MemberGuard: Protecting Machine Learning Privacy from Membership Inference. Machine Learning has become a core part of many real-world applications. However, machine learning models are vulnerable to membership inference attacks. In these attacks, an adversary can infer if a given data record has been part of the model's training data. In this project, the team aims to develop new techniques that can be used to counter these attacks, such as 1) new analytical models for membership leakage, 2) ne ....MemberGuard: Protecting Machine Learning Privacy from Membership Inference. Machine Learning has become a core part of many real-world applications. However, machine learning models are vulnerable to membership inference attacks. In these attacks, an adversary can infer if a given data record has been part of the model's training data. In this project, the team aims to develop new techniques that can be used to counter these attacks, such as 1) new analytical models for membership leakage, 2) new methods for susceptibility diagnosis, 3) new defences that leverage privacy and utility. Data-oriented services are estimated to be valuable assets in the future. These techniques can help Australia gain cutting edge advantage in machine learning security and privacy and protect its intellectual property on these services.Read moreRead less
New methods for drug discovery by NMR spectroscopy. This project aims to advance nuclear magnetic resonance (NMR) spectroscopy methods in the field of drug discovery. It addresses a long-standing bottleneck for medicinal chemists in drug development: the rapid determination of how ligand molecules bind to proteins, where they bind and their orientation in the binding site. The methods include techniques for the attachment of NMR tags to ligands and target proteins, installation of new unnatural ....New methods for drug discovery by NMR spectroscopy. This project aims to advance nuclear magnetic resonance (NMR) spectroscopy methods in the field of drug discovery. It addresses a long-standing bottleneck for medicinal chemists in drug development: the rapid determination of how ligand molecules bind to proteins, where they bind and their orientation in the binding site. The methods include techniques for the attachment of NMR tags to ligands and target proteins, installation of new unnatural amino acids in proteins, and software for automated assignment of NMR spectra and 3D structure modelling of proteins using sparse distance restraints measured by electron paramagnetic resonance (EPR) spectroscopy. The outcome is to benefit the early stages of drug discovery in the biotech industries.Read moreRead less
Privacy-preserving data processing on the cloud. This project aims to address the current lack of privacy of user data processed by common cloud computing web servers, including email, business data, and confidential files. This project aims to develop new techniques in cryptography. The anticipated outcome is a suite of practical tools enabling common cloud computing processing operations such as search, statistical analysis, and multi-user access control, to be performed efficiently while pres ....Privacy-preserving data processing on the cloud. This project aims to address the current lack of privacy of user data processed by common cloud computing web servers, including email, business data, and confidential files. This project aims to develop new techniques in cryptography. The anticipated outcome is a suite of practical tools enabling common cloud computing processing operations such as search, statistical analysis, and multi-user access control, to be performed efficiently while preserving the data privacy. These tools should provide significant benefits to the privacy of cloud users, as well as financial and reputation benefits to the IT industry, by significantly reducing the likelihood of massive user data privacy breaches in the event of a cyber-hacking attack on the cloud server.Read moreRead less
Verified concurrent memory management on modern processors. This project aims to formally verify automatic memory managers in the presence of concurrency and the weakly ordered memory of modern processors. A new framework for verifying memory managers, reusable for a wide range of managed programming languages, target hardware, policies, and algorithms will be developed. Expected technical outcomes include improved techniques to ensure trustworthiness of the foundations on which critical softwar ....Verified concurrent memory management on modern processors. This project aims to formally verify automatic memory managers in the presence of concurrency and the weakly ordered memory of modern processors. A new framework for verifying memory managers, reusable for a wide range of managed programming languages, target hardware, policies, and algorithms will be developed. Expected technical outcomes include improved techniques to ensure trustworthiness of the foundations on which critical software infrastructures are built. This will significantly enhance the security of public and private cyber assets, and deliver applications that are more robust and trustworthy, across a range of critical infrastructure such as transportation, communication, energy and defence.Read moreRead less
Next generation garbage collection: discovery, design, and development. This project aims to improve the performance of programming languages used by millions of Australians every day, such as Java, JavaScript and PHP by developing improved memory-management algorithms. These languages use what is referred to as “garbage collection” to ensure memory is managed without data loss, but do so conservatively and consequently cause performance challenges and energy overheads. This project expects to p ....Next generation garbage collection: discovery, design, and development. This project aims to improve the performance of programming languages used by millions of Australians every day, such as Java, JavaScript and PHP by developing improved memory-management algorithms. These languages use what is referred to as “garbage collection” to ensure memory is managed without data loss, but do so conservatively and consequently cause performance challenges and energy overheads. This project expects to provide these languages with improved memory-management algorithms, and provides researchers and industry with a framework for innovation. This project will enable safe software that is more efficient on today's hardware and able to exploit emerging hardware. This project should lead to better performance and energy savings for server applications, phones, watches, and smart appliances, while ensuring memory safety.Read moreRead less
Automatic Training Data Search and Model Evaluation by Measuring Domain Gap. We aim to investigate computer vision training data and test data, using automatically generated data sets for facial expression recognition and object re-identification. This project expects to quantify and understand the domain gap, the distribution difference between training and test data sets. Expected outcomes of this project are insights on measuring the domain gap, the ability to estimate model performance witho ....Automatic Training Data Search and Model Evaluation by Measuring Domain Gap. We aim to investigate computer vision training data and test data, using automatically generated data sets for facial expression recognition and object re-identification. This project expects to quantify and understand the domain gap, the distribution difference between training and test data sets. Expected outcomes of this project are insights on measuring the domain gap, the ability to estimate model performance without accessing expensive test labels and improvements to system generalisation. This should provide significant benefits for computer vision applications that currently require expensive labelling, and commercial and economic benefits across sectors such as transportation, security and manufacturing.Read moreRead less
Non-Canonical Amino Acids for Protein Analysis and Peptide Inhibitors. This interdisciplinary project aims to establish new tools to experimentally confirm 3D structure predictions of proteins that are otherwise difficult to study. A combination of innovative biochemistry, modern spectroscopy, and high-performance computing will be applied to study protein-protein and protein-ligand interactions. The project expects to generate new techniques and to test them on established drug targets. Expecte ....Non-Canonical Amino Acids for Protein Analysis and Peptide Inhibitors. This interdisciplinary project aims to establish new tools to experimentally confirm 3D structure predictions of proteins that are otherwise difficult to study. A combination of innovative biochemistry, modern spectroscopy, and high-performance computing will be applied to study protein-protein and protein-ligand interactions. The project expects to generate new techniques and to test them on established drug targets. Expected outcomes include new tools which quickly inform medicinal chemists how drugs interact with their targets and how they can be improved. The developed tools should provide significant benefit to many researchers by accelerating the early stage of drug discovery, and support Australia’s fast growing biotechnology sector.Read moreRead less
Protein Structure and Dynamics by Electron/Nuclear Paramagnetic Resonance. This interdisciplinary project aims to establish new magnetic resonance methods for the analysis of protein structure and motion at low concentrations and in physiological conditions that are otherwise difficult or impossible to study. It brings together four different research groups with expertise in advanced biochemistry, modern magnetic spectroscopy and high-performance computing. The project expects to develop tools ....Protein Structure and Dynamics by Electron/Nuclear Paramagnetic Resonance. This interdisciplinary project aims to establish new magnetic resonance methods for the analysis of protein structure and motion at low concentrations and in physiological conditions that are otherwise difficult or impossible to study. It brings together four different research groups with expertise in advanced biochemistry, modern magnetic spectroscopy and high-performance computing. The project expects to develop tools to study protein structure, protein-protein association and protein-ligand interactions of established drug-targets. Expected outcomes include new techniques that quickly inform how drugs work, providing significant benefits to many researchers studying biomolecules, and supporting Australia’s growing biotechnology sector. Read moreRead less
Learning to Pinpoint Emerging Software Vulnerabilities. This project aims to develop learning-based software vulnerability detection techniques to improve the reliability and security of modern software systems. The existing techniques relying on conventional yet rigid software analysis and testing techniques are ineffective and/or inefficient when detecting a wide variety of emerging software vulnerabilities. The outcomes of this project will be a deep-learning-based detection approach and an ....Learning to Pinpoint Emerging Software Vulnerabilities. This project aims to develop learning-based software vulnerability detection techniques to improve the reliability and security of modern software systems. The existing techniques relying on conventional yet rigid software analysis and testing techniques are ineffective and/or inefficient when detecting a wide variety of emerging software vulnerabilities. The outcomes of this project will be a deep-learning-based detection approach and an open-source tool that can capture precision correlations between deep code features and diverse vulnerabilities to pinpoint emerging vulnerabilities without the need for bug specifications. Significant benefits include greatly improved quality, reliability and security for modern software systems.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