Research map / Optimization / Optimizers
Hypergradient bilevel optimization
Computes approximate hypergradients to solve bilevel problems for hyperparameter optimization.
35 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2018–2026) · an approach in Optimizers, keyword cluster “bilevel · inner · shot”
State History Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
10 papers in 2023–24 and 6 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 4.1% to 2.1%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2018 · Bilevel Programming for Hyperparameter Optimization and Meta-Learning (ICML 2018 · 105 citations)
- 2019 · Self-Tuning Networks: Bilevel Optimization of Hyperparameters using Structured Best-Response Functions (ICLR 2019 · 26 citations)
- 2020 · A Generic First-Order Algorithmic Framework for Bi-Level Programming Beyond Lower-Level Singleton (ICML 2020 · 34 citations)
- 2021 · Bilevel Optimization: Convergence Analysis and Enhanced Design (ICML 2021 · 31 citations)
- 2022 · A Fully Single Loop Algorithm for Bilevel Optimization without Hessian Inverse (AAAI 2022 · 16 citations)
- 2023 · Implicit Bilevel Optimization: Differentiating through Bilevel Optimization Programming (AAAI 2023 · 5 citations)
- 2024 · Functional Bilevel Optimization for Machine Learning (NeurIPS 2024 · 2 citations)
- 2025 · A Stochastic Approach to Bi-Level Optimization for Hyperparameter Optimization and Meta Learning (AAAI 2025 · 0 citations)
- 2026 · Natural Hypergradient Descent: Algorithm Design, Convergence Analysis, and Parallel Implementation (ICML 2026 · 0 citations)
Earliest papers here
- Bilevel Programming for Hyperparameter Optimization and Meta-Learning (ICML 2018 · 105 citations)
- Self-Tuning Networks: Bilevel Optimization of Hyperparameters using Structured Best-Response Functions (ICLR 2019 · 26 citations)
- A Generic First-Order Algorithmic Framework for Bi-Level Programming Beyond Lower-Level Singleton (ICML 2020 · 34 citations)
What has been tried
30 papers here report 303 measured changes (a component added, removed or swapped, compared with a baseline): helped 71%, no clear effect 11%, hurt 14%, mixed 5%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Increase unrolled depth | StreamingQA · F1 | Helped | Memory-Efficient Gradient Unrolling for Large-Scale Bi-level Optimization (NeurIPS 2024 · 1 citation) | |
| Increase the SGLD temperature from tau=10^-10 to tau=10^-2 | Noisy synthetic 1D · lambda* error | 0.0019 -> 0.0008 | Helped | A Stochastic Approach to Bi-Level Optimization for Hyperparameter Optimization and Meta Learning (AAAI 2025 · 0 citations) |
| Add saving inner gradients to mixed mode and block rematerialisation | GPU · HBM (G) | 174.8 -> 54.8 | Helped | Scalable Meta-Learning via Mixed-Mode Differentiation (ICML 2025 · 0 citations) |
| Use conjugate gradient instead of gradient descent for solving the subproblems | Deterministic bilevel optimization · Hessian-vector products | O(\kappa^4 /\epsilon) -> O(\kappa^{3.5} \log \kappa/\epsilon) | Helped | Efficient Curvature-Aware Hypergradient Approximation for Bilevel Optimization (ICML 2025 · 0 citations) |
| Smooth Swish activation search space versus non-smooth activations in NAS | NAS CIFAR-10 · Test Top-1 acc (%) | 95.870 (smooth) vs 96.070 (non-smooth) | Hurt | Moreau Envelope for Nonconvex Bi-Level Optimization: A Single-Loop and Hessian-Free Solution Strategy (ICML 2024 · 0 citations) |
| vary the FOSL penalty parameter λ | CIFAR100 and CelebA · test AUC | No clear effect | First-Order Minimax Bilevel Optimization (NeurIPS 2024 · 0 citations) |
What reviewers will expect
Datasets and baselines named in the 5 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Baselines
- AmIGO · 2 of 5 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 16 papers; from OpenAlex and the papers' affiliations).
- Southern University of Science and Technology · 3 papers
- University of Edinburgh · 2 papers
- Dalian University of Technology · 2 papers
- State University of New York at Buffalo · 2 papers
Use it
Download BibTeX of 31 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Gradient-based meta-learning theory (40 papers)
- Differentiating through inner loops (23 papers)
Elsewhere in the topic
- Stochastic gradient MCMC (142 papers)
- Surrogate losses and robust solvers (119 papers)
- Differentiating through embedded optimization (117 papers)
- Constrained and non-backprop training (97 papers)
- Memory-efficient LLM optimizers (79 papers)
- Adam convergence and adaptive learning rates (78 papers)
- Natural gradient and Fisher approximations (77 papers)
- Randomized gradient estimation (77 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).