Laboratory of Computer and Information Science / Neural Networks Research Centre CIS Lab Helsinki University of Technology

Näitä sivuja ei päivitetä enää. Ole hyvä ja katso tietojenkäsittelytieteen laitoksen WWW-sivuja:

These pages are not any more updated. Please, see web pages of Department of Information and Computer Science (ICS):

T-61.5140 Machine Learning: Advanced Probabilistic Methods (5 cr) P

Lecturer: Jaakko Hollmén, D.Sc.(Tech.)
Course Assistant: Tapani Raiko, D.Sc.(Tech.)
Semester: Spring term, periods III and IV
Credit points: 5 ECTS credit points, also eligible for post-graduate studies
Lectures:Lecture hall T3, Thursdays 10-12 (starting on the 17th of January)
Exercises:Lecture hall T3, Fridays 10-12 (starting on the 25th of January)
Language: English
Course book:Christopher M. Bishop: Pattern Recognition and Machine Learning, Springer, 2006.
Contents: The course covers probabilistic concepts in machine learning starting with the concepts of independence and conditional independence as the basis for subsequent models. During the course, we will study probabilistic modeling with Naive Bayes type of models and finite mixture models and extend the treatment to Bayesian networks. Learning of the model parameters is covered in the framework of maximum likelihood. Computational algorithms for exact and approximate inference will be covered. Other topics include sampling and the use of prior information.

Schedule: Lectures and exercise sessions

Time Topic on the lecture Materials Exercises Materials
17.1. Introduction to the course and the topic Lecture notes
24.1. Graphical Models Bishop: Chapter 8 25.1. problems, solutions
31.1. Graphical Models, cont. Bishop: Chapter 8 1.2. problems, solutions, code
14.2. Graphical models, cont. Lecture notes 7.2. problems, solutions
15.2. Self-study lecture about graphical models Self-study problems Useful formulae (ver 1)
21.2. Winter holiday, no lecture N/A 22.2. problems, solutions
28.2. Mixture models and EM Bishop, Chapter 9
Lecture notes
29.2. problems, solutions
6.3. No lecture, examination period N/A no (7.3.) N/A
13.3. Mixture Models and EM, cont. Bishop, Chapter 9 14.3. problems, solutions
20.3. No lecture, easter holiday N/A no (21.3.) N/A
27.3. Sequential Data Bishop, Chapter 13
Term project
28.3. problems, solutions
3.4. Approximate inference, (Sampling) Bishop, Chapter 11 4.4. problems, unknown_p.m, gaussian.m
10.4. Approximate inference Bishop, Chapter 11 11.4. solutions, matlab code, BUGS demo
17.4. Variational Bayesian learning Bishop Chapter 10,
Lecture notes
18.4. problems, solutions
24.4. Run-through of the course
Exam requirements
Exam requirements 25.4. problems, solutions

Term project

In the term project, you will use the software package BernoulliMix to solve real machine-learning problems using ready-made implementation of finite mixture models of multivariate Bernoulli distributions.

The deadline for the term project work is May 31st. If you are a visiting student, and return your report earlier, you will get feedback before your departure.


The first examination is scheduled to be on the 15th of May, 2008. See the examination schedule for up-to-date information and further examination dates. The exam questions given on 15th of May, 2008.

You are at: CIS → T-61.5140 Machine Learning: Advanced Probabilistic Methods

Page maintained by, last updated Tuesday, 19-Aug-2008 10:51:03 EEST