Research map / Graph learning / Graph algorithms and clustering
LogSpecT: Feasible Graph Learning Model from Stationary Signals with Recovery Guarantees
Shangyuan Liu, Linglingzhi Zhu, Anthony Man-Cho So
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 |
|---|---|---|---|
| replace SpecT with LogSpecT | graph learning from stationary signals · BA graphs, 20-60 nodes | F-measure | Improved |
| replace SpecT with LogSpecT | graph learning · ER graphs, 20-60 nodes | F-measure | No clear effect |
| vary graph filter (lowpass-EXP, highpass-EXP, QUA) | LogSpecT and SpecT · synthetic ER/BA | F-measure | No clear effect |
| increase number of nodes | LogSpecT · BA graphs | F-measure | Improved |
| increase number of nodes | SpecT · BA graphs | F-measure | No clear effect |
| increase sample size n from 10 to 10^6 | rLogSpecT · BA graphs, three filters | F-measure | Improved |
| increase sample size | rSpecT feasibility · Protein | infeasibility frequency: 0.980 -> 0.999 -> 0.994 | No clear effect |
| increase sample size | rSpecT feasibility · Reddit | infeasibility frequency: 0.984 -> 1 -> 1 | No clear effect |
| increase sample size from 10 to 1000 | rSpecT · Protein | mean of delta_min: 38.514 -> 26.012 -> 14.813 | Improved |
| increase sample size from 10 to 1000 | rSpecT · Reddit | mean of delta_min: 1094.788 -> 830.642 -> 531.006 | Improved |
| replace rSpecT with rLogSpecT | graph learning · Protein and Reddit, fixed sample count | recovery of ground truth graph | Improved |
| use stationarity-based models instead of correlation or Kalofolias | rLogSpecT/rSpecT · Protein and Reddit | graph recovery | Improved |
Closest papers by meaning
- WaveNet: Tackling Non-stationary Graph Signals via Graph Spectral Wavelets (AAAI 2024 · similarity 0.85)
- A Poincaré Inequality and Consistency Results for Signal Sampling on Large Graphs (ICLR 2024 · similarity 0.85)
- Out-of-Distribution Generalization on Graphs via Progressive Inference (AAAI 2025 · similarity 0.84)
- Graph Neural Networks Are More Than Filters: Revisiting and Benchmarking from A Spectral Perspective (ICLR 2025 · similarity 0.84)
- SLOG: An Inductive Spectral Graph Neural Network Beyond Polynomial Filter (ICML 2024 · similarity 0.84)
- Taxonomy of reduction matrices for Graph Coarsening (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.
