Searching for near-exact protein models. This project aims to develop novel and efficient heuristic-based algorithms leading to near accurate protein tertiary structure models. Knowledge about protein structures is fundamental to our understanding of living systems. The progress on experimental determination of these structures has been extremely limited and remains an open challenge in molecular biology. Computational prediction of protein structures from sequences is emerging as a promising ap ....Searching for near-exact protein models. This project aims to develop novel and efficient heuristic-based algorithms leading to near accurate protein tertiary structure models. Knowledge about protein structures is fundamental to our understanding of living systems. The progress on experimental determination of these structures has been extremely limited and remains an open challenge in molecular biology. Computational prediction of protein structures from sequences is emerging as a promising approach, but its accuracy is far from satisfactory. The software systems developed in this project will be used in structural identification of target proteins in drug design. This will make drug design process more efficient, saving time and cost, potentially saving lives.Read moreRead less
Automatic detection and modelling of acoustic markers of speech timing. This project aims to create new automatic sensing, analysis and assessment of cognitive, affective, mental and physical state from voice for mobile and computing devices. This project expects to generate new understanding of the effects of these states on detailed timing indicators of speech motor control, and new signal processing and machine learning methods that best exploit it. Expected outcomes from this project include ....Automatic detection and modelling of acoustic markers of speech timing. This project aims to create new automatic sensing, analysis and assessment of cognitive, affective, mental and physical state from voice for mobile and computing devices. This project expects to generate new understanding of the effects of these states on detailed timing indicators of speech motor control, and new signal processing and machine learning methods that best exploit it. Expected outcomes from this project include a new and accurate deep neural network framework for learning, analysing and detecting human states from speech automatically using articulatory timing markers. This should provide significant benefits, such as individually-tailored, frequent and low-cost automatic detection, monitoring and analytics for adverse states.Read moreRead less
Intelligence and national security: ethics, efficacy and accountability. This project aims to generate an ethically informed set of practice and policy guidelines for viable security intelligence collection and analysis of electronic data by liberal democracies. In the context of global terrorism and the resurgence of technologically sophisticated authoritarian states, effective intelligence collection and analysis of electronic data is crucial for the national security of liberal democratic sta ....Intelligence and national security: ethics, efficacy and accountability. This project aims to generate an ethically informed set of practice and policy guidelines for viable security intelligence collection and analysis of electronic data by liberal democracies. In the context of global terrorism and the resurgence of technologically sophisticated authoritarian states, effective intelligence collection and analysis of electronic data is crucial for the national security of liberal democratic states. Yet intelligence agencies in Australia, United States, European Union and so on, are not only under pressure to perform, but must also meet a variety of ethical challenges, notably privacy constraints and democratic accountability. This project will contribute to Australia's national security policy making environment, and to privacy and broader human rights debates, by providing an evidenced based, ethically informed set of practice and policy guidelines for viable national security intelligence practice in liberal democracies.Read moreRead less
Visual Simultaneous Localisation and Mapping in Deformable Environments. This project aims to investigate the problem of building a three-dimensional map of a deformable environment in real-time using images and at the same time localising the camera within the map. This project expects to generate new knowledge in the area of simultaneous localisation and mapping in deformable environments using visual sensors. Expected outcomes include in-depth understanding of the fundamental sensing requirem ....Visual Simultaneous Localisation and Mapping in Deformable Environments. This project aims to investigate the problem of building a three-dimensional map of a deformable environment in real-time using images and at the same time localising the camera within the map. This project expects to generate new knowledge in the area of simultaneous localisation and mapping in deformable environments using visual sensors. Expected outcomes include in-depth understanding of the fundamental sensing requirements for the problem to be solvable, the achievable accuracy, and efficient algorithms for achieving accurate three-dimensional reconstruction of deformable environments. The research outcomes from this project offer significant benefits to diverse areas such as minimally invasive robotic surgery.Read moreRead less
Advanced Machine Learning with Bilevel Optimization. There is an urgent need to develop a new machine learning (ML) paradigm that can overcome data-privacy and model-size constraints in real-world applications. This project aims to develop an advanced paradigm of ML with bilevel optimisation, called bilevel ML. A theoretically-guaranteed fast approximate solver and a new fuzzy bilevel learning framework will be developed to achieve the aim in complex situations; a methodology to transfer knowled ....Advanced Machine Learning with Bilevel Optimization. There is an urgent need to develop a new machine learning (ML) paradigm that can overcome data-privacy and model-size constraints in real-world applications. This project aims to develop an advanced paradigm of ML with bilevel optimisation, called bilevel ML. A theoretically-guaranteed fast approximate solver and a new fuzzy bilevel learning framework will be developed to achieve the aim in complex situations; a methodology to transfer knowledge and an approach to fast-adapt bilevel optimization solutions when required computing resources change. The anticipated outcomes should significantly improve the reliability of ML with benefits for safety learning and computing resource optimisation in ML-based data analytics.Read moreRead less
Robotic Perception with Unconventional Sensors . Autonomy in robotic systems currently relies on conventional sensors such as lasers and cameras. Alternative sensing modalities as in the case of active electromagnetic sensors are commonly used to detect flaws, cracks and assess infrastructure’s integrity, however, fundamental research questions preclude their use for robotic perception. This project will develop the theory and algorithms to enable perception tasks such as localisation, mapping a ....Robotic Perception with Unconventional Sensors . Autonomy in robotic systems currently relies on conventional sensors such as lasers and cameras. Alternative sensing modalities as in the case of active electromagnetic sensors are commonly used to detect flaws, cracks and assess infrastructure’s integrity, however, fundamental research questions preclude their use for robotic perception. This project will develop the theory and algorithms to enable perception tasks such as localisation, mapping and recognition with unconventional sensors. The outcomes of this research have the potential to improve the effectiveness of critical civil infrastructure maintenance technology through accurate and reliable inspections, and the reduced need for human intervention.Read moreRead less
Toward Human-guided Safe Reinforcement Learning in the Real World. This project aims to investigate human-guided safe reinforcement learning (RL). Safe RL is an important topic that could enable real applications of RL systems by addressing safety constraints. Existing safe RL assumes the availability of specified safety constraints in mathematical or logical forms. This project proposes to study learning safety objectives from information provided directly by humans or indirectly via language m ....Toward Human-guided Safe Reinforcement Learning in the Real World. This project aims to investigate human-guided safe reinforcement learning (RL). Safe RL is an important topic that could enable real applications of RL systems by addressing safety constraints. Existing safe RL assumes the availability of specified safety constraints in mathematical or logical forms. This project proposes to study learning safety objectives from information provided directly by humans or indirectly via language models, and human-guided continuous correction for safety improvements. The established theories and developed algorithms will advance frontier technologies in AI and contribute to a wide range of real applications of safe RL, such as robotics and autonomous driving, bringing enormous social and economic benefits. Read moreRead less
Evaluating the Network Neuroscience of Human Cognition to Improve AI. This project will translate the brain’s inherent complexity into a set of explorable networks that will test the network theory of intelligence, and also be used to drive advances in next generation artificial neural networks. Our approach will catalyse new knowledge regarding how the complexity of the brain gives rise to cognition using innovative analyses inspired by physics and engineering. This fresh perspective on cogniti ....Evaluating the Network Neuroscience of Human Cognition to Improve AI. This project will translate the brain’s inherent complexity into a set of explorable networks that will test the network theory of intelligence, and also be used to drive advances in next generation artificial neural networks. Our approach will catalyse new knowledge regarding how the complexity of the brain gives rise to cognition using innovative analyses inspired by physics and engineering. This fresh perspective on cognition will accelerate understanding of normal cognitive function and also advance the development of advances in artificial neural network performance. Expected outcomes include methods to describe the computational signature of how cognition emerges from dynamic brain network activity and novel AI algorithms. Read moreRead less
Adversarial Learning of Hybrid Representation. This project aims to design and implement a foundational deep representation learning framework for early detection, classification and defense of emerging malware by capturing their underlying behaviours via structured and unstructured heterogeneous information through hybrid representation learning, behaviour graph mining, and symbolic adversarial learning to discover and defend unknown malware families, thereby significantly boosting the accuracy ....Adversarial Learning of Hybrid Representation. This project aims to design and implement a foundational deep representation learning framework for early detection, classification and defense of emerging malware by capturing their underlying behaviours via structured and unstructured heterogeneous information through hybrid representation learning, behaviour graph mining, and symbolic adversarial learning to discover and defend unknown malware families, thereby significantly boosting the accuracy and robustness of existing classifiers and detectors. The resulting representation learning framework will enhance the national security to protect user privacy, reducing the multi-million-dollar loss caused by fraudulent transactions, and defending against cyber attacks.Read moreRead less
Generative Visual Pre-training on Unlabelled Big Data. This project aims to develop a generative visual pre-training of large-scale deep neural networks on unlabelled big data. Developing pre-trained visual models that are accurate, robust, and efficient for downstream tasks is a keystone of modern computer vision, but it poses challenges and knowledge gaps to existing unsupervised representation learning. Expected outcomes include new theories and algorithms for unsupervised visual pre-training ....Generative Visual Pre-training on Unlabelled Big Data. This project aims to develop a generative visual pre-training of large-scale deep neural networks on unlabelled big data. Developing pre-trained visual models that are accurate, robust, and efficient for downstream tasks is a keystone of modern computer vision, but it poses challenges and knowledge gaps to existing unsupervised representation learning. Expected outcomes include new theories and algorithms for unsupervised visual pre-training, which are anticipated to deepen our understanding of visual representation and make it easier to build and deploy computer vision applications and services. Examples of benefits include modernising machines in manufacturing and farming with visual intelligence. Read moreRead less