Research map / Learning theory and algorithms / Combinatorial and search optimization
Consistent Second-Order Conic Integer Programming for Learning Bayesian Networks
Simge Kucukyavuz, Ali Shojaie, Hasan Manzour, Linchuan Wei, Hao-Hsiang Wu
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 linear-ordering (LN) instead of topological-ordering (TO) cycle constraints (ablation) | MISOCP · Complete instances, m=20 | RGAP: 0.272 -> 0.195 | Improved |
| use LN instead of TO cycle constraints (ablation) | MISOCP · Moral instances, m=20 | RGAP: 0.021 -> 0.006 | Improved |
| use LN instead of TO cycle constraints (ablation) | MISOCP · m=30 and m=40 | feasible solution found | Improved |
| use LN instead of cutting-plane (CP) cycle constraints (ablation) | MISOCP · n=100 synthetic | RGAP and time | Improved |
| use MISOCP+LN instead of SIMILP+LN | MISOCP · n=100 synthetic | RGAP, upper bound, time | Improved |
| use second-order conic MISOCP+LN instead of big-M MIQP+LN | MIQP+LN · Complete instances, m=40, lambda_n=18.4 | RGAP: 0.465 -> 0.374 | Improved |
| use MISOCP+LN instead of MIQP+LN | MIQP+LN · Moral instances, m=40, lambda_n=4.6 | RGAP: .037 -> .042 | Hurt |
| use MISOCP+LN instead of MIQP+LN | MIQP+LN · Moral instances, m=40, lambda_n=18.4 | instances solved to optimality: eight out of ten vs two out of ten | Improved |
| increase lambda_n from log(n) to 4 log(n) | MISOCP+LN · Complete instances, m=20 | RGAP: .195 -> .113 | Improved |
| use MISOCP+LN instead of MIQP+LN | MIQP+LN · Complete instances | branch-and-bound nodes and relaxation value | Improved |
Closest papers by meaning
- Learning Identifiable Gaussian Bayesian Networks in Polynomial Time and Sample Complexity (NeurIPS 2017 · similarity 0.90)
- DAGs with No Fears: A Closer Look at Continuous Optimization for Learning Bayesian Networks (NeurIPS 2020 · similarity 0.90)
- Efficient Bayesian Network Structure Learning via Parameterized Local Search on Topological Orderings (AAAI 2021 · similarity 0.88)
- Causal Discovery via Bayesian Optimization (ICLR 2025 · similarity 0.88)
- Inexact Column Generation for Bayesian Network Structure Learning via Difference-of-Submodular Optimization (NeurIPS 2025 · similarity 0.86)
- ProDAG: Projected Variational Inference for Directed Acyclic Graphs (NeurIPS 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.
