Research map / Probabilistic and scientific modeling / Molecules and proteins
Simultaneous Identification of Sparse Structures and Communities in Heterogeneous Graphical Models
Dapeng Shi, Tiandong Wang, Zhiliang Ying
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 adaptive weights instead of non-adaptive l1 penalty (NonAPMLE) (ablation) | Proposed adaptive l1 estimator · Latent community GGM, n=1000 | TR_L: 0.010 -> 0.960 | Improved |
| use adaptive l1 estimator instead of LVGGM to recover zero blocks of L | Proposed adaptive l1 estimator · Latent community GGM, n=8000 | FP_L: 1.000 -> 0.014 | Improved |
| use adaptive l1 estimator instead of hard thresholding LVGGM for rank recovery | Proposed adaptive l1 estimator · Latent community GGM, n=8000 | TR_L: 0.000 -> 1.000 | Improved |
| use adaptive l1 estimator instead of LVGGM in a sparse graph with edges between and within communities | Proposed adaptive l1 estimator · Latent community GGM, n=1000 | FP_S: 0.003 -> 0.045 | Hurt |
| use adaptive l1 estimator instead of NonAPMLE for rank recovery in grouped latent GGM | Proposed adaptive l1 estimator · Grouped latent GGM, n=2000 | TR_L: 0.440 -> 0.910 | Improved |
| cluster on L-hat from proposed estimator rather than LVGGM when community eigenvalue is weak (a=1.5) | K-means on L-hat · Grouped latent GGM, n=1000 | Hamming error rate: 0.157 -> 0.121 | Improved |
| cluster on L-hat from proposed estimator at strong eigenvalue (a=3.5) | K-means on L-hat · Grouped latent GGM, n=1000 | Hamming error rate: 0.000 -> 0.000 | No clear effect |
| use Cor(abs(L-hat)) instead of L-hat as clustering input when eigenvalue is weak (a=1.5, proposed) | K-means, proposed estimator · Grouped latent GGM, n=1000 | Hamming error rate: 0.121 -> 0.537 | Hurt |
| use Cor(abs(L-hat)) instead of L-hat as clustering input at larger sample size (n=4000, proposed) | K-means, proposed estimator · Grouped latent GGM, larger within-community variation | Hamming error rate: 0.172 -> 0.047 | Improved |
| use adaptive l1 estimator instead of LVGGM for clustering as sample size grows (n=4000, Cor input) | K-means on Cor(abs(L-hat)) · Grouped latent GGM, larger within-community variation | Hamming error rate: 0.051 -> 0.047 | Improved |
Closest papers by meaning
- Community-Based Group Graphical Lasso (JMLR 2020 · similarity 0.91)
- Detecting Overlapping and Correlated Communities without Pure Nodes: Identifiability and Algorithm (ICML 2019 · similarity 0.88)
- Analysis of spectral clustering algorithms for community detection: the general bipartite setting (JMLR 2019 · similarity 0.88)
- Contextual Stochastic Block Models (NeurIPS 2018 · similarity 0.88)
- Learning to Discover Sparse Graphical Models (ICML 2017 · similarity 0.88)
- Learning Local Neighborhoods of Non-Gaussian Graphical Models (AAAI 2025 · similarity 0.86)
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.
