4TH YEAR STUDIES IN COMPUTER SCIENCE

Home

General Information

Important Dates

Units

Project

Assessment

Students

Resources

Miscellaneous Links
Graham Nolan
nolang01@student.uwa.edu.au

Entry year: 2008
Enrolment status: checked and confirmed
Degree: BE(SE)
Degree status: continuing

Project: Improving the k-Nearest Neighbour Algorithm with CUDA
Supervisor(s): Amitava Datta
Project status: expects to complete end of semester 2, 2009


Data classification is an important task for many modern software applications ranging from data mining systems to complex machine learning algorithms. One of the most frequently used classification methods is the k-Nearest Neighbour (kNN) algorithm. Despite it's relative ease of implementation, kNN can become very computationally intensive when applied to large or highly variable sets of n-dimensional data, hence there has been much research into ways to improve it's efficiency.
The recent release of CUDA by nVidia gives programmers the ability to perform parallel computation using commercially available graphics processors. This project aims to modify the kNN algorithm to take advantage of CUDA and determine what kind of performance improvement (if any) can be obtained through parallel processing.

Proposal
Last update: Thu Sep 3 15:51:21 2009
For further enquiries, please contact the 4th Year Coordinator, Luigi Barone.
Copyright © 2009
School of Computer Science & Software Engineering
The University of Western Australia
CRICOS Provider Code: 00126G
Last modified: Thu Sep 3 15:52:23 2009