| CARVIEW |
Haggai Maron
I am an Assistant Professor and the Robert J. Shillman Fellow at the Faculty of Electrical and Computer Engineering at the Technion and a senior research scientist at NVIDIA Research at NVIDIA's lab in Tel Aviv. My primary research interest is in machine learning, with a focus on deep learning for structured data. Specifically, I study how to apply deep learning techniques to sets, graphs, point clouds, surfaces, weight spaces and other mathematical objects that have an inherent symmetry structure. My goal is twofold: first, to understand and design deep learning architectures from a theoretical perspective, for example, by analyzing their expressive power; and second, to demonstrate their practical effectiveness on real-world problems involving structured data. I completed my Ph.D. in 2019 at the Weizmann Institute of Science under the supervision of Prof. Yaron Lipman. You can get an idea of what I am working on by taking a look at these talks I gave:
- Geometric Deep Learning for Neural Artifacts: Symmetry-Aware Learning across Trained Model Weights, Internal Representations, and Gradients.
- The expressive power of GNNs - short tutorial at the Simons Institute.
- Equivaraint architectures for learning in deep weight spaces
- Subgraph-based networks for expressive, efficient, and domain-independent graph learning
- Leveraging Permutation Group Symmetries for Equivariant Neural Networks
Email: haggaimaron (at) technion.ac.il, Google scholar page
Group
Current Group Members: Fabrizio Frasca (Postdoc, 2024-today), Yam Eitan (PhD, 2024 - today), Guy Bar-Shalom (PhD, 2023 - today, joint supervision with Ran El-Yaniv), Yoav Gelberg (PhD, 2024-today, joint supervision with Michael Bronstein), Ran Elbaz (Msc, 2023-today), Yaniv Galron (MSc, 2023-today, joint supervision with Eran Treister)Past Group Members: Yuval Aidan (MSc, 2023-2025, joint supervision with Ayellet Tal).
Edan Kinderman (MSc, 2023-2025, joint supervision with Daniel Soudry), Ofir Haim (MSc, 2023-2025, joint supervision with Shie Mannor)
Close student / postdoc collaborators: Theo (Moe) Putterman (UC Berkeley, 2023-today), Beatrice Bevilacqua (PhD Candidate, Purdue University, 2021-2025), Derek Lim (PhD Candidate, MIT CSAIL, 2021-2025), Moshe Eliasof (University of Cambridge, BGU, 2022-today), Aviv Navon (BIU, Aiola, 2021-today), Aviv Shamsian (BIU, Aiola, 2022-today).
News
- I gave keynotes at the Learning on Graphs Conference (LoG 2025) and the Topology Algebra and Geometry in Data Science Conference (TAG-DS 2025) on Geometric Deep Learning for Neural Artifacts: Symmetry-Aware Learning across Trained Model Weights, Internal Representations, and Gradients.
- See my recent short tutorial on the expressive power of GNNs at the Graph Learning Meets Theoretical Computer Science workshop we organized at the Simons Institute.
- Proud to be the recipient of the Alon scholarship for the Integration of Outstanding Faculty.
- See my recent talk on Equivaraint architectures for learning in deep weight spaces .
- A new blog post on Equivaraint architectures for learning in deep weight spaces .
- I will serve as an area chair at NeurIPS 2023.
- See our recent tutorial on Exploring the practical and theoretical landscape of expressive graph neural networks given at the Learning on Graphs Conference . With With Fabrizio Frasca and Beatrice Bevilacqua.
- A new blog post on Subggraph GNNs on Towards Data Science.
- Our ICML 2020 paper On Learning Sets of Symmetric Elements received the Outstanding paper award (best paper award). See this interview for more details.
Teaching
- 2025/Spring (Technion): Introduction to Machine Learning
- 2025/Spring (Technion): Deep Learning and groups
- 2024/winter (Technion): Topics in learning on graphs
- 2024/Spring (Technion): Deep Learning and groups
- 2023/winter (Technion): Topics in learning on graphs
- 2019/spring (WIS): Geometric and Algebraic Methods in Deep Learning
- 2018/winter (WIS): Geometry and Deep Learning
Publications
Beyond Next Token Probabilities: Learnable, Fast Detection of Hallucinations and Data Contamination on LLM Output Distributions
Guy Bar-Shalom*, Fabrizio Frasca*, Derek Lim, Yoav Gelberg, Yftah Ziser, Ran El-Yaniv, Gal Chechik, Haggai Maron
The 40th Annual AAAI Conference on Artificial Intelligence (AAAI 2026)
A shorter version was also presented as a spotlight presentation at the ICLR 2025 workshop 'Quantify Uncertainty and Hallucination in Foundation Models'
Abstract
Paper
Code
A Flexible, Equivariant Framework for Subgraph GNNs via Graph Products and Graph Coarsening
Guy Bar-Shalom, Yam Eitan, Fabrizio Frasca, Haggai Maron
38th Annual Conference on Neural Information Processing Systems (NeurIPS 2024)
The Empirical Impact of Neural Parameter Symmetries, or Lack Thereof
Derek Lim*, Theo (Moe) Putterman*, Robin Walters, Haggai Maron, Stefanie Jegelka
38th Annual Conference on Neural Information Processing Systems (NeurIPS 2024)
(A shorter version also appeared in HiLD 2024: 2nd Workshop on High-dimensional Learning Dynamics @ ICML 2024 winning the best paper award)
Improved Generalization of Weight Space Networks via Augmentationss
Aviv Shamsian, Aviv Navon, David W. Zhang, Yan Zhang, Ethan Fetaya, Gal Chechik, Haggai Maron
International Conference on Machine Learning (ICML) 2024
(A preliminary version appeared in Symmetry and Geometry in Neural Representations Workshop, 37th Annual Conference on Neural Information Processing Systems (NeurIPS 2023) as an Oral Presentation).
Equivariant Subgraph Aggregation Networks
Beatrice Bevilacqua*, Fabrizio Frasca*, Derek Lim*, Balasubramaniam Srinivasan, Chen Cai, Gopinath Balamurugan, Michael M. Bronstein, Haggai Maron (*equal contribution)
International Conference on Learning Representations (ICLR) 2022
Spotlight presentation (5% acceptance rate)
Abstract Paper GitHub Blog post Talk
StyleGAN-NADA: CLIP-Guided Domain Adaptation of Image Generators
Rinon Gal, Or Patashnik , Haggai Maron , Gal Chechik , Daniel Cohen-Or
ACM SIGGRAPH 2022
