Seminar by Dr. Hervé Lombaert (Inria Sophia-Antipolis, France)

Speaker: Dr. Hervé Lombaert
                Inria Sophia-Antipolis, France

Title: The Human Body as Big Data - Moving beyond conventional learning approaches

Date: Monday, February 15th, 2016

Time: 2:30 - 4PM

Place: EV3.309

ABSTRACT

Big Data, machine learning, and pattern analysis - One of the most fascinating Big Data system is perhaps the Human body.  It is studied using data from multiple sources, where terabytes of data can be created for single subjects.  Preeminent research in understanding, for example, the human mind now needs population studies that generate even larger amounts of data.  The analysis of such data reaches the limits of conventional computational methods and requires to go beyond our most advanced research techniques.  This presentation will show one example, on how new manifold learning methods can be used to drastically improve applications in neuroscience, and more importantly, on how to exploit Big Data in increasingly larger population studies.  The presentation will be illustrated using medical images but will remain general enough to cover broader applications in different fields that involve very large data.  Topics will cover spectral graph theory, random decision forests, and data correspondences between reduced embeddings.

BIO

Hervé Lombaert is now a Starting Researcher at Inria Sophia-Antipolis, France.  His research interests include the Geometry of Shapes and Data, using Machine Learning and Computer Vision techniques.  He earned his PhD in Computer Engineering at Polytechnique Montreal in 2012, and had the chance to progress in multiple centers, including Microsoft Research Cambridge, Siemens Corporate Research, INRIA Sophia-Antipolis, McGill University, and Polytechnique Montreal.