.
- Loredo, From Laplace to Supernova SN 1987A: Bayesian Inference in Astrophysics.
- Mackay, Information Theory, Inference, and Learning Algorithms: Chapter IV
- Heckerman, A Tutorial on Learning Bayesian Networks.
- Buntine, Operations for Learning with Graphical Models.
- Jeffrey, Probabilistic thinking
- Jaynes, Probability Theory: The Logic of Science.
- Howson & Urbach: Bayesian reasoning in science