Can Machines Unlearn? Toward Next-Generation Safe Artificial Intelligence. This project aims to develop new principles, theories, and methods to unlearn undesirable artifacts from an Artificial Intell

Description

Can Machines Unlearn? Toward Next-Generation Safe Artificial Intelligence. This project aims to develop new principles, theories, and methods to unlearn undesirable artifacts from an Artificial Intelligence (AI) system. It expects to produce new knowledge, algorithms, tools, techniques, and intellectual property to a new field of machine unlearning. Expected outcomes can be used to efficiently implement responsible use of AI such as protecting users’ data from existing trained AI models or safeguarding contents generated by them. It will enhance Australia's leadership in AI research and practice; deliver trustworthy technology which benefits not only scientific and translational knowledge advancement but also in accelerating AI innovations and filling the AI skills demand in Australia.. Scheme: Discovery Projects. Field: 4611 - Machine Learning. Lead: Prof Dinh Phung