Research map / Learning theory and algorithms / Sample complexity
Generalization and limits under data constraints
Bounds learning under adaptivity, active learning, missing data and noisy examples.
72 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Sample complexity, keyword cluster “al · bounds · et”
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.
24 papers in 2023–24 and 18 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 7.2% to 5.5%.
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
- 2016 · Provable Algorithms for Inference in Topic Models (ICML 2016 · 16 citations)
- 2017 · Information Theoretic Properties of Markov Random Fields, and their Algorithmic Applications (NeurIPS 2017 · 21 citations)
- 2018 · Do GANs learn the distribution? Some Theory and Empirics (ICLR 2018 · 111 citations)
- 2019 · Wasserstein Dependency Measure for Representation Learning (NeurIPS 2019 · 19 citations)
- 2020 · Learning-Augmented Data Stream Algorithms (ICLR 2020 · 9 citations)
- 2021 · Rate-Distortion Analysis of Minimum Excess Risk in Bayesian Learning (ICML 2021 · 3 citations)
- 2022 · Learning Bayesian Networks in the Presence of Structural Side Information (AAAI 2022 · 6 citations)
- 2023 · Bypass Exponential Time Preprocessing: Fast Neural Network Training via Weight-Data Correlation Preprocessing (NeurIPS 2023 · 4 citations)
- 2024 · The Impact of Initialization on LoRA Finetuning Dynamics (NeurIPS 2024 · 6 citations)
- 2025 · Forecasting Competitions with Correlated Events (AAAI 2025 · 1 citation)
- 2026 · How High is ‘High’? Rethinking the Roles of Dimensionality in Topological Data Analysis and Manifold Learning (ICML 2026 · 1 citation)
Earliest papers here
- Provable Algorithms for Inference in Topic Models (ICML 2016 · 16 citations)
- On the Recursive Teaching Dimension of VC Classes (NeurIPS 2016 · 7 citations)
- The Limits of Learning with Missing Data (NeurIPS 2016 · 7 citations)
What has been tried
51 papers here report 320 measured changes (a component added, removed or swapped, compared with a baseline): helped 71%, no clear effect 8%, hurt 14%, mixed 7%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Count last-layer FLOPs instead of excluding the head from model size | OpenWebText2 and RefinedWeb · compute-optimal exponent a | 0.864 -> 0.699; 0.835 -> 0.706 | Helped | Resolving Discrepancies in Compute-Optimal Scaling of Language Models (NeurIPS 2024 · 1 citation) |
| Use one-bit instead of bit-wise outer-level mutation | Mcc · expected runtime | O\left(rM\ln(rM)\right); \Omega(rM\ln{M}) | No clear effect | Runtime Analysis of Evolutionary NAS for Multiclass Classification (ICML 2025 · 0 citations) |
| Use the distribution-free threshold sequence instead of Monte Carlo threshold estimation | MNIST digit 0 to digits 1–9 · Expected detection delay | Hurt | Optimal Online Change Detection via Random Fourier Features (NeurIPS 2025 · 0 citations) |
What reviewers will expect
Datasets and baselines named in the 11 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- MNIST · 4 of 11 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 42 papers; from OpenAlex and the papers' affiliations).
- Massachusetts Institute of Technology · 3 papers
- University of California, San Diego · 3 papers
- Google · 3 papers
- Inria · 2 papers
- Stanford University · 2 papers
- University of Washington · 2 papers
- Yale University · 2 papers
- University of Michigan - Ann Arbor · 2 papers
- Tel Aviv University · 2 papers
- Purdue University · 2 papers
Use it
Download BibTeX of 39 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
- Statistical-query lower bounds (85 papers)
- Robust and surrogate-loss learning bounds (83 papers)
- Computational thresholds in graphs and structure (84 papers)
Elsewhere in the topic
- Minimax estimation rates (109 papers)
- Sample complexity of learning distributions (101 papers)
- Optimal mean estimation and subsampling (95 papers)
- Density and property estimation (86 papers)
- Learning halfspaces under noise (84 papers)
- Sample complexity of ranking and selection (80 papers)
- Learning under log-concave distributions (80 papers)
- PAC learnability theory (79 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).