Research map / Architectures and efficiency / Quantization
Network Binarization via Contrastive Learning
Yuzhang Shang, Dan Xu, Ziliang Zong, Liqiang Nie, Yan Yan
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 CMIM loss | IR-Net ResNet-18 · CIFAR-100 | top-1 accuracy: 64.5 -> 71.2 | Improved |
| add CMIM loss | RBNN ResNet-18 · CIFAR-100 | top-1 accuracy: 65.3 -> 71.4 | Improved |
| add CMIM loss | RBNN ResNet-20 · CIFAR-10 | top-1 accuracy: 87.0 -> 87.6 | Improved |
| add CMIM loss | RBNN VGG-small · CIFAR-100 | top-1 accuracy: 67.4 -> 71.0 | Improved |
| add CMIM loss | RBNN ResNet-18 · ImageNet | top-1 accuracy: 59.9 -> 62.5 | Improved |
| add CMIM loss | RBNN ResNet-34 · ImageNet | top-1 accuracy: 63.1 -> 65.0 | Improved |
| add CMIM loss | ReActNet · ImageNet | top-1 accuracy: 69.4 -> 71.0 | Improved |
| add CMIM loss with lambda 1.6 (ablation) | IR-Net ResNet-18 · CIFAR-10 | top-1 accuracy: 87.59 -> 92.18 | Improved |
| add CMIM loss with lambda 6.4 (ablation) | IR-Net ResNet-18 · CIFAR-100 | top-1 accuracy: 64.53 -> 71.19 | Improved |
| add CMIM loss with lambda 0.8 (ablation) | IR-Net ResNet-18 · ImageNet-1K | top-1 accuracy: 58.03 -> 61.22 | Improved |
| raise lambda from 0.8 to 12.8 (ablation) | IR-Net ResNet-18 · ImageNet-1K | top-1 accuracy: 61.22 -> 59.7 | Hurt |
| raise lambda from 1.6 to 12.8 (ablation) | IR-Net ResNet-18 · CIFAR-10 | top-1 accuracy: 92.18 -> 91.01 | Hurt |
Closest papers by meaning
- Understanding Neural Network Binarization with Forward and Backward Proximal Quantizers (NeurIPS 2023 · similarity 0.90)
- Distribution-Sensitive Information Retention for Accurate Binary Neural Network (IJCV 2022 · similarity 0.90)
- Sparsity-Inducing Binarized Neural Networks (AAAI 2020 · similarity 0.89)
- High-Capacity Expert Binary Networks (ICLR 2021 · similarity 0.89)
- BiPer: Binary Neural Networks using a Periodic Function (CVPR 2024 · similarity 0.86)
- 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.
