Research map / Learning theory and algorithms / Sample complexity
Variance estimation in graphs with the fused lasso
Oscar Hernan Madrid Padilla
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 proposed Het. estimator instead of U-LP kernel estimator (homoscedastic truth, scenario 3, v0=1.0) | variance estimation on 2D grid · Homoscedastic grid, Scenario 3, n=400^2 | 100x MSE: 4.81 -> 0.29 | Improved |
| use Hom. estimator instead of U-LP | variance estimation on 2D grid · Homoscedastic grid, Scenario 1, n=400^2, v0=0.5 | 100x MSE: 0.61 -> 0.20 | Improved |
| use Het. estimator instead of Hom. when true variance is constant and larger | variance estimation on 2D grid · Homoscedastic grid, Scenario 3, n=400^2, v0=1.0 | 100x MSE: 1.03 -> 0.29 | Improved |
| use Het. estimator instead of Hom. at small variance (v0=0.5) | variance estimation on 2D grid · Homoscedastic grid, Scenario 1, n=400^2, v0=0.5 | 100x MSE: 0.20 -> 0.70 | Hurt |
| use Het. estimator instead of local polynomial regression (heteroscedastic grid) | variance estimation on 2D grid · Heteroscedastic grid, Scenario 4, n=400^2 | 10x MSE: 3.15 -> 0.18 | Improved |
| use Het. estimator instead of Laplacian smoothing (heteroscedastic grid) | variance estimation on 2D grid · Heteroscedastic grid, Scenario 6, n=400^2 | 10x MSE: 7.41 -> 0.29 | Improved |
| use Het. estimator instead of local polynomial regression on K-NN graph | variance estimation with K-NN graph · Heteroscedastic K-NN graph, Scenario 7, d=2, n=20000 | 10x MSE: 1.25 -> 0.27 | Improved |
| increase dimension d from 2 to 3 for the Het. estimator on K-NN graph (ablation) | Het. estimator · Heteroscedastic K-NN graph, Scenario 7, n=20000 | 10x MSE: 0.27 -> 0.89 | Hurt |
| increase grid size n from 100^2 to 400^2 for Het. estimator (ablation) | Het. estimator · Heteroscedastic grid, Scenario 4 | 10x MSE: 1.34 -> 0.18 | Improved |
| use graph fused lasso instead of Laplacian smoothing for mean estimation with Laplace errors | mean estimation, Scenario 4 · Laplace errors, grid n=400^2 | 10x MSE: 1.11 -> 0.13 | Improved |
| choose lambda by BIC with homoscedastic variance estimate instead of oracle best lambda (ablation) | graph fused lasso mean estimation · Scenarios 1-3, sigma in {0.25,0.5,0.75,1} | selected lambda | No clear effect |
Closest papers by meaning
- The DFS Fused Lasso: Linear-Time Denoising over General Graphs (JMLR 2018 · similarity 0.92)
- Generalized Sparse Learning of Linear Models Over the Complete Subgraph Feature Set (TPAMI 2016 · similarity 0.87)
- Equivalence of Graphical Lasso and Thresholding for Sparse Graphs (JMLR 2016 · similarity 0.87)
- Graphical Lasso and Thresholding: Equivalence and Closed-form Solutions (JMLR 2019 · similarity 0.87)
- Bayesian Joint Estimation of Multiple Graphical Models (NeurIPS 2019 · similarity 0.87)
- Fair GLASSO: Estimating Fair Graphical Models with Unbiased Statistical Behavior (NeurIPS 2024 · similarity 0.85)
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.
