Research map / Architectures and efficiency / RNNs and normalization
End-to-End Learning for Optimization via Constraint-Enforcing Approximators
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Train forecasts end-to-end with ProjectNet rather than with a two-stage prediction objective | Multi-Product Newsvendor · 50 products | average cost: 19.5 -> 18.5 | Helped |
| Use ProjectNet instead of OptNet | Multi-Product Newsvendor · 50 products | average cost: 18.7 -> 18.5 | Helped |
| Train forecasts end-to-end with ProjectNet rather than with a two-stage prediction objective | Multi-Product Newsvendor · 100 products | average cost: 6.3 -> 5.6 | Helped |
| Use ProjectNet instead of OptNet | Multi-Product Newsvendor · 100 products | average cost: 5.9 -> 5.6 | Helped |
| Use ProjectNet instead of the Vlastelica et al. end-to-end method | Warcraft Shortest Path · test set | optimal path found: 86.6% -> 83.0% | Hurt |
| Use ProjectNet instead of the Vlastelica et al. end-to-end method | Warcraft Shortest Path · test set | runtime per epoch: 81.3 -> 68.3 | Helped |
| Replace SAA with gradient descent using approximate projection | No-feature Newsvendor · capacity 10 | average cost: 32.528 -> 32.53 | No clear effect |
| Replace SAA with gradient descent using approximate projection | No-feature Newsvendor · capacity 20 | average cost: 9.661 -> 9.669 | No clear effect |
| Replace SAA with gradient descent using approximate projection | No-feature Newsvendor · capacity 30 | average cost: 4.173 -> 4.181 | No clear effect |
| Use ProjectNet rather than projected gradient descent | Maximum Matching · n=50, smaller iteration counts | relative regret: up to 12.5% improvement | Helped |
Closest papers by meaning
- BPQP: A Differentiable Convex Optimization Framework for Efficient End-to-End Learning (NeurIPS 2024 · similarity 0.86)
- Exploring Landscapes for Better Minima along Valleys (NeurIPS 2025 · similarity 0.84)
- Gradient Descent as Loss Landscape Navigation: a Normative Framework for Deriving Learning Rules (NeurIPS 2025 · similarity 0.83)
- DOGE-Train: Discrete Optimization on GPU with End-to-End Training (AAAI 2024 · similarity 0.83)
- Efficient Bisection Projection to Ensure Neural-Network Solution Feasibility for Optimization over General Set (ICML 2025 · similarity 0.83)
- FSNet: Feasibility-Seeking Neural Network for Constrained Optimization with Guarantees (NeurIPS 2025 · similarity 0.83)
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.