Abstract

Bio: Olivia Wiles is a Staff Research Scientist at DeepMind working on robustness and evaluation of large models in machine learning, focussing on application driven research ranging from medical applications to large, multimodal foundational models. Prior to this, she was a PhD student at Oxford with Andrew Zisserman studying self-supervised representations for 3D.

Abstract: AI breakthroughs for medical applications are happening at pace, but it is important to consider how to ensure trustworthiness of these solutions before adoption. While true for ML as a whole , these questions are especially vital in the medical domain. I will discuss two angles of trustworthiness—fairness/robustness and privacy—and how we can build solutions that aim to ensure these requirements are met from the ground up by leveraging generative models. While these approaches show promising results, they are not a panacea, and a holistic approach is required to identify and mitigate challenges for deploying AI solutions in medical applications.