Pim de Haan

Machine Learning Researcher

I currently work as a machine learning researcher at CuspAI on the discovery of new materials for carbon capture, using simulation and ML.

Previously, I was a senior researcher at Qualcomm AI Research and a PhD student in machine learning under supervision of Max Welling in the AMLab at the University of Amsterdam. In the PhD, I worked on deep learning in the presence of symmetries, generalizing symmetry groups to groupoids and beyond, with applications to geometric data, such as graphs, point clouds and meshes. Additionally, I’ve studied various applications to physical domains, including fluids, quantum lattice systems, and particle physics. Also, I’ve dipped my toes in causality, but decided that it wasn’t for me.

Before my PhD, I obtained masters degrees in theoretical physics in Cambridge and in AI in Amsterdam. I wrote my masters thesis on Causal Imitation Learning as visiting scholar at the Robotic AI and Learning Lab at UC Berkeley supervised by Sergey Levine.

Have a look at my Resume.

• Tycho van der Ouderaa, Mark van der Wilk, Pim de Haan:
  Noether’s Razor: Learning Conserved Quantities
  NeurIPS 2024 [to appear]
• Jonas Spinner*, Victor Bresó*, Pim de Haan, Tilman Plehn, Jesse Thaler, Johann Brehmer:
  Lorentz-Equivariant Geometric Algebra Transformers for High-Energy Physics
  NeurIPS 2024 [arxiv]
• Pim de Haan, Taco Cohen, Johann Brehmer:
  Euclidean, Projective, Conformal: Choosing a Geometric Algebra for Equivariant Transformers
  AISTATS 2024 [arxiv]
• Risto Vuorio*, Pim de Haan*, Johann Brehmer, Hanno Ackermann, Daniel Dijkman, Taco Cohen:
  Deconfounding Imitation Learning with Variational Inference
  TMLR, 2024 [openreview]
• Julian Suk, Pim de Haan, Phillip Lippe, Christoph Brune, Jelmer M. Wolterink
  Mesh neural networks for SE(3)-equivariant hemodynamics estimation on the artery wall
  Computers in Biology and Medicine, 2024 [link]
• Johann Brehmer*, Pim de Haan*, Sönke Beherends, Taco Cohen:
  Geometric Algebra Transformer
  NeurIPS 2023 [arxiv]
• Mathis Gerdes*, Pim de Haan*, Corrado Rainone, Roberto Bondesan, Miranda CN Cheng:
  Learning Lattice Quantum Field Theories with Equivariant Continuous Flows
  SciPost Physics, 2023 [arxiv]
• Johann Brehmer*, Pim de Haan*, Phillip Lippe, Taco Cohen:
  Weakly supervised causal representation learning
  NeurIPS 2022 [arxiv]
• Pim de Haan*, Maurice Weiler*, Taco Cohen, Max Welling:
  Gauge Equivariant Mesh CNNs: Anisotropic convolutions on geometric graphs
  ICLR 2021 (spotlight) [openreview]
• Pim de Haan, Taco Cohen, Max Welling:
  Natural Graph Networks
  NeurIPS 2020 [arxiv]
• Pim de Haan, Dinesh Jayaraman, Sergey Levine:
  Causal Confusion in Imitation Learning
  NeurIPS 2019 (oral, 0.5% of submissions) [arxiv]
• Luca Falorsi, Pim de Haan, Tim R. Davidson, Patrick Forré:
  Reparameterizing Distributions on Lie Groups
  AISTATS 2019 (oral) [arxiv]
• Luca Falorsi*, Pim de Haan*, Tim R. Davidson*, Nicola De Cao, Maurice Weiler, Patrick Forré, Taco S. Cohen:
  Explorations in Homeomorphic Variational Auto-Encoding
  ICML 2018 workshop on Theoretical Foundations and Applications of Deep Generative Models [arxiv]