PeopleofSuperintelligence

Researcher

Siamak Ravanbakhsh

Siamak Ravanbakhsh is a machine learning researcher whose work explores symmetry in neural networks and also includes reinforcement learning. He earned a Ph.D. in computing science at the University of Alberta in 2015; his dissertation on message passing and combinatorial optimization received the Faculty of Science Doctoral Dissertation Award. In 2017, he proposes parameter-sharing schemes for equivariant neural networks, and his 2020 paper proves universality for a class of equivariant multilayer perceptrons. A 2023 paper examines symmetry breaking while another uses Koopman theory for interactive-environment dynamics; as of October 2026, he is an associate professor at McGill, a core academic member of Mila, and a Canada CIFAR AI Chair.

Works
86

Publication numbers from OpenAlex, as of 27 October 2026. Citations and h-index are hidden because this OpenAlex record may be incomplete or mix in other authors.

Timeline

  1. 2009

    An M.Sc. thesis in optimization

    In 2009, he completes an M.Sc. at the University of Alberta with a thesis on stochastic optimization for NMR spectra; the thesis is nominated for the best M.Sc. thesis award.

  2. 2015

    Doctoral thesis recognition

    He earns a Ph.D. in computing science at the University of Alberta in 2015. The university names his dissertation, Message Passing and Combinatorial Optimization, for its Faculty of Science Doctoral Dissertation Award; Russ Greiner supervises the thesis.

  3. By October 2017

    From CMU to UBC

    By October 2017, he is a faculty member in UBC’s Department of Computer Science after a postdoctoral fellowship at Carnegie Mellon’s Machine Learning Department and Robotics Institute. His research focuses on inference in structured, complex, and combinatorial domains, with applications in astronomy and cosmology.

  4. 2017

    Parameter sharing for equivariance

    In 2017, with Jeff Schneider and Barnabás Póczos, he proposes parameter-sharing schemes that make neural network layers equivariant to discrete group symmetries.

  5. August 2019

    McGill School of Computer Science

    In August 2019, he becomes an assistant professor at McGill's School of Computer Science after a similar faculty position at UBC.

  6. 2020

    Universal equivariant MLPs

    In 2020, he publishes Universal Equivariant Multilayer Perceptrons at ICML; the paper proves that a broad class of equivariant MLPs with one hidden layer is universal.

  7. 2023

    Symmetry breaking and RL models

    In 2023, with Sékou-Oumar Kaba, he introduces relaxed equivariance to address the inability of equivariant functions to break symmetry at individual data samples. A second 2023 paper uses Koopman theory to model interactive-environment dynamics for model-based planning and model-free reinforcement learning.

  8. As of October 2026

    Associate professor at McGill

    As of October 2026, he is an associate professor at McGill's School of Computer Science, a core academic member of Mila, and a Canada CIFAR AI Chair.

Career

  1. NowMcGill UniversityAssociate Professor
  2. NowMilaCore Academic Member
  3. From 2019McGill UniversityAssistant Professor
  4. Carnegie Mellon UniversityPostdoctoral Fellow
  5. University of British ColumbiaFaculty Member

Education

  1. 2015University of AlbertaPh.D. in computing science
  2. 2009University of AlbertaM.Sc.

From the biography.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Siamak Ravanbakhsh yet. Send us a free-licensed photo.

  1. Publications — Siamak Ravanbakhsh cs.mcgill.ca. Checked 27 October 2026.
  2. Rouzbeh Maani, Martha White, Siamak Ravanbakhsh to receive Faculty of Science Dissertation Awards gsa.ualberta.ca. Checked 27 October 2026.
  3. UBC Computer Science Welcomes Eight New Faculty Members cs.ubc.ca. Checked 27 October 2026.
  4. Equivariance Through Parameter-Sharing arxiv.org. Checked 27 October 2026.
  5. Reasoning and Learning Lab | Siamak Ravanbakhsh rl.cs.mcgill.ca. Checked 27 October 2026.
  6. Universal Equivariant Multilayer Perceptrons proceedings.mlr.press. Checked 27 October 2026.
  7. Symmetry Breaking and Equivariant Neural Networks arxiv.org. Checked 27 October 2026.
  8. Efficient Dynamics Modeling in Interactive Environments with Koopman Theory arxiv.org. Checked 27 October 2026.
  9. Siamak Ravanbakhsh | Mila mila.quebec. Checked 27 October 2026.