Research map / Graph learning / Graph algorithms and clustering
Improved Graph Laplacian via Geometric Self-Consistency
Dominique Joncas, Marina Meila, James McQueen
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 |
|---|---|---|---|
| Choose epsilon by geometric consistency (GC) instead of cross-validation | GC bandwidth selection · Digit1 | classification error (%): 3.32 -> 2.11 | Improved |
| Choose epsilon by GC instead of CV | GC · USPS | classification error (%): 5.18 -> 3.89 | Improved |
| Choose epsilon by GC instead of CV | GC · COIL | classification error (%): 7.02 -> 8.81 | Hurt |
| Choose epsilon by GC instead of CV | GC · BCI | classification error (%): 49.22 -> 48.67 | Improved |
| Choose epsilon by GC instead of CV | GC · g241c | classification error (%): 13.31 -> 12.77 | Improved |
| Choose epsilon by GC instead of reconstruction error (Rec) | GC · g241c | classification error (%): 23.93 -> 12.77 | Improved |
| Choose epsilon by GC instead of reconstruction error (Rec) | GC · g241d | classification error (%): 18.39 -> 8.76 | Improved |
| Include matrix inversion in algorithm (GC^-1 vs GC) (ablation) | GC · USPS | classification error (%): 3.89 -> 12.00 | Hurt |
| Include matrix inversion in algorithm (GC^-1 vs GC) (ablation) | GC · COIL | classification error (%): 8.81 -> 16.31 | Hurt |
| Increase projection dimension d' from 1 to 2 or 3 (ablation) | GC · g241c | estimated epsilon: 7.38 -> 9.83 | Mixed |
| Increase d' from 1 to 3 (ablation) | GC^-1 · Digit1 | estimated epsilon: 0.743 -> 0.305 | Mixed |
| Use 2-fold CV versus 5- or 20-fold CV (ablation) | CV baseline · SSL benchmarks | classification error | No clear effect |
Closest papers by meaning
- Manifold Random Features (NeurIPS 2026 · similarity 0.85)
- Toward a Unified Geometry Understanding : Riemannian Diffusion Framework for Graph Generation and Prediction (NeurIPS 2025 · similarity 0.85)
- Recovering Manifold Structure Using Ollivier Ricci Curvature (ICLR 2025 · similarity 0.84)
- The Noisy Laplacian: a Threshold Phenomenon for Non-Linear Dimension Reduction (ICML 2025 · similarity 0.84)
- Curvature-aware Graph Attention for PDEs on Manifolds (ICML 2025 · similarity 0.84)
- Euclidean Embedding of Data Using Local Distances (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.
