Research projects in Information Technology

According to Australian Network on Disability, over 4 million people (about 20% of the whole population) in Australia have some form of disability. At the same time, it is estimated that 15% Australians were aged 65 and over. For people with disabilities and older users, mobile phones and other mobile devices can provide increased freedom by allowing users to act independently while remaining in contact with friends, family, and caregivers.  However, studies of relatively small groups of mobile apps found that there still exists significant accessibility barriers.

Mobile apps now have become the most popular way of accessing the Internet as well as performing daily. Different from traditional desktop applications, mobile apps are typically developed under the time-to-market pressure and facing fierce competition — over 3.8 million Android apps and 2 million iPhone apps are striving to gain users on Google Play and Apple App Store, the two primary mobile app markets. Therefore, for app developers and companies, it is crucial to accelerate the mobile app development process.

This practice-based research involves further development of the AirSticks, a hardware/software package which allows the triggering and manipulation of sound and visuals in a 3D playing space, as a gestural instrument for live electronic music performance, music education and general health and wellbeing in collaboration with our interdisciplinary team at SensiLab. This can be done through new performances, new software or new hardware. How can we reinvent the connect between our bodies, our ears and our creativity, and what new applications for the AirSticks can be discovered?

This project will seek to further the research into and development of machine learning techniques that may be used to triage, classify, and otherwise process material of a distressing nature (such as child exploitation material). It will involve the use of deep neural networks for image, video, audio, social network, and/or text classification.

This project aims to identify novel methods for inferring where and when photographs and videos were recorded from features of the material itself. A key requirement of image processing in a Law Enforcement (LE) context is to augment classification of material by identifying its spatio-temporal context.

Advanced strongly typed languages like Haskell and emerging type systems like refinement types (as implemented in Liquid Haskell) offer strong guarantees about the correctness of programs.  However, when type errors occur it can be difficult for programmers to understand their cause.  Such errors are particularly confusing for people learning the language.  The situation is not helped by the cryptic error messages often produced by compilers.

We live and work in a world of complex relationships between data, systems, knowledge, people, documents, biology, software, society, politics, commerce and so on.  We can model these relationships as networks or graphs in the hope of reasoning about them - but the tools that we have for understanding such network structured data (whether algorithmic analytics or visualisation tools) remain crude.  Emerging display and interaction devices such as augmented and virtual reality headsets offer new ways to visualise and interact with data in the world around us rather than on screens.  This…

Testing self-driving cars is extremely difficult, as one has to account for a very large space of possible scenarios. In this project, we will explore the application of automated testing techniques, mainly in the area of search-based software testing to verify that the AI components of self-driving cars work as they should. This project is in collaboration with Professor Hai Vu and the Monash Connected Autonomous Vehicle (MCAV) team.

Defect prediction has been developed for more than four decades. Yet, a multitude of human aspects (i.e., both developers and end-users) have been rarely considered and incorporated. Thus, this project aims to focus on inventing theories and approaches for human-centric defect prediction to efficiently predict and explain non-functional requirement defects (e.g., accessibility issues and usability issues in Mobile Apps) that have the largest impact on end-users and humanity.

Adversarial Machine Learning (AML) is a technique to fool a machine learning model through malicious input. Due to its significance in many scenarios, including security, privacy, and health application, AML has attracted a large amount of attention in recent years. However, the underlying theoretical foundation for AML still remains unclear and how to design effective and efficient attack and defence algorithms are remain a challenge in the research community. Furthermore, most existing  AML algorithms can only apply to Euclidean space.