Timothy Hospedales

WORK HISTORY

EDUCATION

SKILLS

ABOUT TIMOTHY HOSPEDALES

I am a Full Professor at University of Edinburgh, and Head of Samsung AI Research, Europe.My research has focused on improving the data and computation efficiency of machine learning, AI models’ robustness to real-world distribution shifts, and recently AI trustworthiness guarantees and governance. I have addressed these issues in a variety of contexts including vision, embodied agents, “Generative AI” large language models and diffusion models, finance, and more. This work has led to over 100 publications in top-ranked highly selective journals and conferences (PAMI, NeurIPS, ICML, CVPR, ICCV, ICLR, IJCAI, etc), including several best paper prizes and nominations. Influential examples include the few-shot learner Relation Net, mutual learning for distillation, and TucKER approach to knowledge graph completion. I have been involved as Co-Investigator or Principal Investigator in several major academic projects. In the SPADS CDT (£8.7M, 2024-29), we are training 60 PhDs to become the next generation of scientists and engineers in Sensing, Processing, and AI for Defense and Security. In the UDRC Phase 3 (£4M, 2018-24), I led the machine learning track that developed cutting edge verifiable AI and generative AI methods in the security context. In ORCA (£15M, 2017-22), we developed autonomous robotics for offshore infrastructure maintenance. In DREAM (€2.7M, 2015-18) we developed robots that dramatically improve autonomy via dream-like processes of mental restructuring. In my industrial role at Samsung AI, I support teams across Europe in developing AutoML tools and efficient learning algorithms that accelerate research, as well as the next generation of lightweight Generative AI foundation models across diverse modalities, to underpin future Samsung products. Other Activities- Fellow- ELLIS Fellow- Member of UKRI Strategic Advisory Team: Robotics and AI- Associate Program Chair of 2022- Ex Associate Editor of T-PAMI- Google Scholar: