{"links":{"self":"http://dataportal.arc.gov.au/NCGP/API/grants/DE260100019"},"data":{"type":"grant-details","id":"DE260100019","attributes":{"code":"DE260100019","administering-organisation":"RMIT University","announcement-administering-organisation":"RMIT University","scheme-name":"Discovery Early Career Researcher Award","grant-status":"Active","funding-commencement-year":2026,"years-funded":3,"project-start-date":"2026-01-01","anticipated-end-date":"2028-12-31","grant-summary":"Secure Deep Learning Inference with Privacy Protection. This project aims to investigate output privacy risks and develop corresponding mitigations for secure deep learning inference. This project expects to advance knowledge of how prediction outputs from secure inference are exploitable, the extent of privacy breaches, and strategies to safeguard output privacy. Expected outcomes of this project include a formal trust model characterising output privacy in secure inference, principled attack methodologies unveiling the risks, lightweight privacy-enhancing mitigation techniques, and a practical system solution for real-world applications. This should provide significant benefits such as facilitating AI-powered industries to uplift their businesses in a secure and trustworthy fashion.","funding-current":532907.00,"funding-at-announcement":528678,"investigators-current":[{"title":"Dr","firstName":"Xiaoning","familyName":"Liu","roleName":"Discovery Early Career Researcher Award","roleCode":"DECRA","isFellowship":true,"orcidIdentifier":"0000-0002-9874-8839 "}],"investigators-at-announcement":[{"title":"Dr","firstName":"Xiaoning","familyName":"Liu","roleName":"Discovery Early Career Researcher Award","roleCode":"DECRA","isFellowship":true,"orcidIdentifier":"0000-0002-9874-8839 "}],"organisations-current":[{"organisationName":"RMIT University","roleName":"Administering Organisation","state":"VIC"}],"organisations-at-announcement":[{"organisationName":"RMIT University","roleName":"Administering Organisation","state":"VIC"}],"field-of-research":[{"isPrimary":true,"code":"4604","name":"Cybersecurity and Privacy","type":"FOR20"},{"isPrimary":false,"code":"460403","name":"Data Security and Protection","type":"FOR20"},{"isPrimary":false,"code":"460407","name":"System and Network Security","type":"FOR20"},{"isPrimary":false,"code":"460505","name":"Database Systems","type":"FOR20"}],"socio-economic-objective":[{"code":"220302","name":"Electronic Information Storage and Retrieval Services","type":"SEO20"},{"code":"220405","name":"Cybersecurity","type":"SEO20"}],"international-collaboration":[],"lief-register":[],"achievement-summary":null,"national-interest-test-statement":"Secure deep learning inference enables organisations with AI models to provide Deep Learning as a Service (DLaaS) for their end-users without revealing sensitive data carried by the data and AI models to each other. It is considered as a key enabler of trustworthy AI services. Legitimate secure inference results can still be exploited by AI attacks to recover the AI model and its training data information, eventually resulting in an  output privacy breach. This project will investigate output privacy risks that hinder the practical deployment of secure deep learning inference and then enable new mitigation technologies to remedy the risks. The proposed technologies will directly benefit enterprises, governments, and Australian citizens by safeguarding their valuable AI models and private end-users data. The outcomes will enable Australia to strengthen cybersecurity and AI sectors and further promote disruptive technologies such as AI-powered applications in healthcare and finance. Beyond economic benefits, this project will enhance national cybersecurity capabilities thwarting AI-centred cybercrimes, alleviate public privacy concerns in adopting AI, and foster confidence in a trustworthy AI ecosystem. A practical system solution will be developed to maximise real-world adoption, complemented by collaborative demonstration projects and targeted training workshops to promote these advancements across digital and AI-driven industries."}}}