Research Assistant: Spatial and Temporal Learning

Research Assistant Position: Spatial and Temporal Learning

A position is available for a 6-month full time/12-month part-time research project in using machine learning techniques to detect spatial and temporal patterns. It is well suited to a part time final year Computer Science/Computer Engineering/Software Engineering student or part-time PhD student . Other people (e.g. recent graduates) may also wish to consider the appointment. The project consists of two sub-projects:

• Entering the SIGKDD 2007 Time Series Challenge: An international competition has been set up to look at different approaches to classifying time series [1] as part of this years SIGKDD conference (SIGKDD is the Special Interest Group in Knowledge Discovery and Data-mining). I have in the past created software well-suited for this kind of problem (in a project called TClass). In this part of the project, you will adjust TClass to function effectively on 20 "practice" datasets. On June 13th, the competition will provide the real datasets. We will then have 24 hours to apply our techniques to these unseen datasets before submission. This will also involve a trip (if the results are in the top half) to SIGKDD in California in August for the researcher to present our results.

• Exploring Spatio-Temporal Classification approaches: The techniques above work on time-series. In a more open-ended research project, you will be asked to see if the same techniques for temporal problems generalise to spatial problems (such as image processing). For example, can these techniques be used to detect several people in a scene in different poses?

Pay and timing

• Position to start as soon as possible. Either full time for 6 months (preferred) or part time for 12 months are options. Remuneration will be based on that of a UNSW Level 6 General staff member [2]. Times, days, etc can be negotiated.

Essential and Desirable characteristics

Essential:

• Understanding of machine learning techniques.
• Java programming skills.
• Interest in research.
• Must be completing/have completed an undergraduate degree in Comp Sci/Comp Eng/Soft Eng or related.
• Australian Permanent Resident/Citizen.

Desirable:

• Completing/have completed a PhD in Comp Sci/Comp Eng/Soft Eng or related.
• Experience with Weka.
• Attends/works at UNSW to simplify organising meetings.

Process

• Send a resume with contact details to waleed@cse.unsw.edu.au by March 26. If selected, I will organise a short one-on-one interview.