Declarative Networks: Towards Robust and Explainable Deep Learning. The aim of this project is to develop declarative machine learning techniques that exploit inherent structure and models of the world. Deep learning has become the dominant approach for machine learning with many products and promises built on this technology. But deep learning is expensive, opaque, brittle and relies solely on human labelled data. This project intends to make deep learning more reliable by establishing theory a ....Declarative Networks: Towards Robust and Explainable Deep Learning. The aim of this project is to develop declarative machine learning techniques that exploit inherent structure and models of the world. Deep learning has become the dominant approach for machine learning with many products and promises built on this technology. But deep learning is expensive, opaque, brittle and relies solely on human labelled data. This project intends to make deep learning more reliable by establishing theory and algorithms that allow physical and mathematical models to be embedded within a deep learning framework, providing performance guarantees and interpretability. This would likely benefit machine learning based products that can understand the world and interact with humans naturally through vision and language.Read moreRead less
Data sharing with strong privacy against inference attacks. This project aims to develop theories and techniques for strong protection of personal information in sharing large datasets such as national health data or census records. It intends to achieve this through developing new information theoretic methods for synthesising datasets with proven high fidelity and protection against re-identification and inference attacks, where attackers try to learn probability of sensitive data. The expecte ....Data sharing with strong privacy against inference attacks. This project aims to develop theories and techniques for strong protection of personal information in sharing large datasets such as national health data or census records. It intends to achieve this through developing new information theoretic methods for synthesising datasets with proven high fidelity and protection against re-identification and inference attacks, where attackers try to learn probability of sensitive data. The expected outcomes are algorithms for public and private sector data curators to dial up or down their data access arrangements based on privacy risks and fidelity demands linked with different data types and uses. This project intends to enable Australians to securely benefit from valuable data in decision making.Read moreRead less