Eric McCreath's Home Page

Research - Current Focus

My research focus is on employing novel architectures, such as the cell processor, GPUs, and FPGAs, for computationally intensive tasks. The two domains I have been looking at are computational chemistry and computer vision. I am a member of the Computer Systems research group.

Research Students

  • Anish Varghese(2013), Ultrasound simulation on a GPGPU using the Westervelt equation comp8750 - Computer & Programming Language Systems Project

  • Joshua Nelson(2012), Fast Multipole Algorithms vs the Particle Mesh Ewald Method comp3006 - Computer Science Research Project

  • Timothy Cosgrove(2012), Enhanced Visualization and Linking of Legislation comp3006 - Computer Science Research Project

  • Rui Chen(2012), Ultrasound Simulation on a GPGPU using the Westervelt Equation comp8790 - Software Engineering Project

  • Harry Slatyer and Malcolm Gill(2011), Optimising Networked Multiplayer Games by Transferring Computation to Clients ACS Project

    pdf

  • Andrew Haigh(2011), Implementation of rotation-based operators for Fast Multipole Method in X10 Summer Scholar

    pdf

  • Robin Srivastava(2010), Machine Learning on the Cell Processor comp8740 - Artificial Intelligence Project

  • Xiang Lv(2010), Email Usage Stats Collector comp8770 - eResearch Project

  • Temi Varghese(2010), Email Tagging on RainDrop comp8780 - Human Centrered Computer Project

  • Xiao Cui(2010), Implementation of a user interface for a new language for high performance computing comp8790 - Software Engineering Project

  • Luke Michael Vincent(2010), Android Bluetooth Remote comp3006 - Computer Science Research Project

  • Ya Jun Wang(2010), Diagram Drawing using Shape Recognition comp8780 - Human Centrered Computer Project

  • Yun Wang(2010), E-mail Usage Stats Collector comp8780 - Human Centrered Computer Project

  • Tao Wang(2010), Synchronisation Across Blocks on a CUDA GPU comp8790 - Software Engineering Project

  • Christopher John Fraser(2009), Fast Multipole Method on the Cell BCS /

    pdf

  • Ng Ye Hao(2008), Critical Success Factors in Municipal Wireless Networks Implementation BIT - honours

  • Yoon Chang(2008), An Intelligent Email Sorter Extension for Thunderbird MIT / COMP6703 (12 units)

  • Michael Chisholm(2008), Calculating Optical Flow using FGPAs BIT / COMP3750 (6 units)

  • Ben Griffiths(2008), Simple Rostering System BIT / COMP3760 (6 units)

  • Li Zhou(2008), Particle Simulation on the Cell Broadband Engine – Extending the types of interactions COMP3750 - Project Work in Comp Systems

  • Robert Bridle(2003-2008), Machine Learning on Mobile Phones PhD

    This project involves applying machine learning approaches onto mobile phones. The research looked both at predicting menu selection and inducing short cuttings.

  • James Navin(2006), Depth Estimation via Active Illumination BSE - honours research project

    This project investigated a pixel-based active illumination approach for depth estimation. Two images of a scene are taken from the same location, one image has a light source close to the camera on. The difference in the pixel intensities enables depth estimation. The results are promising and could be used as a low cost way of obtaining such depth estimations. pdf

  • Nick Piggin(2006), A Lockless Pagecache in Linux BSE - honours research project

    This project proposed and evaluated a lockless approach for the synchronization of the pagecache. This improved scalability on multiprocessor systems under various workloads. pdf

  • Warren Armstrong(2005), Using Learning to Improve the Computational Performance of the Simulation of a Modeled Environment BIT - honours

    This projected looked at the combination of speed and accuracy in a metric for inducing decision trees. That is, decision trees that executed quickly were chosen over slower trees. This was apply to the simulation of 2D bouncing balls, to improve the overall speed performance of the simulation. pdf

  • Andrew Bartlett(2004), Samba4 - Active Directory BSE - honours research project

    Description of authentication in particular Microsoft's proprietary NTLM and Microsoft's extensions to Kerberos

  • Thomas Alexander O'Callaghan(2003), A Hybrid Legal Expert System BS - honours

    Considered the combination of rule-based and case-based reasoning within expert systems. (James Popple was also a supervisor)

  • Leigh Michael Thompson(2003), Induction in a Large Temporal Domain - The Intelligent File Pre-fetcher for Linux BIT - honours

    Considered the use of machine learning to create a predictive pre-fectcher for linux

  • Rob Sangster(2000), Induction of Functions With Constants BCST - honours

    This project investigates induction of unknown constant values for learning functional logic programs.

Invited Presentations (from 2006)

  • Eric McCreath, "Using the Cell Broadband Engine and NVIDIA 8800 GPU for Computational Science: A Particle Dynamics Comparison" Department of Computer Science and Engineering The Chinese University of Hong Kong, August 1, 2008
  • Eric McCreath, "The Intelligent E-Mail Sorter", (details) (slides) Basser Seminar Series, Sydney University Australia,10 May, 2006

Research - old projects

Phd and Lime

I complete my Phd in 1999 and graduated form the Universtity of New South Wales in 2000. A major part of my thesis was the development of an ILP system called lime.

Inductive Logic Program

Inductive Logic Programming (ILP) is concerned with learning logic programs as hypothesis from examples. ILP can be viewed in many different ways:
  • machine learning where the hypothesis language is a subclass of first-order logic,
  • logic program synthesis,
  • system explanation and prediction, or
  • data compression.
The idea of learning with predicate logic as a description language goes back to Banerji in 1964. Plotkin in 1971 and Shapiro and Sammut in 1981 laid important foundations for ILP. In and mid-1970's Vere wrote programs based on Plotkin's and Reynold's work. Also, in the late 1970's Michalski developed first order learning systems. Another early system was Confucius developed by Cohen. However, the field has seen a resurgence in the 1990's and has been named inductive logic programming by Muggleton (There is currently a debate about a new name for the field.). Muggleton described it as the intersection of machine learning and logic programming. This is appropriate as the field utilizes these two well established areas. ILP, which is a particular setting for computational induction, uses first-order logic as the language of description and inference. The learner is also given background knowledge, which is a finite set of clauses, to aid and bias the induction. For a more complete review of the origins of ILP we refer the reader to Sammut's prehistoric tale.

The two aspects within ILP that I have focused on are learning from fixed example set sizes and learning when the data contains noise. The main artifact that was generated from my PhD was an ILP system called Lime. I am currently writing a new version that is more flexible in terms of the training data, also this new version is written in Java making it a simpler research tool for others to use and modify.

For a more in-depth look at ILP I refer the reader to my thesis.

An Intelligent Email Managers

I have also been involved in some research relating to an intelligent email manager. This research is in collaboration with Judy Kay and Elisabeth Crawford from Sydney University.

We have developed a demonstration system that learns how people filter email. this is then used to filter new messages within the inbox. Below is a screen shot of the user interface.

Smart Internet Technology CRC

I am also involved in the CRC(Cooperative Research Centre) for Smart Internet Technology. This will involve collaboration with a number of researcher within Australia including: John Lloyd, Paul Compton, Judy Kay, Ron van der Meyden, Supriya Singh and others. I also supervised Robert Bridle and we published some interesting results relating to smart mobile phone interfaces.