A safety-preserving ecosystem for autonomous driving. In this project, Macquarie University will collaborate with UTS and SilverQuest to develop an innovative safety-preserving ecosystem for autonomous driving. This system will not only be adopted by SilverQuest’s customers (automotive companies) to secure their latest autonomous driving models, but also be commercialised as a toolset that can be plugged into existing autonomous vehicles to detect and prevent malicious attacks on autonomous driv ....A safety-preserving ecosystem for autonomous driving. In this project, Macquarie University will collaborate with UTS and SilverQuest to develop an innovative safety-preserving ecosystem for autonomous driving. This system will not only be adopted by SilverQuest’s customers (automotive companies) to secure their latest autonomous driving models, but also be commercialised as a toolset that can be plugged into existing autonomous vehicles to detect and prevent malicious attacks on autonomous driving models. The project will lead to two innovations: in theory design an attack detection and prevention ecosystem for autonomous driving and in application implement a safety analysis toolset for industry-scale autonomous systems.Read moreRead less
Discovery Early Career Researcher Award - Grant ID: DE210100274
Funder
Australian Research Council
Funding Amount
$415,675.00
Summary
Graph Neural Networks for Efficient Decision-making towards Future Grids. This project aims to develop a breakthrough framework for decision-focused learning by integrating explainable graph neural networks and efficient computational methods. It expects to create new methodologies of graph representation learning for unlocking data insight with spatiotemporal knowledge while to build new accelerated optimisation theories for speeding up decision-focused learning model. The expected outcomes wil ....Graph Neural Networks for Efficient Decision-making towards Future Grids. This project aims to develop a breakthrough framework for decision-focused learning by integrating explainable graph neural networks and efficient computational methods. It expects to create new methodologies of graph representation learning for unlocking data insight with spatiotemporal knowledge while to build new accelerated optimisation theories for speeding up decision-focused learning model. The expected outcomes will advance big spatiotemporal data analytics and nonlinear optimisation theory for solving decision-making tasks towards a future energy system. This should promote the Australian power industry transition to a sustainable future grid based on a digitalisation approach to efficient energy management against climate changes.Read moreRead less
Discovery Early Career Researcher Award - Grant ID: DE200100245
Funder
Australian Research Council
Funding Amount
$410,518.00
Summary
Bayesian nonparametric learning for practical sequential decision making. This project aims to develop new methods to support practical sequential decision making under uncertainty. It expects to pave the way for the next generation of sequential decision making uniquely characterised by uncertainty modelling, high sample-efficiency, efficient environment change adaptation, and automatical reward function learning. The expected outcomes will advance machine learning knowledge with a new deep lea ....Bayesian nonparametric learning for practical sequential decision making. This project aims to develop new methods to support practical sequential decision making under uncertainty. It expects to pave the way for the next generation of sequential decision making uniquely characterised by uncertainty modelling, high sample-efficiency, efficient environment change adaptation, and automatical reward function learning. The expected outcomes will advance machine learning knowledge with a new deep learning schema for data modelling and sequential decision-making knowledge with a novel deep reinforcement learning methodology. These developments have immediate applications in autonomous vehicles, advanced manufacturing, and dynamic pricing, with scientific, economic, and social benefits for Australia and the world.Read moreRead less
Transfer Learning for Genome Analysis and Personalised Recommendation. This project aims to improve the accuracy, adaptability, and comprehensiveness of health characteristic predictions and provide personalised recommendations for healthcare service and disease prevention. The deliverables include uncertainty learning and multi-source transfer learning methodologies for predictions based on genome analysis that distils and transfers useful knowledge from multiple sources into an Australian geno ....Transfer Learning for Genome Analysis and Personalised Recommendation. This project aims to improve the accuracy, adaptability, and comprehensiveness of health characteristic predictions and provide personalised recommendations for healthcare service and disease prevention. The deliverables include uncertainty learning and multi-source transfer learning methodologies for predictions based on genome analysis that distils and transfers useful knowledge from multiple sources into an Australian genome analysis model. A federated cross-domain recommender system will be developed to profile individuals and generate personalised recommendations. The outcomes are expected to create a paradigm shift in learning-based prediction and personalised recommendations to support healthcare services in complex environments. Read moreRead less
Sequential decision-making in dynamic and uncertain environments. Current machine learning and optimisation methods cannot well support sequential prediction and decision-making due to the dynamic nature and pervasive presence of big data. This project aims to create a foundation and technology for sequence and uncertainty learning, sequential and dynamic optimisation, and their integration. It is expected to improve robustness and mitigate the vulnerabilities of machine learning algorithms, to ....Sequential decision-making in dynamic and uncertain environments. Current machine learning and optimisation methods cannot well support sequential prediction and decision-making due to the dynamic nature and pervasive presence of big data. This project aims to create a foundation and technology for sequence and uncertainty learning, sequential and dynamic optimisation, and their integration. It is expected to improve robustness and mitigate the vulnerabilities of machine learning algorithms, to increase prediction accuracy and reliability in dynamic sequences, and to support decision-making in complex situations to achieve robust and adaptive results. Anticipated outcomes can help data scientists with state-of-the-art skills to manage sequential data and benefit data-enabled innovation in Australia.Read moreRead less
Discovery Early Career Researcher Award - Grant ID: DE220101075
Funder
Australian Research Council
Funding Amount
$415,820.00
Summary
Fuzzy transfer learning for real-time decision making under uncertainty. This project’s objective is to build new tools for the next generation of real-time decision making. As the datasphere grows more complex, meaningful decision support already requires a strong capacity for knowledge transfer, substantial robustness to uncertainty, and real-time analytics. Today’s methods are struggling to meet these challenges. The new schema to be devised combines fuzzy logic, transfer learning, reinforcem ....Fuzzy transfer learning for real-time decision making under uncertainty. This project’s objective is to build new tools for the next generation of real-time decision making. As the datasphere grows more complex, meaningful decision support already requires a strong capacity for knowledge transfer, substantial robustness to uncertainty, and real-time analytics. Today’s methods are struggling to meet these challenges. The new schema to be devised combines fuzzy logic, transfer learning, reinforcement learning and deep neural networks. These integrations will lay the foundations for real-time decision-making solutions over the next decade and will advance machine learning under uncertainty. Immediate applications include structural health monitoring, climate prediction and telecommunications maintenance. Read moreRead less
Robust meta learning for risk-aware recommender systems. Recommender systems are the core of many online services but they are highly vulnerable to risks like shilling attacks, privacy leaks, and unexpected change. This project aims to develop new adversarial Bayesian-based, privacy-preserved and self-adaptive fuzzy meta learning methods and meta recommender systems that are robust to these risky, uncertain and dynamic environments. The anticipated outcomes should significantly improve the relia ....Robust meta learning for risk-aware recommender systems. Recommender systems are the core of many online services but they are highly vulnerable to risks like shilling attacks, privacy leaks, and unexpected change. This project aims to develop new adversarial Bayesian-based, privacy-preserved and self-adaptive fuzzy meta learning methods and meta recommender systems that are robust to these risky, uncertain and dynamic environments. The anticipated outcomes should significantly improve the reliability of recommender systems with particular benefits for online personalised service systems, e.g., e-government, e-business and e-Learning. The outcomes will also advance machine learning knowledge with a new robust meta learning schema for general data analytics and applications.Read moreRead less
Australian Laureate Fellowships - Grant ID: FL190100149
Funder
Australian Research Council
Funding Amount
$3,280,000.00
Summary
Autonomous learning for decision making in complex situations. The project aims to create a novel research direction – autonomous machine learning for data-driven decision-making – that innovatively and effectively learns from big data to support decision-making in complex (massive, uncertain, dynamic) situations. A set of new theories, methodologies and algorithms will give artificial intelligence the ability to learn autonomously from data to enable machine learning capability to effectively h ....Autonomous learning for decision making in complex situations. The project aims to create a novel research direction – autonomous machine learning for data-driven decision-making – that innovatively and effectively learns from big data to support decision-making in complex (massive, uncertain, dynamic) situations. A set of new theories, methodologies and algorithms will give artificial intelligence the ability to learn autonomously from data to enable machine learning capability to effectively handle tremendous uncertainties in data, learning processes and decision outputs, particularly enabling smart learning in massive domains, massive streams, and massive-agent sequentially changing environments. The project’s outcomes are expected to improve data-driven decision-making in multiple industry sectors.Read moreRead less
Discovery Early Career Researcher Award - Grant ID: DE200101310
Funder
Australian Research Council
Funding Amount
$426,918.00
Summary
Dimension-reduced Reinforcement Learning for Large-scale Fleet Management. This project aims to address the problems in large-scale fleet management to ensure the efficiency of tomorrow’s transportation models, such as on-demand ride-hailing and mobility-as-a-service. The expected outcomes of this project include improved techniques for optimising the utility of large fleets of vehicles, and particularly robust dimension-reduced reinforcement learning algorithms that are capable of handling the ....Dimension-reduced Reinforcement Learning for Large-scale Fleet Management. This project aims to address the problems in large-scale fleet management to ensure the efficiency of tomorrow’s transportation models, such as on-demand ride-hailing and mobility-as-a-service. The expected outcomes of this project include improved techniques for optimising the utility of large fleets of vehicles, and particularly robust dimension-reduced reinforcement learning algorithms that are capable of handling the complex dynamics of supply and demand in transportation. The results should advance both research and technology in academia and the transportation industry and will also provide significant benefits to Australia and the international community by enhancing the energy-efficiency of and access to the mobility of the future.Read moreRead less
Discovery Early Career Researcher Award - Grant ID: DE200100479
Funder
Australian Research Council
Funding Amount
$427,116.00
Summary
A Unified Framework to Rapidly Fabricate Individualised Activity Sensors. This proposal aims to develop a unified computational framework which enables non-expert users to co-design and fabricate specialised physical activity sensors to address individualised sensing problems in applications such as rehabilitation, age-care and sports. Specifically, we will develop an analytical framework to classify complex sensing problems into fabricable primitive classes, namely i) conditional – limits of ac ....A Unified Framework to Rapidly Fabricate Individualised Activity Sensors. This proposal aims to develop a unified computational framework which enables non-expert users to co-design and fabricate specialised physical activity sensors to address individualised sensing problems in applications such as rehabilitation, age-care and sports. Specifically, we will develop an analytical framework to classify complex sensing problems into fabricable primitive classes, namely i) conditional – limits of activity, ii) differential – frequency of activity and iii) integrational – cumulative activity. And a co-design interface to synthesize them into complex activity sensors to fit individualised needs. Finally, we will evaluate the framework by deploying the created sensors in real-world settings and gathering data.Read moreRead less