The Information Technology Discipline academics possess world class expertise in artificial intelligence, particularly in knowledge modelling, software robots, image processing and their applications in e-health, tourism, social network analytics, education systems and active living. In 2015, Computer Science at Victoria University was ranked top 4% in the QS WU Subject Ranking. In 2017, Computer Science at Victoria University was ranked 2% by Times HE Subject Ranking, which was 2nd in Melbourne.
We have strong interests in applying intelligent technologies in smart environments such as smart homes and smart cities. If you would like to explore collaboration and knowledge transfer, please contact the Discipline Head, Professor Yuan Miao at Yuan.Miao@vu.edu.au .
The followings are some recent research projects.
- Data Exchange and Service Integration with Applications in Health Information Systems, ARC Linkage
- Privacy Preserving Data Sharing in Electronic Health Environment, ARC Linkage
- Accurate and Online Abnormality Detection in Multiple Correlated Time Series, ARC DECRA
- Using Data Mining Methods to Remove Uncertainties in Sensor Data Streams, ARC Discovery
- Real-time and Self- Adaptive Stream Data Analyser for Intensive Care Management, ARC Linkage
Intelligent software agent for heterogeneous systems data exchange and integration
System integration is often far more expensive than people thought. Modification to an existing system costs about 50 to 100 times more than that the change be made at the design phase. Further, if the systems are purchased from a third party vendor and the agreement for the vendor to make the corresponding change if the existing API does not support, the management and negotiation cost is also high. With this technology, it is possible to achieve data exchange among existing systems without accessing source codes, API or even encrypted database. It was demonstrated through two medical record systems which the team has no access to source codes and APIs.
Lightweight V&V for Intelligent Technologies
The technology has been applied in auto pilot car testing. On road testing is not only costly but also inefficient in some case. For example, it may take months to have some test cases in foggy weather. Through this technology, through this technology, test cases can be generated from existing ones. In this case, it has increased neuron coverage by ~100%. More than 100 erroneous behaviours were successfully detected. Retraining of the DNN improved the accuracy by up to 46%. The following images shows two fatal cases via generated foggy and rotated images.
Privacy Preserving, Access Control and Internal Attack Prevention
Many system can be vulnerable to internal unauthorised access. Large scale cases have been reported in government or national wide organisations. To enable stronger access control while without comprise usability, research on ambient intelligence is being developed to enable smart environment that protects systems and at the same time provides friendly access.
Big Data Analytics for Smart City
Social media has fast response to events and has powerful prediction potential. It can out perform the public health reporting system to provide in time status update of pandemic outbreaks. The analysis has been conducted through twitter data about tourists’ visit, interested topics and etc., and visualised over the google earth for city councils to plan and better support the tourism.
UAV surveillance, multi sensor fusion
This work involves video stabilization (shake removal), panorama generation, target tracking and multi sensor fusion.