Research map / Architectures and efficiency / Quantization
SiMaN: Sign-to-Magnitude Network Binarization
Mingbao Lin, Rongrong Ji, Zihan Xu, Baochang Zhang, Fei Chao, Chia‐Wen Lin, Ling Shao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add half-half binarization (top-half magnitude to +1) on top of l2-regularized weights (ablation) | SiMaN · ResNet-18 ImageNet | top-1 accuracy (%): 55.1 -> 59.2 | Improved |
| remove l2 regularization on binarized weights (no half-half) (ablation) | SiMaN · ResNet-18 ImageNet | top-1 accuracy (%): 55.1 -> 57.3 | Improved |
| remove l2 regularization with half-half binarization (ablation) | SiMaN · ResNet-18 ImageNet | top-1 accuracy (%): 59.2 -> 60.1 | Improved |
| remove l2 regularization with half-half binarization (ablation) | SiMaN · ResNet-18 ImageNet | top-5 accuracy (%): 81.5 -> 82.3 | Improved |
| add half-half binarization without l2 regularization (ablation) | SiMaN · ResNet-18 ImageNet | top-1 accuracy (%): 57.3 -> 60.1 | Improved |
| replace RBNN sign-based binarization with SiMaN | binarized ResNet-18 · ImageNet | top-1 accuracy (%): 59.9 -> 60.1 | Improved |
| replace RBNN with SiMaN | binarized ResNet-34 · ImageNet | top-1 accuracy (%): 63.1 -> 63.9 | Improved |
| replace RBNN with SiMaN | binarized VGG-small · CIFAR-10 | top-1 accuracy (%): 91.3 -> 92.5 | Improved |
| replace RBNN with SiMaN | binarized ResNet-18 · CIFAR-10 | top-1 accuracy (%): 92.2 -> 92.5 | Improved |
| replace RBNN with SiMaN | binarized ResNet-20 · CIFAR-10 | top-1 accuracy (%): 87.8 -> 87.4 | Hurt |
| use SiMaN on ReActNet structure instead of ReActNet training | ReActNet ResNet-18 · ImageNet | top-1 accuracy (%): 65.9 -> 66.1 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- XNOR-Net — XNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networks (ECCV 2016) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Sparsity-Inducing Binarized Neural Networks (AAAI 2020 · similarity 0.91)
- Understanding Neural Network Binarization with Forward and Backward Proximal Quantizers (NeurIPS 2023 · similarity 0.90)
- Forward and Backward Information Retention for Accurate Binary Neural Networks (CVPR 2020 · similarity 0.88)
- OvSW: Overcoming Silent Weights for Accurate Binary Neural Networks (ECCV 2024 · similarity 0.88)
- Distribution-Sensitive Information Retention for Accurate Binary Neural Network (IJCV 2022 · similarity 0.88)
- Training Binary Neural Networks via Gaussian Variational Inference and Low-Rank Semidefinite Programming (NeurIPS 2024 · similarity 0.86)
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.
