Trustworthy Transfer Learning: A Survey

The paper 'Trustworthy Transfer Learning: A Survey' published on arXiv on November 13, 2025, delves into the critical aspects of transfer learning, focusing on knowledge transferability and trustworthiness. It poses essential research questions regarding the quantitative measurement and enhancement of knowledge transfer across different domains, as well as the reliability of the transferred knowledge. The review encompasses various dimensions, including problem definitions, theoretical analyses, empirical algorithms, and practical applications. It emphasizes understanding knowledge transferability under IID and non-IID assumptions, while also addressing the impact of trustworthiness factors such as adversarial robustness, algorithmic fairness, and privacy-preserving constraints. By summarizing recent advancements and identifying open questions, the paper aims to pave the way for future research in creating reliable and trustworthy transfer learning systems.