Research map / Label-efficient and robust learning / Recognition and long-tail
Unsupervised Co-Learning on $G$-Manifolds Across Irreducible Representations
Yifeng Fan, Tingran Gao, Zhizhen Jane Zhao
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 multi-irrep power-spectrum affinity instead of scalar edge weights | Spectral clustering with SO(2) · K=2, p=0.16 | Rand index: 0.569 ± 0.069 -> 0.670 ± 0.065 | Improved |
| Use multi-irrep optimal-alignment affinity instead of VDM | Spectral clustering with SO(2) · K=2, p=0.16 | Rand index: 0.526 ± 0.036 -> 0.687 ± 0.011 | Improved |
| Use multi-irrep bispectrum affinity instead of scalar edge weights | Spectral clustering with SO(2) · K=10, p=0.20 | Rand index: 0.948 ± 0.015 -> 0.997 ± 0.003 | Improved |
| Use multi-irrep affinity rather than single-representation VDM | Spectral clustering with SO(3) · K=2, p=0.16 | Rand index: 0.600 ± 0.048 -> 0.921 ± 0.038 | Improved |
| Use multi-irrep bispectrum affinity rather than VDM | Spectral clustering with SO(3) · K=10, p=0.20 | Rand index: 0.919 ± 0.013 -> 0.943 ± 0.009 | Improved |
| Use multi-irrep optimal-alignment affinity rather than VDM | S² nearest-neighbor identification · p=0.10 | percentage of identified neighbor pairs with inner product > 0.95: 27.56 -> 87.72 | Improved |
| Increase cutoff m_k from 2 to 20 (ablation) | S² nearest-neighbor identification with Opt affinity · p=0.10 | percentage of identified neighbor pairs with inner product > 0.95: 15.03 -> 87.72 | Improved |
| Increase cutoff m_k from 20 to 100 (ablation) | S² nearest-neighbor identification with Opt affinity · p=0.10 | percentage of identified neighbor pairs with inner product > 0.95: 87.72 -> 82.66 | Hurt |
| Increase maximum frequency k_max from 2 to 50 (ablation) | S² nearest-neighbor identification with Opt affinity · p=0.10 | percentage of identified neighbor pairs with inner product > 0.95: 43.42 -> 92.07 | Improved |
| Increase maximum frequency k_max from 2 to 100 (ablation) | Spectral clustering with SO(2), K=10 · p=0.16 | Rand index: 0.920 ± 0.019 -> 0.985± 0.003 | Improved |
| Increase maximum frequency k_max from 2 to 100 (ablation) | Spectral clustering with SO(2), K=10 · p=0.25 | Rand index: 0.997 ± 0.001 -> 1 ± 0 | No clear effect |
| Increase cutoff m_k beyond the cluster count (ablation) | Spectral clustering with SO(2), K=10 · p=0.16, 0.20, 0.25 | Rand index | Mixed |
Closest papers by meaning
- Co-manifold learning with missing data (ICML 2019 · similarity 0.88)
- Recovering Manifold Structure Using Ollivier Ricci Curvature (ICLR 2025 · similarity 0.84)
- Finsler Multi-Dimensional Scaling: Manifold Learning for Asymmetric Dimensionality Reduction and Embedding (CVPR 2025 · similarity 0.84)
- Geometric Analysis of Nonlinear Manifold Clustering (NeurIPS 2024 · similarity 0.84)
- Partition First, Embed Later: Laplacian-Based Feature Partitioning for Refined Embedding and Visualization of High-Dimensional Data (ICML 2025 · similarity 0.84)
- Building Transformation Layers for Riemannian Neural Networks (NeurIPS 2026 · 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.
