Honours and Minor Thesis projects

Proteomics data generated by cutting-edge mass spectrometers play a crucial part in early disease diagnosis, prognosis and drug development in the biomedical sector. It can be used to understand the expression, structure, function, interactions and modifications of virtually any protein in any cell, tissue or organ. Moreover, proteomics can be used in conjunction with other “omics” technologies such as genomics, transcriptomics or metabolomics to further unravel the complexity of signalling pathways and other subcellular systems.
…

Lipids such as cholesterol or triglycerides are involved in a plethora of medical disorders and diseases ranging from cardiovascular diseases (including obesity and artherosclerosis) to neurodegenerative disorders such as Parkinson’s disease. An in-depth analysis of individual lipid classes and species is often indispensable to unravel the mechanisms underlying disease onset and progression.

DeepLabCut™ is an efficient method for 3D markerless pose estimation based on transfer learning with deep neural networks that achieves excellent results across a broad collection of behaviours. This project will utilise the DeepLabCut package to analyse the behaviour of rats and mice as they are trained and tested on reward-based learning tasks designed to examine aspects of attention, memory and impulsive behaviour.

Activity and movement are fundamental diagnostic parameters of animal behaviour. However, measuring long-term individual movement within groups was not possible until recently. Our ActivityMonitor provides accurate individual movement data in a fully automated way. This is a unique solution for the 24/7 long-term tracking of individual animals living in groups, which utilises an array of RFID readers positioned under the home cage of rats and mice that are implanted with RFID transponders.

Working remotely under the COVID-19 pandemic has given rise to the demand for cloud-based technology, including online file sharing and cloud storage services. However, attackers have recently abused these platforms and propagated the emails that contain a file-sharing link to bypass the email filter. A typical example is that criminals can easily create and share phishing forms through legitimate form builders, e.g., Google Form to trick users into handing over sensitive information such as password or credit card number.

People are continuously receiving unsolicited emails where phishers impersonate legitimate organisations or trusted sender to harvest victim credentials. The rapid advance of AI boosts recent automatic detection of phishing attempts but also provides hackers with the opportunities to build increasingly sophisticated phishing tactics to bypass the filter. While attackers leverage social engineering to exploit human weakness, human skills can be a powerful component in cyber defence such as cognitive function and professional judgment.

Deep learning has achieved ground-breaking performance in many 2D vision tasks in the recent years. With more and more 3D data available such as those captured by Lidar, the next research trend is doing advanced perception on 3D data. The objective of this project is to study the state-of-the-art object detection techniques for 3D point clouds such as PointNet and PointVoxel.

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.

Modern machine learning is increasingly applied to create amazing new technologies and user experiences, many of which involve training machines to learn responsibly from sensitive data, such as personal photos or email. Ideally, the parameters of trained machine-learning models should encode general patterns rather than facts about specific training examples.

It is quite challenging to access to videos for people who are blind or have low vision (BLV), particularly creating audio descriptions that describe the scenes without interfering the dialogues in a video. There is also the challenge of providing additional information using multi-modal feedback, that is using non-speech audio and haptics.