Honours and Masters project

Commonsense reasoning refers to the ability of capitalising on commonly used knowledge by most people, and making decisions accordingly. This process usually involves combining multiple commonsense facts and beliefs to draw a conclusion or judgement. While human trivially performs such reasoning, current Artificial Intelligence models fail, mostly due to challenges of acquiring relevant knowledge and forming logical connections between them. This project aims to develop and evaluate machine learning models for commonsense reasoning, with question answering as the key application. 

Developing quality AI tools for legal texts is the focus of enormous industry, government and scholarly attention. The potential benefits include greater efficiency, transparency and access to justice. Moving beyond the hype requires novel transdisciplinary effort to combine IT and Law expertise. This project engages this challenge by developing a semi-structured knowledge base (KB) and reasoners for statutes and cases. The project will also construct corresponding training and evaluation datasets.

This project is within the scope of the project “Artificial Intelligence in carDiac arrEst” (AIDE), which was led by Ambulance Victoria (AV) in Australia, involving a team of researchers at Monash University. This AIDE project has developed an Artificial Intelligence (AI) tool to recognise potential Out-of-Hospital-Cardiac Arrest (OHCA) during the Triple Zero (000) call by using transcripts produced by Microsoft Automatic Speech Recognition service. In the next step, we aim to optimise the workflow of call-takers and investigate which workflows can lead to earlier identification of OHCA.…

A lot of decision support systems have been developed to predict or suggest a diagnosis about the health conditions of patients with the aim to assist clinicians in their decisional process. One of the techniques that is proved to present an efficient tool for medical healthcare decision making is Bayesian networks (BNs). BNs are recognized as efficient graphical models that can be used to explain the relationships between variables.

Digital signatures are asymmetric cryptographic schemes used to validate the authenticity and integrity of digital messages or documents. The signer uses their private key to generate a signature on a message. Then, this signature can be validated by any verifier who knows the signer’s corresponding public key. Sometimes a digital message might require signatures from a group of signers. The naïve method to achieve this goal is collecting distinct signatures from all signers.

Visually discriminating the identity of multiple (similar looking) objects in a scene and creating individual tracks of their movements over time, namely multi-object tracking (MOT), is one of the basic yet most crucial vision tasks, imperative to tackle many real-world problems in surveillance, robotics/autonomous driving, health and biology.

The ability to forecast human trajectory and/or body motion (i.e. pose dynamics and trajectory) from camera or other visual sensors is an essential component for many real-world applications, including robotics, healthcare, detection of perilous behavioural patterns in surveillance systems.

Human behaviour understanding in videos is a crucial task in autonomous driving cars, robot navigation and surveillance systems. In a real scene comprising of several actors, each human is performing one or more individual actions. Moreover, they generally form several social groups with potentially different social connections, e.g. contribution toward a common activity or goal.

3D localisation, reconstruction and mapping of the objects and human body in dynamic environments are important steps towards high-level 3D scene understanding, which has many applications in autonomous driving, robotics interaction and navigation. This project focuses on creating the scene representation in 3D which gives a complete scene understanding i.e pose, shape and size of different scene elements (humans and objects) and their spatio-temporal relationship.

In this project, the goal is to develop a new method (using computer vision and machine learning techniques) for robotic navigation in which goals can be specified at a much higher level of abstraction than has previously been possible. This will be achieved using deep learning to make informed predictions about a scene layout and navigating as an active observer in which the predictions inform actions.