Research map / Learning theory and algorithms / Combinatorial and search optimization
Branch & Learn for Recursively and Iteratively Solvable Problems in Predict+Optimize
Xinyi HU, Jasper C.H. Lee, Jimmy H.M. Lee, Allen Z. Zhong
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 |
|---|---|---|---|
| B&L vs linear regression | MCFP · USANet real-life, n=100 | mean regret: 177.55±154.53 -> 141.37±128.52 | Improved |
| B&L vs linear regression | MCFP · GEANT artificial, n=100 | mean regret: 765.13±345.07 -> 733.78±294.23 | Improved |
| B&L vs SPO Forest | MCFP · GEANT real-life, n=300 | mean regret: 132.07±77.23 -> 68.44±50.76 | Improved |
| B&L vs SPOT | MCFP · GEANT real-life, n=100 | mean regret: 143.57±107.88 -> 72.43±63.58 | Improved |
| B&L vs linear regression | MCVC · POLSKA real-life, n=100 | mean regret: 3.56±2.07 -> 2.22±1.36 | Improved |
| B&L vs linear regression | MCVC · PDH artificial, n=300 | mean regret: 56.59±6.70 -> 53.71±5.62 | Improved |
| B&L vs SPO | MCVC · PDH artificial, n=300 | mean regret: 69.46±6.63 -> 53.71±5.62 | Improved |
| Increase training set size from 100 to 300 | B&L · MCFP USANet real-life | mean regret: 141.37±128.52 -> 122.66±85.52 | Improved |
| Increase training set size from 100 to 300 | B&L · MCFP USANet artificial | mean regret: 1687.10±729.36 -> 1699.24±688.96 | No clear effect |
| Artificial (nonlinear) vs real-life data | B&L · MCFP relative error | relative error: 7.46%-15.68% -> 1.12%-2.07% | Improved |
| B&L vs SPOT on larger instances | MCFP GEANT · real-life, n=300 | runtime (s): 439.19 -> 59.17 | Improved |
| B&L vs SPO Forest on larger instances | MCVC POLSKA · real-life, n=300 | runtime (s): 4277.86 -> 1965.00 | Improved |
Closest papers by meaning
- Parameterizing Branch-and-Bound Search Trees to Learn Branching Policies (AAAI 2021 · similarity 0.89)
- A Solver-Free Training Method for Predict-then-Optimize (ICML 2026 · similarity 0.88)
- Solver-Free Decision-Focused Learning for Linear Optimization Problems (NeurIPS 2025 · similarity 0.87)
- A New Branch-and-Bound Pruning Framework for $\ell_0$-Regularized Problems (ICML 2024 · similarity 0.87)
- A Generalization Result for Convergence in Learning-to-Optimize (ICML 2025 · similarity 0.86)
- Sample Complexity of Algorithm Selection Using Neural Networks and Its Applications to Branch-and-Cut (NeurIPS 2024 · 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.
