Olivia Wiles

Olivia Wiles is a Senior Researcher at DeepMind, focusing on adversarial robustness, distribution shift, and computer vision. She earned her DPhil from the University of Oxford under Andrew Zisserman in the Visual Geometry Group (VGG), following a Computer Science degree at the University of Cambridge. Her work spans view synthesis, self-supervised learning, and robust model design.

• Education: DPhil (Oxford), Computer Science (Cambridge)
• Key Collaborators: Georgia Gkioxari, Justin Johnson, Richard Szeliski (FAIR), Andrew Zisserman
• Current Role: Senior Researcher at DeepMind

Her research emphasizes adversarial robustness, 3D reconstruction, and self-supervised learning, with notable contributions to view synthesis (SynSin), image matching (Co-Attention), and robustness under distribution shifts. Publications at CVPR, NeurIPS, and ECCV highlight her work in generative models, physical prediction, and multi-view geometry.

Scientific Awards:

• Best Poster, BMVC 2017
• Best Paper, NeurIPS ML Safety Workshop 2022
• Outstanding Reviewer, ECCV 2020, ICCV 2020/2021

Olivia contributes to academia via community service, including roles as Area Chair (CVPR, ICCV) and reviewer for top-tier conferences (NeurIPS, SIGGRAPH) and journals (PAMI). Her Google Scholar profile reveals ongoing work in unsupervised physics modeling, GANs, and intuitive physics from visual data.