Research map / Learning theory and algorithms / Sample complexity
Neural Injective Functions for Multisets, Measures and Graphs via a Finite Witness Theorem
Tal Amir, Steven J. Gortler, Ilai Avni, Ravina Ravina, Nadav Dym
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| analytic activations instead of piecewise-linear activations in MPNN | MPNN with graph convolution layers · TUDataset 600 graphs, WL separation | graphs where coloring differs from 1-WL: about 1% | Improved |
| analytic activations with hidden dimension 1 | MPNN · TUDataset 600 graphs | equivalence to 1-WL | Improved |
| piecewise-linear vs analytic activations at low embedding dimension | moment-based multiset functions · random multisets, d=3, n=1000 | c/C ratio | Improved |
| increase embedding dimension m | moment-based multiset functions · random multisets, d=3, n=1000 | c/C ratio | Improved |
| high embedding dimension, analytic vs PwL activation | moment-based multiset functions · bi-Lipschitzness | c/C ratio | No clear effect |
| analytic activations, least-separated node features | MPNN · TUDataset 600 graphs | feature distance of non-equivalent nodes: 0.1 -> 10^-7 | Mixed |
Closest papers by meaning
- Globally injective and bijective neural operators (NeurIPS 2023 · similarity 0.85)
- A Function Space Analysis of Finite Neural Networks With Insights From Sampling Theory (TPAMI 2022 · similarity 0.84)
- Globally Injective ReLU Networks (JMLR 2022 · similarity 0.84)
- A Functional Perspective on Learning Symmetric Functions with Neural Networks (ICML 2021 · similarity 0.84)
- On the Lipschitz Continuity of Set Aggregation Functions and Neural Networks for Sets (ICLR 2026 · similarity 0.84)
- On the Hölder Stability of Multiset and Graph Neural Networks (ICLR 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
