Research map / Graph learning / Graph algorithms and clustering
Learning the piece-wise constant graph structure of a varying Ising model
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 a group-fused ℓ2 penalty across consecutive parameter vectors instead of Tesla's ℓ1 fused penalty | TVI-FL versus Tesla · Synthetic piece-wise constant Ising graphs | h-score: 0.046 ± (0.024) vs 0.106 ± (0.087); 0.221 ± (0.035) vs 0.184 ± (0.051) | Mixed |
| Use a group-fused ℓ2 penalty across consecutive parameter vectors instead of Tesla's ℓ1 fused penalty | TVI-FL versus Tesla · Synthetic piece-wise constant Ising graphs | F1-score: 0.694 ± (0.103) vs 0.649 ± (0.190); 0.876 ± (0.030) vs 0.841 ± (0.041) | Helped |
| Select hyperparameters by AIC rather than maximizing AUC | TVI-FL · Synthetic Ising graphs and change-points | h-score: 0.046 ± (0.024) vs 0.221 ± (0.035) | Mixed |
| Select hyperparameters by maximizing AUC rather than AIC | TVI-FL · Synthetic Ising graphs and change-points | F1-score: 0.694 ± (0.103) vs 0.876 ± (0.030) | Helped |
| Use a group-fused penalty that couples all dimensions of adjacent parameter vectors | TVI-FL · Piece-wise constant Ising structure recovery | change-point recovery | Helped |
| Increase the number of observations per timestamp | TVI-FL · Synthetic Ising, d=2, AIC-selected | F1-score: 0.694 ± (0.103) -> 0.816 ± (0.073) -> 0.833 ± (0.095) | Helped |
Closest papers by meaning
- Generalized Direct Change Estimation in Ising Model Structure (ICML 2016 · similarity 0.92)
- Learning Changes in Graphon Attachment Network Models (ICML 2025 · similarity 0.83)
- When Edge Independence Fails: Joint Graph Diffusion with Latent Sociability Priors (NeurIPS 2026 · similarity 0.83)
- Pard: Permutation-Invariant Autoregressive Diffusion for Graph Generation (NeurIPS 2024 · similarity 0.83)
- Beyond Spatio-Temporal Representations: Evolving Fourier Transform for Temporal Graphs (ICLR 2024 · similarity 0.83)
- Fixed-Confidence Multiple Change Point Identification under Bandit Feedback (ICML 2025 · similarity 0.83)
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.