Eugene Belilovsky bio photo

Eugene Belilovsky

Postdoctoral Researcher in Machine Learning

Email Twitter Github

Eugene Belilovsky

I am a Postdoctoral Researcher at the MILA lab at the University of Montreal working with Aaron Courville. I recently completed a (joint) PhD at CentraleSupelec and KU Leuven supervised by Matthew Blaschko. During my PhD I primarily worked at the Center for Visual Computing and was also part of the VISICS lab. I also visited the University of Toronto Machine Learning Group working with Richard Zemel, and Raquel Urtasun and interned in the core machine learning groups at Apple and Amazon.

My primary research interests are in Machine Learning, Deep Learning, Computer Vision, and Graphical Models. My PhD thesis developed several machine learning methods with applications in limited sample setting such as those found in the analysis of neuro-imaging. During my PhD I have also had the great pleasure to work extensively on a broad scope of problems in machine learning, from deep generative modeling, multi-modal learning, bayesian optimization, to developing more efficient large scale image recognition and detection and working towards furthering our understanding of succesful deep learning methods in the context of computer vision.


  • I succesfully completed my PhD
  • I am co-organizing the NIPS 2017 Workshop: Learning with Limited Labeled Data: Weak Supervision and Beyond.
  • One paper accepted at ICCV 2017
  • One paper accepted at ICML 2017
  • New working paper and python package
  • Two papers accepted at ICLR 2017 workshop
  • I have received a NIPS travel award
  • One paper accepted for presentation at NIPS 2016
  • I have received the University Paris Saclay Best Contribution Award (2nd place)
  • I have received a MITACS-INRIA Globalink award to visit the University of Toronto with Raquel Urtasun and Richard Zemel
  • Two new working papers on graphical model structure discovery here and here
  • Received an ICLR student award
  • Our paper on model selection in deep generative models was accepted at ICLR 2016. link