Research map / Optimization / Optimizers
A Generic First-Order Algorithmic Framework for Bi-Level Programming Beyond Lower-Level Singleton
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 |
|---|---|---|---|
| Aggregate upper-level and lower-level descent information when updating the lower-level variable | BDA vs. RHG · MNIST data hyper-cleaning | accuracy: 88.96 -> 89.12; 89.73 -> 90.12; 90.13 -> 90.57; 90.19 -> 90.81; 90.15 -> 90.86 | Helped |
| Aggregate upper-level and lower-level descent information when updating the lower-level variable | BDA vs. RHG · MiniImageNet 5-way 1-shot | accuracy: 48.89 -> 49.08 | Helped |
| Use BDA instead of RHG | few-shot learning · MiniImageNet 5-way 1-shot, near 44% accuracy | UL iterations: 3300 -> 2500 | Helped |
| Use BDA instead of the strongest listed baseline | few-shot learning · Omniglot 5-way 1-shot | accuracy: 98.70 -> 99.04 | Helped |
| Use BDA instead of MAML | few-shot learning · Omniglot 5-way 5-shot | accuracy: 99.91 -> 99.62 | Hurt |
| Use BDA instead of the strongest listed baseline | few-shot learning · Omniglot 20-way 1-shot | accuracy: 95.80 -> 96.50 | Helped |
| Use BDA instead of the strongest listed baseline | few-shot learning · Omniglot 20-way 5-shot | accuracy: 98.95 -> 99.10 | Helped |
| Increase the number of lower-level iterations (ablation) | BDA · MNIST data hyper-cleaning | accuracy: 89.12 -> 90.86 | Helped |
| Remove upper-level information from the lower-level update (set alpha_k=0) (ablation) | BDA · Synthetic BLP | solution quality | Hurt |
| Use a fixed aggregation parameter instead of an adaptive schedule (ablation) | BDA · Synthetic BLP | convergence speed | Hurt |
| Use an adaptive aggregation parameter schedule (ablation) | BDA · Synthetic BLP | convergence behavior | Helped |
| Increase lower-level iterations for RHG (ablation) | RHG · Synthetic BLP | solution quality | No clear effect |
Closest papers by meaning
- Constrained Bi-Level Optimization: Proximal Lagrangian Value Function Approach and Hessian-free Algorithm (ICLR 2024 · similarity 0.87)
- qNBO: quasi-Newton Meets Bilevel Optimization (ICLR 2025 · similarity 0.86)
- Memory-Efficient Gradient Unrolling for Large-Scale Bi-level Optimization (NeurIPS 2024 · similarity 0.86)
- Moreau Envelope for Nonconvex Bi-Level Optimization: A Single-Loop and Hessian-Free Solution Strategy (ICML 2024 · similarity 0.86)
- Select-then-differentiate: Solving Bilevel Optimization with Manifold Lower-level Solution Sets (NeurIPS 2026 · similarity 0.86)
- Optimal Hessian/Jacobian-Free Nonconvex-PL Bilevel Optimization (ICML 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.