Research map / Probabilistic and scientific modeling / Variational and generative sampling
Simultaneous Change Point Inference and Structure Recovery for High Dimensional Gaussian Graphical Models
Bin Liu, Xinsheng Zhang, Yufeng Liu
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 node-wise threshold-variable regression instead of fused graphical lasso for change point location | change point estimation · Model 2 (p=100, tau*=0.5, alpha=10) | Rmse: 0.123 -> 0.008 | Improved |
| use node-wise method instead of FGL as dimension grows | change point estimation · Model 2 (p=200, tau*=0.3) | Rmse: 0.100 -> 0.030 | Improved |
| use node-wise thresholding instead of FGL for E(1) recovery | structure recovery · Model 1 (p=100, tau*=0.5, alpha=20) | precision: 0.19 -> 0.79 | Improved |
| use node-wise method instead of FGL at larger dimension | structure recovery · Model 1 (p=300, tau*=0.5, alpha=20) | precision: 0.08 -> 0.88 | Improved |
| use node-wise method instead of FGL at p=100, alpha=25 | change point estimation · Model 1 (p=100, tau*=0.5, alpha=25) | Rmse: 0.001 -> 0.007 | No clear effect |
| increase dimension p from 100 to 300 | FGL · Model 1 (tau*=0.5, alpha=20) | Rmse of change point: 0.001 -> 0.074 | Hurt |
| increase dimension p from 100 to 300 | Node-wise · Model 1 (tau*=0.5, alpha=20) | E(1) F-score: 0.88 -> 0.80 | Hurt |
| decrease similarity between graphs (alpha 25 to 10) | Node-wise · Model 1 (p=100, tau*=0.5) | Rmse: 0.007 -> 0.047 | Hurt |
| move change point toward boundary (tau* 0.5 to 0.3) | Node-wise · Model 2 (p=200, alpha=20) | Rmse: 0.007 -> 0.030 | Hurt |
| move change point toward boundary (tau* 0.5 to 0.3) with p=300 | change point detection classifier · Model 2 (alpha=100) | type II error: 0 -> 0.04 | Hurt |
| use known change point with separate neighborhood selection (Oracle 1) instead of Algorithm 2 thresholding (Oracle 2) (ablation) | structure recovery · Model 1 (p=100, tau*=0.5, alpha=10) | E(1) precision: 0.82 -> 0.98 | Improved |
| estimate change point jointly (Node-wise) instead of assuming known tau* (Oracle 2) | structure recovery · Model 1 (p=100, tau*=0.5, alpha=10) | E(1) precision: 0.82 -> 0.82 | No clear effect |
Closest papers by meaning
- Sequential change-point detection in high-dimensional Gaussian graphical models (JMLR 2020 · similarity 0.90)
- Joint Structural Estimation of Multiple Graphical Models (JMLR 2016 · similarity 0.88)
- Change-Point Computation for Large Graphical Models: A Scalable Algorithm for Gaussian Graphical Models with Change-Points (JMLR 2018 · similarity 0.88)
- An Efficient Two Step Algorithm for High Dimensional Change Point Regression Models Without Grid Search (JMLR 2019 · similarity 0.87)
- Fast Bayesian Inference of Sparse Networks with Automatic Sparsity Determination (JMLR 2020 · similarity 0.87)
- Learning Local Neighborhoods of Non-Gaussian Graphical Models (AAAI 2025 · similarity 0.87)
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.
