Research map / Architectures and efficiency / Neural architecture search
Latency-aware architecture search
Searches network width and structure under latency, memory, and robustness constraints.
91 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2017–2026) · an approach in Neural architecture search, keyword cluster “differentiable · image · latency”
State History Benchmarks Tried Checklist Preprints Who Export
Where it stands
Quiet None of the tracked benchmarks has a report from 2025–26.
10 papers in 2023–24 and 5 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 13.3% to 10.0%.
No longer reported by papers here: ImageNet (Top-1 accuracy, last in 2023); CIFAR-10 (accuracy, last in 2022); CIFAR-10 test (error (%), last in 2019); CIFAR-10 test (test error (%), last in 2020); CIFAR-100 (accuracy, last in 2022).
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
- 2017 · Neural Architecture Search with Reinforcement Learning (ICLR 2017 · 4,370 citations)
- 2018 · Learning Transferable Architectures for Scalable Image Recognition (CVPR 2018 · 6,603 citations)
- 2019 · MnasNet: Platform-Aware Neural Architecture Search for Mobile (CVPR 2019 · 3,416 citations)
- 2020 · Searching Central Difference Convolutional Networks for Face Anti-Spoofing (CVPR 2020 · 649 citations)
- 2021 · EfficientNetV2: Smaller Models and Faster Training (ICML 2021 · 1,141 citations)
- 2022 · SplitNets: Designing Neural Architectures for Efficient Distributed Computing on Head-Mounted Systems (CVPR 2022 · 25 citations)
- 2023 · MDL-NAS: A Joint Multi-Domain Learning Framework for Vision Transformer (CVPR 2023 · 17 citations)
- 2024 · Hypergraph Neural Architecture Search (AAAI 2024 · 4 citations)
- 2025 · Jet-Nemotron: Efficient Language Model with Post Neural Architecture Search (NeurIPS 2025 · 0 citations)
- 2026 · BIOARC: Discovering Optimal Neural Architectures for Biological Foundation Models (ICML 2026 · 0 citations)
Earliest papers here
- Neural Architecture Search with Reinforcement Learning (ICLR 2017 · 4,370 citations)
- Learning Transferable Architectures for Scalable Image Recognition (CVPR 2018 · 6,603 citations)
- Efficient Neural Architecture Search via Parameters Sharing (ICML 2018 · 1,269 citations)
Benchmark records
The best number each year that beat every earlier one, among papers of this approach reporting their own method. Numbers are extracted by AI from each paper's text; check the paper before citing.
ImageNet · Top-1 accuracy (higher is better)
9 papers, 2019–2023 · not reported since 2023
| Year | Best so far | Set by |
|---|---|---|
| 2019 | 77.1 | Discovering Neural Wirings (NeurIPS 2019 · 61 citations) |
| 2020 | 80 | Once-for-All: Train One Network and Specialize it for Efficient Deployment (ICLR 2020 · 99 citations) |
CIFAR-10 · accuracy (higher is better)
6 papers, 2019–2022 · not reported since 2022
| Year | Best so far | Set by |
|---|---|---|
| 2019 | 98.01 | NAT: Neural Architecture Transformer for Accurate and Compact Architectures (NeurIPS 2019 · 71 citations) |
| 2021 | 99.1 | EfficientNetV2: Smaller Models and Faster Training (ICML 2021 · 1,141 citations) |
CIFAR-10 test · error (%) (lower is better)
4 papers, 2018–2019 · not reported since 2019
| Year | Best so far | Set by |
|---|---|---|
| 2018 | 3.63 | Hierarchical Representations for Efficient Architecture Search (ICLR 2018 · 226 citations) |
| 2019 | 2.6 | IRLAS: Inverse Reinforcement Learning for Architecture Search (CVPR 2019 · 39 citations) |
CIFAR-10 test · test error (%) (lower is better)
4 papers, 2018–2020 · not reported since 2020
| Year | Best so far | Set by |
|---|---|---|
| 2018 | 2.3 | Path-Level Network Transformation for Efficient Architecture Search (ICML 2018 · 150 citations) |
| 2019 | 2.08 | ProxylessNAS: Direct Neural Architecture Search on Target Task and Hardware (ICLR 2019 · 284 citations) |
CIFAR-100 · accuracy (higher is better)
4 papers, 2019–2022 · not reported since 2022
| Year | Best so far | Set by |
|---|---|---|
| 2019 | 73.83 | Learnable Embedding Space for Efficient Neural Architecture Compression (ICLR 2019 · 27 citations) |
| 2020 | 79.51 | Network Adjustment: Channel Search Guided by FLOPs Utilization Ratio (CVPR 2020 · 15 citations) |
| 2021 | 92.3 | EfficientNetV2: Smaller Models and Faster Training (ICML 2021 · 1,141 citations) |
What has been tried
84 papers here report 827 measured changes (a component added, removed or swapped, compared with a baseline): helped 75%, no clear effect 6%, hurt 16%, mixed 2%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Replace HyperNAS search with random architecture selection | Cora, Citeseer, and Pubmed · accuracy | Cora: 83.9 -> 82.8; Citeseer: 74.1 -> 73.6; Pubmed: 81.3 -> 80.7 | Hurt | Hypergraph Neural Architecture Search (AAAI 2024 · 4 citations) |
| use learned full-attention layer placement instead of uniform placement | MMLU · accuracy | Helped | Jet-Nemotron: Efficient Language Model with Post Neural Architecture Search (NeurIPS 2025 · 0 citations) |
What reviewers will expect
Datasets and baselines named in the 5 papers of this approach from 2024–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- ImageNet · 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 15 papers; from OpenAlex and the papers' affiliations).
- Huawei · 3 papers
- University of Alberta · 2 papers
- Tsinghua University · 2 papers
Use it
Download BibTeX of 28 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
- AutoML pipeline search (57 papers)
- Differentiable architecture search (DARTS) (32 papers)
Elsewhere in the topic
- Efficient NAS search strategies (86 papers)
- Bi-level differentiable NAS (84 papers)
- One-shot and few-shot supernets (30 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).