Exploring Bad Minima in Transfer Learning

Abstract:Transfer learning is a highly successful learning paradigm in deep learning which leverages knowledge from one task and applies it to another. Despite its success, the necessary conditions for knowledge transfer are unclear and require further research. In this dissertation, our aim is to understand transferability a little better by answering the following question: Does good generalization on a source task necessarily imply transferability to a related target task? We hypothesize the answer is negative, and attempt to find counterexamples via adversarial attacks on the transfer learning procedure. Over three types of adversarial attacks, our empirical results show a strong relationship between generalization on the source task and transferability. These results do not rule out generalization on the source task as a sufficient condition for successful transfer.

57 pages, 13 figures, 3 tables. MPhil thesis, University of Cambridge, 2022. Supervised by Ferenc Huszár

Machine Learning (cs.LG); Cryptography and Security (cs.CR)

arXiv:2610.41807 [cs.LG]

(or arXiv:2610.41807v1 [cs.LG] for this version)

https://doi.org/10.48550/arXiv.2610.41807 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.