Data: 4 de junho
Hora: 13h
Local: Anfiteatro PA2, Técnico – Campus Alameda
Orador: Catarina Barata (IST/ISR)
Título: “Human-Centered Explainable AI for Healthcare”
Resumo:
Healthcare research has changed dramatically in the last decade. From the development of top algorithms for patient diagnosis and profiling, including the recent advances on chatbots such as Med-Gemini, to the public release of increasingly larger and more challenging medical datasets, it is undeniable that we are living an exciting era for those working in the interface between machine learning and healthcare. However, working on safety critical applications brings an additional level of responsibility. Here, a prediction error may have dire consequences, as we often work with life or death situations. Thus, it becomes critical to understand our models, their decisions and failure modes. Moreover, this understanding should be grounded on the actual knowledge and needs of those that will work with the models: doctors, patients, insurance companies, or even the developers. In this talk I will present an overview of our path towards the development of healthcare models for cancer analysis. This path has followed two core ideas: i) human expectations and knowledge must be integrated in the model development; and ii) we must be able to understand the model decisions. This has led to the proposal of several approaches that combine a human-centered vision with the concept of explainable AI.
Nota biográfica:
Catarina Barata holds a Msc Degree in Biomedical Engineering and a PhD in Electrical and Computer Engineering (Instituto Superior Técnico – IST – 2011 and 2017 respectively). In the Fall of 2022, she was a Visiting Scholar at Carnegie Mellon University. Presently, she is a tenure-track Assistant Professor at the ECE Department of IST and a Researcher at Institute for Systems and Robotics (ISR), where she is a member of the Computer and Robot Vision Laboratory (VisLab). Her main research interests are in the interface between machine learning, computer vision, and healthcare, where she has been collaborating and leading various projects together with hospitals and other healthcare institutions. An example is her work on the discovery of therapeutic biomarkers for melanoma, for which she received a Google Research Award in 2021.
A Priberam integra a Comunidade IST Spin-Off®.
Os “Priberam Machine Learning Lunch Seminars” são de entrada livre, mediante inscrição prévia.
