CPSC 540 course projects


Name

Contact

Project description

Dave Thompson

 

graph cuts for medical image segmentation

Jian Xu,
Hongrae Lee

{xujian} at cs dot ubc dot ca

automatic XML schema matching and mapping in p2p system ??

Ewout van den Berg,
Man Hon Chan,
Stanley Kai Him Chiu

{ewout78, mhchan, skhchiu}at cs dot ubc dot ca

Reimplement Vibes in Matlab

Lei Duan,
Shuang Hao

{lduan} at cs dot sfu dot ca,
{haos05} at cs.ubc.ca

Implementing structured penalties for GLS+

Nels Anderson

{andersnc} at cs dot ubc dot ca

Language processing using Mallet (I'm planning on using Mallet to give a demo of greedy discriminative feature induction. Time permitting, there may be some anaylsis on the features.)

Daniel Eaton

 

Swendsen-Wang on a toy problem

Disha Al Baqui

 

Topic Modelling [Language application]

Anirban Sinha

 

Comparing of WinBugs & Vibes for their speed/accuracy trade off & possibly also use Gaussian mixture models on them to see how both of them perform.

Mike Cora

{mcora} at cs dot ubc dot ca

Region classification using texture features in Matlab, based on the work of Julia Vogel and Tim Rees.

M. Argun Alparslan

 

Blob World

Aline Tabet,
Steve Kanters

 

Feature selection using RJMCMC

Alfred Pang

 

construct a suite of demos/tutorials for MatBUGS

Sharath J George

{sharathg} at cs dot ubc dot ca

Object Detection using classifiers

Andrew Carbonetto

 

Evaluating the segmentation algorithms mentioned in the Lectures (Blob World, the Superpixels, and the Felzenszalb segmentation). I'll evaluate the algorithms over the LabelMe database. I hope to provide some segmentation solutions to Mike Cora's texture classification project.

Brad Bingham

 

Either graph cuts (as described on slide 15) or "Fast BP 3" (as described on slide 12)

Maryam Mahdaviani

 

"L_1 Feature Selection for 1-D CRFs": Modifying the penalty in 1-D CRFs from l_2 norm to l_1 and testing it on a toy problem.

Mike Li

{mike} at gmail dot com

Summary and demo of the software under "gRaphical models in R" project (gR).
If I have time I would also port the function calls btw BNT and gR...

Wei Li

 

Fast BP: Distance Transform (Slide 12)

Akshay Gattani, Flavia Moser

 

LANGUAGE MODELING : "Combining n-gram models and Probabilistic Context Free Grammars via Directed Markov Random Fields".

Sebastian Streg

 

Survey of machine learning toolkits (Slice30)

Lin Xu,
Lan Wu

 

I will do lasso as my feature selection method. I plan to use lasso compare to our feature selection method (regression feature) to see which one is better, are they select the same set of features and so on. There is another student in CSPS540 interesting in that as well.

Chris Elliott

 

Bayesian Updating with Discretization