Three Concepts: Probability

Project IV: Preparing a Poster

Presented posters

NB: Persons shown in the picture are not necessarily the authors.

Instructions

A part of the course (30 % of the grade) is to prepare a poster presentation for the joint poster session. The posters will be prepared either in pairs or individually. The default is in pairs so that you can take turns so that the other talks in front of your poster board while the other walks around getting acquainted with the other posters.

A poster session is an occasion where you present your topic to the public with the help of posters on a bulletin board. Your role is to stand nearby your poster, explain details and answer questions to the public. Because everyone is doing this at the same time, you do not need to give a lecture, and the public can wander around the poster session room and concentrate on posters whose topics interest them. The session is meant to be relaxed and rather informal.

Poster report

After the poster session, each student has to write a poster report containing a short (one paragraph) evaluation of each of the posters (execpt his/her own). The poster reports are to be done individually, not in pairs. In your evaluation, pay attention to both the poster layout (How well does the material explain the poster topic? Does it support the verbal poster presentation well?) and to the verbal presentation (Did the presenter/presenters seem to know their business? Were they capable of explaining the material well?). Please give the name or names of the people who were presenting the poster to you. Give also an overall score of the poster using the scale + (below average), ++ (average) and +++ (above average). Note that this means that each student has to visit all the posters during the poster session!. I strongly advice you to take notes during the session, as this will be a great help when writing the poster report. The deadline for the poster reports is May 11 (at Midnight).

Guidelines for poster presentation

Some guidelines for designing the poster can be found from the following addresses:

» Henry's slides on posters (Post Script)
» Poster preparation guidelines in Chemistry
» SIAM guidelines for preparing posters
» ... and many, many more from different search engines.

The size of the poster board is 118,5 cm x 150 cm (excluding frames). Copies of the poster material should be delivered to the instructors after the poster session.

Poster topics

You may choose freely a poster topic for yourself from the following list. The topics that have been already assigned have the names of the presentators after the topic. The topics are not ordered by their difficultness, for example - they are in random order.

For each topic, one relevant article is given. However, the idea is not to give a poster on the contents of that particular article only, but to explain more generally what the topic is about. So in your poster (and later in the term paper), try to answer the following questions:

• What is the problem addressed?
• Why is this problem being addressed?
• What is the general framework where the problem belongs to? What other interesting problems can be found in this framework, but are not addressed here?
• What is the solution suggested in the given paper? Is the result correct/feasible? Are there other solutions?
• What do the results mean, how important are they?
• What questions remained unsolved? Could the suggested solution be improved?

Depending on the length of the given article, you may need to acquaint yourself with the references given in the paper and/or do literature search in the Internet. In most cases the questions above cannot be answered and a good poster prepared by reading the single article only.

Note: You may also do the poster from a topic of your own. Ask Petri whether your topic needs improving or whether it's ok.

1. Bayesianism and Causality
   • Reference: J. Pearl, Bayesianism and Causality, or, Why I am Only a Half-Bayesian.
   • Presenters: Mikko Ajoviita and Pekka Maksimainen
2. Fuzzy vs. probabilistic reasoning
   • Reference: P. Cheeseman, Probabilistic Versus Fuzzy Reasoning. Harcopy can be found in the library: Uncertainty in Artifical Intelligence, L.N. Kanal and J.F. Lemmer, editors, pp. 85-102, Elsevier Science Publishers B.V. (North Holland), 1986.
   • Presenters: Juhani Bergström and Mikko Jormalainen
3. Probabilistic inference with Bayesian networks is hard
   • Reference: Dagum & Chavez, Approximating Probabilistic Inference in Bayesian Belief Networks.
     ( Dagum & Ruby, Approximating probabilistic inference in Bayesian belief networks is NP-hard, Artificial Intelligence, Volume 60, Issue 1, March 1993, Pages 141-153 ) --
   • Presenters:
4. Probabilistic inference via local message passing
   • Reference: Judea Pearl, Fusion, propagation, and structuring in belief networks. Artificial Intelligence, Volume 29, Issue 3, September 1986, Pages 241-288.
   • Presenters:
5. Probabilistic inference via conditioning
   • Reference: Darwiche, Any-Space Probabilistic Inference
     ( Suermondt & Cooper, Probabilistic inference in multiply connected belief networks using loop cutsets. International Journal of Approximate Reasoning, Volume 4, Issue 4, July 1990, Pages 283-306 )
   • Presenters:
6. Stochastic probabilistic inference with Bayesian Networks
   • Reference: Pradhan & Dagum, Optimal Monte Carlo Estimation of Belief Network Inference
   • Presenter:
7. Model averaging over Bayesian network structures
   • Reference: Friedman & Koller, Being Bayesian about Network Structure.
   • Presenters:
8. Bayesian clustering
   • Reference: P. Cheeseman, J. Stutz, Bayesian Classification (AutoClass): Theory and Results.
   • Presenters:
9. Supervised vs. unsupervised learning of Bayesian networks
   • Reference: P.Kontkanen, P.Myllymäki, T.Silander, and H.Tirri, On Supervised Selection of Bayesian Networks.
   • Presenters:
10. Learning with hidden variables
    • Reference: Elidan & Friedman, learning the dimensionality of hidden variables.
    • Presenters: Esa Elovaara and Jaakko Sorri
11. Computationally efficient learning of Bayesian trees
    • Reference: M. Meila, An Accelerated Chow and Liu Algorithm: Fitting Tree Distributions to High-Dimensional Sparse Data.
    • Presenter:
12. Bayesian Spam Filtering
    • Reference: Mehran Sahami, Susan Dumais, David Heckerman and Eric Horvitz: A Bayesian Approach to Filtering Junk E-Mail. (see also Bayesian spam filtering in Wikipedia and the references therein)
    • Presenters: Markus Aukeala and Hannu Maksimainen
13. Bayesian Text Analysis
    • Reference: W. Buntine and A. Jakulin: Discrete Components Analysis.
    • Presenters:
14. On Classification Error in Previously Unseen Examples
    • Reference: T. Roos, P. Grünwald, P. Myllymäki and H. Tirri: Generalization to Unseen Cases..
    • Presenters:
15. A Simple Approach for Finding the Globally Optimal Bayesian Network Structure
    • Reference: T. Silander and P. Myllymäki: A Simple Approach for Finding the Globally Optimal Bayesian Network Structure. .
    • Presenters:
16. Bayesian poker
    • Reference: K. Korb, A. Nicholson, N. Jitnah: Bayesian Poker.
    • Presenters: Mikko Heilimo and Paula Silvonen and Michael Malka.

Three Concepts: Probability	2007