Research map / Graph learning / Graph algorithms and clustering
BASiS: Batch Aligned Spectral Embedding Space
Or Streicher, Ido Cohen, Guy Gilboa
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 BASiS batch-aligned supervised spectral learning instead of the strongest listed baseline | Spectral embedding models · MNIST, Fashion-MNIST, SVHN, CIFAR-10 | d_G: 0.204 ± 0.058 -> 0.107 ± 0.038 (MNIST); 0.375 ± 0.132 -> 0.284 ± 0.073 (Fashion-MNIST); 1.909 ± 0.238 -> 1.656 ± 0.170 ( | Improved |
| Use BASiS batch-aligned supervised spectral learning instead of Diffusion-Net | Spectral embedding models · MNIST, Fashion-MNIST, SVHN, CIFAR-10 | d_perp: 0.535 ± 0.365 -> 0.247 ± 0.076 (MNIST); 0.823 ± 0.664 -> 0.590 ± 0.144 (Fashion-MNIST); 1.532 ± 0.354 -> 0.488 ± 0.098 ( | Improved |
| Use BASiS batch-aligned supervised spectral learning instead of the strongest listed baseline | Spectral embedding models · MNIST, Fashion-MNIST, SVHN, CIFAR-10 | NMI: 0.944 ± 0.041 -> 0.961 ± 0.001 (MNIST); 0.761 ± 0.011 -> 0.798 ± 0.001 (Fashion-MNIST); 0.701 ± 0.009 -> 0.736 ± 0.001 ( | Improved |
| Use BASiS batch-aligned supervised spectral learning instead of the strongest listed baseline | Spectral embedding models · MNIST, Fashion-MNIST, SVHN, CIFAR-10 | ACC: 0.966 ± 0.007 -> 0.986 ± 0.001 (MNIST); 0.815 ± 0.029 -> 0.865 ± 0.003 (Fashion-MNIST); 0.813 ± 0.015 -> 0.880 ± 0.001 ( | Improved |
| Use BASiS batch-aligned supervised spectral learning instead of the strongest listed baseline | Spectral embedding models · MNIST, Fashion-MNIST, SVHN, CIFAR-10 | Accuracy (%): 97.026 ± 0.546 -> 98.522 ± 0.065 (MNIST); 86.207 ± 0.196 -> 87.202 ± 0.187 (Fashion-MNIST); 87.469 ± 0.130 -> 88.021 ± 0 | Improved |
| Align batch embeddings using shared anchors and an affine transformation | BASiS · Batch-based spectral embedding learning | Embedding-coordinate consistency | Improved |
| Align embeddings to the previous anchor space after feature changes | BASiS with iterative feature change · MNIST | Clustering performance | Improved |
Closest papers by meaning
- Laplacian Canonization: A Minimalist Approach to Sign and Basis Invariant Spectral Embedding (NeurIPS 2023 · similarity 0.88)
- On the Generalization of Equivariant Graph Neural Networks (ICML 2024 · similarity 0.85)
- Spectral Graph Sparsification Preserves Representation Geometry in Graph Neural Networks (NeurIPS 2026 · similarity 0.85)
- Towards an Explainable Comparison and Alignment of Feature Embeddings (ICML 2025 · similarity 0.85)
- Position: Spectral GNNs Rely Less on Graph Fourier Basis than Conceived (ICML 2025 · similarity 0.84)
- Spectral Graph Neural Networks are Incomplete on Graphs with a Simple Spectrum (NeurIPS 2025 · similarity 0.84)
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.
