Pattern
Recognition and Analysis (BBL514E)
This year we have a number of projects. Please see papers related to each project and let me know which one you will be working on by October 22.
The output of your project will be a paper and a power point presentation.
The paper should include: Problem Definition, Previous Work, Current Work, Data, Experimental Methodology, Results, Conclusions, References.
The presentation will be performed in the class (20 minutes)
and should include the contents of the paper.
You can use prtools, cluto or another program to help you with your implementation.
Evaluation of the project:
1) [Mandatory][due Oct 22, in class] Tell me which papers/subjects you chose. Tell me also which three weeks of the class between Oct 22-Dec 17 you are available to meet at 9 am so that we can discuss your projects.
2) [Mandatory] [due November 26, 10am] Submit through Ninova a report which is a preliminary version of your final report and which includes Problem Definition, Previous Work, Current Work, Data, Experimental Methodology parts.
3) [Mandatory] [due Jan 21, 2008, 8:55am] Upload your paper and power point presentation to the ninova course web site.
4) Presentation [due Jan 21, 2008, 9-13] (If you are in a group of two, you should present the part of the work that you have done yourself):
a. [5 points] Clarity of written and spoken English
b. [5] Presentation format
c. [5] Problem Definition,
d. [10] Previous Work,
e. [20] Current Work,
f. [10] Data,
g. [15] Experimental Methodology,
h. [20] Results,
i. [5] Conclusions,
j. [5] References.
5) Your paper will contribute 5% and your presentation will contribute the rest 20% of your 25% total project score.
Project Topics (see papers at the ninova web site):
1) Bioinformatics (function prediction (see also SurveyPapers)):
a. SVMs for Protein Function Prediction (2+ people)
b. Text and sequence based approaches (2 people)
c. Bayesian Networks (2 people)
d. Motifs (1 person)
e. Feature Selection (1 person)
f. Protein protein interactions (1 person)
2) Classifier Combination for Music Genre Recognition (1 person)
3) Feature Selection (2+ people)
4) Backward ICA/PCA feature selection (1 person)
5) Wheat classification (2 people)