Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
An Asymptotically Optimal Coordinate Descent Algorithm for Learning Bayesian Networks from Gaussian Models
Tong Xu, Simge Küçükyavuz, Ali Shojaie, Armeen Taeb
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 CD-l0 instead of CCDr-MCP (l0 vs minimax concave penalty) | coordinate descent · Pathfinder(109) | d_cpdag: 206.6 -> 95.0 | Improved |
| use CD-l0 instead of GES | structure learning · Diabetes(413) | d_cpdag: 458.7 -> 158.4 | Improved |
| use CD-l0 instead of MICODAG on moderate graphs within time limit | structure learning · Factors(27) | d_cpdag: 60.4 -> 25.2 | Improved |
| use CD-l0 instead of TD on Hepar | structure learning · Hepar(70) moderate heteroscedasticity | d_cpdag: 54.2 -> 38.5 | Improved |
| use CD-l0 instead of NOTEARS (likelihood vs least squares) | structure learning · Factors(27) | d_cpdag: 41.7 -> 25.2 | Improved |
| use CD-l0 instead of TD on Hfinder | structure learning · Hfinder(56) moderate heteroscedasticity | d_cpdag: 38.3 -> 45.1 | Hurt |
| increase noise variance heterogeneity from {0.6,1,1.2} to {0.2,1,5} | CD-l0 · Hepar(70) | d_cpdag: 38.5 -> 91.5 | Hurt |
| use CD-l0 instead of GES under severe heteroscedasticity | structure learning · Hepar(70) severe heteroscedasticity | d_cpdag: 68.5 -> 91.5 | Hurt |
| use CD-l0 instead of GES under severe heteroscedasticity | structure learning · Factors(27) severe heteroscedasticity | d_cpdag: 74.1 -> 40.0 | Improved |
| use CD-l0 instead of exact MICODAG for solution time | structure learning · synthetic DAGs m=10 | solve time: near-optimal solution at about 0.1% of time | Improved |
| use CD-l0 instead of GES for reaching the optimal objective at moderately large n | structure learning · synthetic DAGs m=10 | normalized objective gap | Improved |
| use CD-l0 instead of GES on causal chambers data | structure learning · Causal chambers (m=20) | d_cpdag | No clear effect |
Closest papers by meaning
- Learning Identifiable Gaussian Bayesian Networks in Polynomial Time and Sample Complexity (NeurIPS 2017 · similarity 0.89)
- DAGs with No Fears: A Closer Look at Continuous Optimization for Learning Bayesian Networks (NeurIPS 2020 · similarity 0.88)
- Consistent Second-Order Conic Integer Programming for Learning Bayesian Networks (JMLR 2023 · similarity 0.88)
- Structure Learning in Bayesian Networks of a Moderate Size by Efficient Sampling (JMLR 2016 · similarity 0.88)
- A Study of Bayesian Neural Network Surrogates for Bayesian Optimization (ICLR 2024 · similarity 0.87)
- Bayesian Optimization via Continual Variational Last Layer Training (ICLR 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.
