03 février 2020
12h30 - 14h00
Woluwé-Saint-Lambert
Auditoire Gerty Cori - Tour Laennec +1
Machine learning (ML) is the science of getting computers to act without being explicitly programmed. ML typically follows a data-driven methodology where models are built from observed data before making predictions on new data. This talk will present some of the challenges raised when applying ML to precision medicine, an area of medicine where decisions, treatment and follow-up are aimed to be tailored to each individual patient.
Prototypical examples will be briefly presented, including early diagnosis of undifferentiated arthritis, treatment response prediction of an immunotherapy against melanoma, or breast cancer prognosis. Such examples motivate the discovery of stable biomarkers that should be informative and predictive of the task at hand. We will show that a direct brute-force approach to this problem is intrinsically intractable, hence calling for dedicated computational methods.
We conclude our discussion by stressing that the current Artificial Intelligence “revolution” is mostly attributed to machine learning and, more specifically, to so-called deep learning approaches. Yet, we will argue why deep learning is currently not the most promising way to address precision medicine from omics data.
Date: February 3, 2020, 12:30 PM (followed by walking lunch)
Venue: Auditoire Gerty Cori (Tour Laennec +1, Campus Woluwe)
Speaker Biography
Pierre Dupont received a Ph.D. in Computer Science from l’Ecole Nationale Supérieure des Télécommunications, Paris. He has been a post-doctoral researcher at Carnegie Mellon University, Pittsburgh, PA, USA. He is now Full Professor at the Louvain School of Engineering, UCLouvain, and member of the Institute for Information and Communication, Electronics and Applied Mathematics. He is co-founder of DNAlytics, a UCLouvain spin-off developing innovative data-driven approaches to precision medicine.
His current research interests include machine learning methods applied to computational biology, feature selection methods, biomarker discovery from omics data, survival data analysis and bio-statistics.