Vivien Cabannes

PhD Candidate


Welcome to my webpage

Along my schooling, I have developed strong interest in rigorous knowledge offered by science, as well as their applications in the real-world. I found motivations in learning, building, interacting with bright people, and tackling complex challenges.

After a detour in the quantitative finance industry, this has led me to pursue a PhD in machine learning. My research is focus on building a framework to learn with weakly supervised data, which could significantly advance the current scope of machine learning.

During my spare time, you will find me involved in public life, through political organizations, and, to a lower extend, artistic reflexions. Finally, cooking and staying active are recreational activities that have the luck to enjoy in my daily routine.

You can find my old webpage including school projects and internship reports at https://www.lapasserelle.com/vivien/2017/html/.


  • Machine Learning
  • Weakly Supervised Learning
  • Structured Prediction


  • MSc in Machine Learning, 2017

    ENS Paris-Saclay

  • BSc in Mathematics, 2015

    ENS Paris

  • BSc in Computer Science, 2015

    ENS Paris


Laplacian regularization

Overcoming the curse of dimensionality with Laplacian regularization in semi-supervised learning, 2021, Preprint.

Disambiguating weak supervision

Disambiguation of weak supervision with exponential convergence rates, 2021, ICML.

Fast rates

Disambiguation of weak supervision with exponential convergence rates, 2021, COLT.

Perception diptychs

Diptychs of human and machine perceptions, 2020, NeurIPS workshop on creativity.

Partial supervision

Structured Prediction with Partial Labelling through the Infimum Loss, 2020, ICML.

Art dialog with A.I.

Dialog on a canvas with a machine, 2019, NeurIPS workshop on creativity.

Selected Experience


PhD Candidate


Sep 2019 – Present Paris, France
Machine learning requires extensive human annotations of data. My thesis addresses this issue by looking for ways to learn with weak source of information. It is theoretically oriented, yet has many potential applications. In particular, my predecessor has developed an algorithm to recognize key moments on Youtube instructional videos. My advisors are Francis Bach and Alessandro Rudi.

Research Analyst Intern

Cubist, Point 72

Jan 2018 – Aug 2018 New York, USA
As the first hire of a new team, I have assisted a portfolio manager to build an entire trading system. This includes data scrapping, cleaning, alpha generation, portfolio optimization and execution. I have returned during summer 2019 to add machine learning all along the team framework. The team was and is still highly profitable.

Short-Term Visiting Scholar

Duke University

Apr 2017 – Sep 2017 North Carolina, USA
I participated in the Autism & Beyond study, which aimed at developping an ergonomic application to help screening for autism in toddler. Under the supervision of Guillermo Sapiro, I have used of machine learning to analyse facial expression in order to learn and flag autistic behaviors.

Graduate Student

Ecole Normale Superieure

Sep 2014 – Apr 2018 Paris, France
Ecole Normale Superieure is made of distinguished students that are offered a position as civil-servants. Ranked 13th at the math entrance examination, I have obtained a bachelor in mathematics and a bachelor in computer science, as well as a master in Machine Learning (MVA) with highest honors (“Félicitations du Jury”, top 2%). I also had the luck to go on research internships at INRIA (10 months) and UCSD (5 months).

Office Directions

  • 2 rue Simone Iff, Paris, Ile-de-France, 75012, France
  • Enter Building C and take the elevators to Office 408