Research map / Architectures and efficiency / Quantization
Binary neural networks
Binarizes weights and activations with information-retention and ensemble techniques.
27 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2017–2026) · an approach in Quantization, keyword cluster “mixed precision · floating · point”
State History Benchmarks Tried Checklist Preprints Export
Where it stands
Few recent reports The tracked benchmarks have at most one report from 2025–26 that could beat the old record, too few to say whether they are still moving.
5 papers in 2023–24 and 7 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 4.2% to 2.1%.
No longer reported by papers here: ImageNet, ResNet-18 (Top-1 accuracy, last in 2024); ImageNet, ResNet-18 (Top-5 accuracy, last in 2024); ImageNet, ResNet-34 (Top-1 accuracy, last in 2024).
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 · How to Train a Compact Binary Neural Network with High Accuracy? (AAAI 2017 · 260 citations)
- 2018 · Apprentice: Using Knowledge Distillation Techniques To Improve Low-Precision Network Accuracy (ICLR 2018 · 148 citations)
- 2019 · Binary Ensemble Neural Network: More Bits per Network or More Networks per Bit? (CVPR 2019 · 149 citations)
- 2020 · Forward and Backward Information Retention for Accurate Binary Neural Networks (CVPR 2020 · 348 citations)
- 2021 · SA-BNN: State-Aware Binary Neural Network (AAAI 2021 · 16 citations)
- 2022 · BiBERT: Accurate Fully Binarized BERT (ICLR 2022 · 33 citations)
- 2023 · Fast and Accurate Binary Neural Networks Based on Depth-Width Reshaping (AAAI 2023 · 3 citations)
- 2024 · BiPer: Binary Neural Networks using a Periodic Function (CVPR 2024 · 12 citations)
- 2025 · Predicting High-precision Depth on Low-Precision Devices Using 2D Hilbert Curves (ICML 2025 · 0 citations)
- 2026 · Zeros can be Informative: Masked Binary U-Net for Image Segmentation on Tensor Cores (ICLR 2026 · 0 citations)
Earliest papers here
- How to Train a Compact Binary Neural Network with High Accuracy? (AAAI 2017 · 260 citations)
- Loss-aware Binarization of Deep Networks (ICLR 2017 · 167 citations)
- Apprentice: Using Knowledge Distillation Techniques To Improve Low-Precision Network Accuracy (ICLR 2018 · 148 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.
CIFAR-10 · accuracy (%) (higher is better)
4 papers, 2018–2025 · one report since 2024, not better
| Year | Best so far | Set by |
|---|---|---|
| 2018 | 89.4 | Heterogeneous Bitwidth Binarization in Convolutional Neural Networks (NeurIPS 2018 · 29 citations) |
| 2023 | 92.6 | Fast and Accurate Binary Neural Networks Based on Depth-Width Reshaping (AAAI 2023 · 3 citations) |
ImageNet, ResNet-18 · Top-1 accuracy (higher is better)
4 papers, 2020–2024 · not reported since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 66.5 | Forward and Backward Information Retention for Accurate Binary Neural Networks (CVPR 2020 · 348 citations) |
ImageNet, ResNet-18 · Top-5 accuracy (higher is better)
4 papers, 2020–2024 · not reported since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 81.9 | Rotated Binary Neural Network (NeurIPS 2020 · 65 citations) |
| 2022 | 82.86 | Equal Bits: Enforcing Equally Distributed Binary Network Weights (AAAI 2022 · 10 citations) |
| 2024 | 83.14 | BiPer: Binary Neural Networks using a Periodic Function (CVPR 2024 · 12 citations) |
ImageNet, ResNet-34 · Top-1 accuracy (higher is better)
4 papers, 2020–2024 · not reported since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 70.4 | Forward and Backward Information Retention for Accurate Binary Neural Networks (CVPR 2020 · 348 citations) |
What has been tried
21 papers here report 220 measured changes (a component added, removed or swapped, compared with a baseline): helped 75%, no clear effect 8%, hurt 15%, 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 |
|---|---|---|---|---|
| Use periodic binarization rather than sign binarization in the second training stage | CIFAR-10, ResNet-18/20 · Precision | 93.01 -> 93.93 | Helped | BiPer: Binary Neural Networks using a Periodic Function (CVPR 2024 · 12 citations) |
| Replace fixed-slope RLeakyReLU (slope 2−3) with quantized RPReLU | ImageNet · Top-1 accuracy | 60.25 -> 61.39; 1.14% | Helped | A&B BNN: Add&Bit-Operation-Only Hardware-Friendly Binary Neural Network (CVPR 2024 · 6 citations) |
| Set the number of EOPM perturbed models per iteration to a larger value | CIFAR-10, ResNet18, 1% BER · perturbed accuracy and stability | Helped | NeuralFuse: Learning to Recover the Accuracy of Access-Limited Neural Network Inference in Low-Voltage Regimes (NeurIPS 2024 · 1 citation) | |
| Use h3 rather than h1 Hilbert curve components | ScanNet v2 test, W8A8 DSP · Abs Rel, % | 1.50 -> 0.93 | Helped | Predicting High-precision Depth on Low-Precision Devices Using 2D Hilbert Curves (ICML 2025 · 0 citations) |
| Use TASU activations (binary layer outputs) instead of RASU | MNIST · accuracy | Hurt | G-Net: A Provably Easy Construction of High-Accuracy Random Binary Neural Networks (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
- CIFAR-10 · 4 of 5 papers
- CIFAR-100 · 2 of 5 papers
Recent preprints
Loading recent arXiv preprints…
Use it
Download BibTeX of 26 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
- Mixed-precision quantization (92 papers)
- Low-precision training formats (58 papers)
- Ultra-low-bit LLM quantization (56 papers)
Elsewhere in the topic
- Post-training quantization of LLMs (136 papers)
- Post-training quantization of ViTs (105 papers)
- Post-training weight quantization (100 papers)
- KV cache quantization (72 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).