Transfer Learning for Reliable Data Detection in Open-set Environments. There is an urgent need to develop a new machine learning scheme in open-set environments to enhance the reliability of machine

Description

Transfer Learning for Reliable Data Detection in Open-set Environments. There is an urgent need to develop a new machine learning scheme in open-set environments to enhance the reliability of machine learning models. This project aims to use transfer learning to enhance the reliability of machine learning models when encountering unfamiliar objects, which are known as out-of-distribution data. The project involves: developing novel machine learning theories to guide method design; novel frameworks that are distribution-robust to transfer knowledge from available related datasets; and novel compatibility-aware frameworks to transfer knowledge from available models. The outcomes are expected to enhance the reliability of machine learning, yielding benefits for responsible artificial intelligence.. Scheme: Discovery Early Career Researcher Award. Field: 4605 - Data Management and Data Science. Lead: Dr Zhen Fang