Project 1: Spam Filter

Type of Project: Implementation

Study various spam filtering techniques based on machine learning.

Summarize the various methods.

Implement one of them

Test the method.

Try to make a prototype and integrate with a standard mailing system

Some references:

1) Bayesian Spam filtering

2) Graham, Paul (2002). A Plan for Spam.

3) Here you will find some spam filter product reviews

4) Some papers: (a) Johan Hovoltd

(b) Androutsopoulos et al.

(c) Androutsopoulos et al

Project 2: Face Recognition

Type of project: Experiment / Implementation

Implement a face recognition algorithm and test it on a standard data set.

Some references:

1) The face recognition page

Project 3: Ensemble Methods

Type of project: Theory/Methodology

Study the various ensemble methods available in literature

Propose a new ensemble technique, may be specific to a classifier.

Some references:

1) Bagging

2) Boosting

3) Kuncheva and Whitaker

4) Rodríguez, Kuncheva and Alonso

Project 4: Graph Clustering

Study some graph clustering algorithms and implement one of them.

Some references

1 http:/www.facweb.iitkgp.ernet.in/~pabitra/paper/barna-sdm07.pdf

2. http://i11www.iti.uni-karlsruhe.de/algo/people/dwagner/papers/bgw-egc-03.pdf

3. http://www.cs.utexas.edu/users/ml/papers/kernel-kdd-05.pdf

Project 5: SVM Regression

Implement the SMO algorithm for Support Vector regression.

Reference:

1) http://citeseer.ist.psu.edu/cache/papers/cs/30467/http:zSzzSzmlg.anu.edu.auzSz~smolazSz.zSzpaperszSzSmoSch03b.pdf/smola03tutorial.pdf

Project 6: K-NN Variants

Study implement and compare some K-NN variants.

References:

1) http://www.geocities.com/ghosh_anilk/k.pdf

2) http://www.geocities.com/ghosh_anilk/nnrule.pdf