Research map / Architectures and efficiency / CNN design
FS-KAN: Permutation Equivariant Kolmogorov-Arnold Networks via Function Sharing
Ran Elbaz, Guy Bar-Shalom, Yam Eitan, Fabrizio Frasca, Haggai Maron
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 |
|---|---|---|---|
| use function-sharing KAN (FS-KAN) instead of DeepSets parameter-sharing MLP | DeepSets · ModelNet40 continual learning, train size 200 | average accuracy: 0.380 -> 0.420 | Improved |
| use FS-KAN instead of DeepSets | DeepSets · ModelNet40 point clouds, train size 200 (acc. A1) | accuracy: 0.514 -> 0.531 | Improved |
| use FS-KAN instead of DeepSets | DeepSets · ModelNet40, train size 1000 | average accuracy: 0.555 vs 0.553 | No clear effect |
| use FS-KAN instead of DeepSets | DeepSets · Forgetting at train size 200 | forgetting: 0.059 -> 0.034 | Improved |
| use FS-KAN instead of DeepSets | DeepSets · Forgetting at train size 1000 | forgetting: 0.027 -> 0.040 | Mixed |
| use efficient FS-KA layer instead of full FS-KA layer (ablation) | FS-KAN · ModelNet40 continual learning avg accuracy, train size 200 | average accuracy: 0.420 -> 0.395 | Hurt |
| use efficient FS-KA layer instead of full (ablation) | FS-KAN · ModelNet40 continual learning avg accuracy, train size 1000 | average accuracy: 0.553 -> 0.525 | Hurt |
| increase training data (high-data regime) | FS-KAN vs parameter-sharing MLP · Rating prediction RMSE | RMSE | No clear effect |
| share functions across symmetric edges | S_n-FS-KAN · synthetic invariant function (interpretability) | interpretability | Improved |
Closest papers by meaning
- Incorporating Arbitrary Matrix Group Equivariance into KANs (ICML 2025 · similarity 0.92)
- KAN: Kolmogorov–Arnold Networks (ICLR 2025 · similarity 0.88)
- On the expressiveness and spectral bias of KANs (ICLR 2025 · similarity 0.88)
- Permutation Equivariant Neural Networks for Symmetric Tensors (ICML 2025 · similarity 0.88)
- REVISITING MULTI-PERMUTATION EQUIVARIANCE THROUGH THE LENS OF IRREDUCIBLE REPRESENTATIONS (ICLR 2025 · similarity 0.87)
- Equivariant Polynomial Functional Networks (ICML 2025 · similarity 0.87)
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.
