Snir Hordan

PhD Candidate at the Technion - Israel Institute of Technology.

Domain: Geometric Deep Learning.

Google Scholar | Homepage

About Me

I am a PhD candidate at the Technion, supervised by Dr. Nadav Dym, where I research machine learning models that process geometric data, such as graphs and point clouds. This is performed via two interlocking paradigms: identifying the shortcomings of geometric models and, consequently, developing geometrically complete models, favorably with low-dimensional features. This yields more powerful models with moderate computational cost.

Publications

• Spectral Graph Neural Networks are Incomplete on Graphs with a Simple Spectrum

  S Hordan, M Bechler-Speicher, G Lifshitz, N Dym

  To be presented at the Conference on Neural Information Processing Systems (NeurIPS), 2025 (Spotlight 3% acceptance rate)

• Weisfeiler Leman for Euclidean Equivariant Machine Learning

  S Hordan, T Amir, N Dym

  Presented at the International Conference on Machine Learning (ICML), 2024

  

• Complete Neural Networks for Euclidean Graphs

  S Hordan, T Amir, SJ Gortler, N Dym

  Presented at the AAAI Conference on Artificial Intelligence (AAAI), 2024

Connect with Me