STEEP Seminar :
Date / time: January 28, 2019 – 11:00 AM
Location : F107 – building INRIA Montbonnot
Title : From medical imaging to climate: machine learning for social good
Lecturer: Sophie Giffard Roisin, post-doctoral researcher at the University of Colorado, Boulder.
By building collaborations with experts in specific areas, we can use the strength of artificial intelligence to help making diagnosis, and also finding concrete and local solutions for our societies. I will first talk about my PhD work in medical imaging and non-invasive cardiac techniques for surgery prediction at Inria Sophia-Antipolis. Together with cardiologists, I developed a non-invasive cardiac electrophysiological model personalisation and I generated a simulated database for training a sparse Bayesian regression. While there is still a strong need for machine learning in medicine, I also realized how it was nearly absent in other critical domains such as climate and computational sustainability. During my post-doctoral position (CNRS Saclay and University of Colorado), I worked on predicting hurricane trajectories from image-like meteorological measures. We developed a fusion of convolutional neural networks, combining past trajectory data and reanalysis atmospheric images (wind and pressure 3D fields). A comparison with current forecast models shows that deep methods could provide a valuable and complementary prediction.