Computer Science Summer Scholarship 2014/15


A limited number of summer scholarships are available to allow computer science students to undertake a small research project over the summer of 2014/15.

The scholarship is for $5,000 (tax free) for approx. 400 hours of work between November and Febuary. The project does not carry any academic credit towards your degree. Payment is in two tranches, the first before Christmas and the second on successful submission of a short project report (download the report guidelines).

Scholarships are available to University of Auckland students, those from other institutions within New Zealand and overseas students interested in coming to Auckland (download the Scholarship guidelines).

To apply you must have a Computer Science academic willing to supervise your project. Below is a list of projects staff  have proposed; however, you're encouraged to approach staff with your own project ideas.

Once you have found a supervisor, fill out the application form, email it back to Ian Watson along with a copy of your university transcript. Please apply as soon as possible.

Scholarships will be awarded primarily on GPA

This is a list of projects that staff are interested in supervising this summer. Please contact the supervisor directly for more information on a particular project. Note, you can propose your own project topic and ask a member of staff to supervise you.

Supervisor
Project description
Ian Watson
A Case-Based Chat Bot - the student will use a library of conversations between people on Internet chat sites to instruct a chat bot using case-based reasoning. The case-based chat bot should be able to recognise similar conversations in its library and respond accordingly. The chat bot will be evaluated using a Turing Test. Skills recquired are: compentent programming, an interest in AI and web development.
Reinhard Klette
Mapping recorded 3D roadside data into a uniform world coordinate system - Stereo vision creates clouds of points while driving a stereo camera system along a road. Those points need to be mapped into a uniform world coordinate system. The project aims at an improvedversion of one selected technique for doing so.
Radu Nicolescu
Evaluating different methods for visual odometry - Visual odometry applies cameras for understanding pose changes of a mobile system. The project aims at a comparative performance evaluation of 3 different methods for visual odometry.

Aniket Mahanti
Comparative evaluation of transfer protocols for big data - Performance evaluation of multiple (TCP and UDP based) transfer protocols for transferring large volumes of data over long distances. Skills needed: Computer networks, Linux shell scripting, Python or Java programming.
Mike Barley
Adaptive Problem Solvers - Help explore systems that solve problems by automatically modifying their conceptualization of the problem, their problem solving techniques, and their heuristics. These modifications are guided by the specific problem, the available computing resources, and the system’s progress in solving the problem.
Yun Sing Koh
Drift Detection in Memory-Constrained Data Stream Mining - With the proliferation of handheld devices and small sensors that can now perform tasks that in the near past required high performance systems. Data streams in and/or produced on such small devices can serve a number of important applications in areas such as astronomy, stock market analysis and social networks. The following two important facts require the processing of data streams to be performed locally on-board small devices with low computational power also known as memory-constrained environments. Focusing on this we now need to consider developing effective techniques that allow change detection data stream mining techniques for memory-constrained environments for continuous data streams.
Christof Lutteroth and Gerald Weber
Eye-gaze user interfaces - A user interface will be developed that makes use of an eye-gaze tracker to support users and help control applications.
Michael Dinneen Utility scripts and ranking algorithms for DOMjudge program submission system - The project will focus on learning php web for front-end tools for the open source DOMjudge contest system.  Additional features needed such as:
1)  custom ranking scoreboard (e.g. for NZPC) with problem point system instead of ACM ICPC rules,
2)  self-registration and team password maintenance,
3) import/export scripts for human and external systems (e.g. problem specs/data and configuration)
Aniket Mahanti Developing a mobile app for benchmarking NZ cellular operators - Themobile app (similar to the speedtest.net app) will perform network measurements for a mobile broadband operator such as latenc, throughput (upload/download), HTTP GET request, and other sensor data from the phone. The app can help benchmark NZ mobile operators. The app could also differentiate between a cellular connection (3G/4G or WiFi) and device type (phone, tablet, etc) among other things. On the backend, a flexible measurement collection platform (e.g., a server) needs to be developed that would collect all the measurement data sent by the app. Skills needed: computer networking, mobile development experience on Android and/or iOS.
Pat Riddle
Adaptive Problem Representation - Working with the planning system that came in 5th in the 2014 International Planning Competition, we will be exploring automatically adapting the representation. This involves determining when/how we can change individual objects into bags of objects and turning individual operators into macro-operators.  This will hopefully allow us to make 1st place in the next IPC competition.
Sam Trowsdale
Code the app “drop in the ocean” – modelling water has taken a radical departure from expert-lead processed-based models to participant-lead multi-user gaming. To get involved, you will develop an app to track a drop of water from street to ocean. You will need skills in computer networking and mobile development with an emphasis on GPS mapping and server development.  You will work alongside a PhD student in the School of Environment and lead the IT development with a small team of researchers. Strong students should apply.
Ian Warren
The BackPocketDriver App - Young drivers, aged 16-24, are twice as likely to be killed or injured as motorists in any other age group. The goal of BackPocketDriver is to develop a smartphone app that changes the behaviour of young drivers, making them safer and to reduce the road toll. BackPocketDriver is a multidisciplinary project that draws on behavioural change theory, psychology and computer science. The project has recently been gaining traction, and has been selected as one of the finalists in the University’s Spark $100K Challenge. Here’s an opportunity to get involved in the project over the summer months.