Nothing Special   »   [go: up one dir, main page]

Gauthier Gidel

Assistant Professor
Université de Montréal - DIRO
Core member of Mila Send an email

I am an assistant professor at Université de Montréal (UdeM) at DIRO and a core faculty member of Mila. I graduated my Ph.D. under the supervision of Simon Lacoste-Julien. During my Ph.D., I have been an intern at Sierra, ElementAI and DeepMind.

Link to my Google scholar.

Prospective Students

Please read this page before sending me any email.

Courses

Winter 2021 - Winter 2022 - Winter 2023: Adversarial Machine Learning -- IFT 6164 (former Game theory and Machine Learning -- IFT 6756).
Fall 2021 - Fall 2022: Data Science -- IFT 6758-A -- Science des Données.

[NEW!]

  • I am teaching the data science class in french this fall.

My Research

The abilities of ChatGPT, released in late 2022, were a shock for many in the machine learning community (including me). This breakthrough ignited an arms race focused on training large generative models with increasing capabilities. In particular, we can foresee a not-so-far future with ML systems empowered with interaction and agency capabilities. In light of these emerging phenomena, my research seeks to explore the forthcoming challenges of the transition of sophisticated ML systems into agents:
  • Adversarial Robustness of LLMs' Safety Alignment: How can we build safer LLMs and improve the evaluation of their robustness?
  • Cooperation and Negotiation in Multi-Agent Contexts: How can we design algorithms that can learn the long-term benefit of cooperation while still maintaining a high value for their objective? In a world where rational agents first serve their interests, negotiation protocols and robustness are essential to reaching a high level of cooperation.
  • Principled Learning Method for (Multi-Agent) RL: Can we design training methods that tackle the challenges of learning in non-stationary environments?
  • Risk and Benefit of Interaction with Synthetic Data: When and to what extent can synthetic data improve (or deteriorate) the performance of models trained on these data? How do generative models affect each other's learning when deployed in shared environments?

I identify to the fields of ML (JMLR, NeurIPS, ICML, AISTATS, COLT, and ICLR) and optimization (SIAM OP)

Students

Alumni

  • Quentin Bertrand (Tenured Research scientist at Inria Lyon)
  • Joey Bose (Postdoc at Oxford)
  • Bilun Sun, Intern
  • Chiara Régniez, Junior Data Scientist, Owkin
  • Leonardo Cunha, Software Engineer, Muvraline
  • Sunand Raghupathi, CEO of Seven Seas Strategies
  • Eduard Gorbunov, Research scientist at MBZUAI
  • Aleksandr Beznosikov Ph.D. at MIPT.
  • Michael Przystupa Ph.D. at University of Alberta.
  • Alexandre Duplessis M.Sc. at Oxford.
  • Chris Junchi Li (Visiting Postdoc co-supervised with Michael I. Jordan)

    • Preprints - Publications - Manuscripts

    With my currents responsabilitities, I have no time to update this website frequently. Thus, this List is not up-to-date, I decided to remove the list on my website, for my latest publications please check my Google Scholar

    Third Person Bio

    Gauthier Gidel is a Canada CIFAR AI Chair, a core member of Mila, and an assistant professor at the Department of Computer Science and Operations Research (DIRO) at Université de Montréal. His research focuses on generative modeling and multi-objective learning, such as alignment or adversarial robustness. He is well known for his work on variational inequality for machine learning. Gauthier co-organized a popular series of workshops on smooth games during NeurIPS. He also co-founded and co-organized the first three iterations of the ICLR blog post track.

    Contact

    Gauthier Gidel
    Pavillon André-Aisenstadt
    2920 Chemin de la Tour, office 3151
    Montreal, QC
    H3T 1J4 CANADA

    © 2019