University of Helsinki homepageSuomeksiPå svenskaIn English
University of Helsinki Department of Computer Science
 

Department of Computer Science

Intelligent Systems Seminar (3cu), Spring 2010. In english!

The Seminar

Teacher: Hannes Wettig, room A324, office hours by appointment or chance, tel. (09) 191 51231, wettig at hiit dot fi

Time and location: 18.1.-26.2 and 15.3.-30.4. Mon 10-12, B119

Material

Your papers should be in UAI format, no longer that eight pages.

latex style file

instructions tex file

instructions pdf file

Of course, other than in these instructions, you should put your name.

A typical paper is structured like this

Here is the review form.

Introduction

The purpose of the seminary is to get acquainted with conference paper submission, review and presentation. You will have to write a paper on a topic of your interest, preferrably something you are working on anyway, and "submit" it. Subsequently it will be reviewed by your peers, which means that you will also have to review your fellow students' papers. The final version should incorporate the reviewers' comments. At the end of the seminar you will give a short presentation.

Structure

The working language of the seminar will be english. All papers, reviews and presentations should be in english.

We will meet once a week or less frequently if that suffices. At the beginning everyone needs to find a topic. It will not hurt at all to have one ready when the seminar starts. The schedule is relatively tight, as the preliminary, submittable version has to be ready in february, before the break.

The review process will start after all papers have been submitted. The last, probably three, sessions are reserved for the presentations, each of length approximately 20 minutes. Please attend all presentation sessions.

Topics

The choice of topics is relatively free, within the borders set by the seminar title, of course. Ideally, you would present a topic that you are currently working on, for example something related to your Master's Thesis. If you do not have a topic of your own, you may choose a scientific paper that has been presented recently at a suitable conference (e.g. IJCAI, UAI). This means more work for you, as you have to get acquainted with the subject thoroughly, present it understandibly to your peers and be able to answer their questions.

If in doubt about the suitability of a topic I'll be happy to discuss that with you already before the seminar starts.

Deliverables

In order to get your three credit units, you will have to do the following:

  • Agree on a topic with me
  • Prepare a paper of no more than 10 pages to be "submitted".
  • Review three papers submitted by your fellow students.
  • In case of disagreeing reviews, discuss with your fellow reviewers to arrive at a common conclusion about the paper in question.
  • Revise your paper in accordance to the reviews you have received. The final version should be ready a week before your presentation.
  • Give a talk of approximately 20 minutes including discussion. Be prepared to answer questions, especially those that have already arisen during the reviewing process.
  • Hand in your slides.
  • Read all papers before their presentation, have questions ready, especially if there is something you do not understand.
  • Participate in the discussions actively.

    • Grades

    Your grades will be based on all of the above with most weight on the final version of your paper. But as the whole picture will decide, I do not put any percentages here.

    Schedule (preliminary)

  • 18.1. First meeting, general discussion, choice of topics

  • 25.1.

  • 1.2.

  • 8.2.

  • 15.2.

  • 22.2. deadline for submitted version, assignment of reviewers to papers

  • 15.3.

  • 22.3.

  • 29.3.

  • 5.4. easter holiday

  • 12.4. presentation session I: images & text

    - Scale-Invariant features from natural images

    - A Probabilistic Model for Document Matching

  • 19.4. presentation session II: miscellaneous

    - Pattern Matching in Financial Markets

    - Reinforcement Learning with Lego Mindstorms NXT

    - Breaking WPA with CUDA and rainbow tables

    - An intelligent models based collaborative filtering algorithm

  • 26.4. presentation session III: networks

    - Manifold Learning in WLAN positioning

    - Route optimization in a probabilistic setting

    - Network Prediction for Adaptive Mobile Applications

    and, postponed from session I,

    - Multilingual Topic Models for Unaligned Text

    (We are not allowed to put student names on web pages anymore)

    Hannes Wettig Last modified: Sun Dec 27 12:31:37 EEST 2009

    •  
    Palaute laitokselle | Tiedekunnan etusivulle | Yliopiston etusivulle