Research map / Architectures and efficiency / CNN design
Bayesian Optimized 1-Bit CNNs
Jiaxin Gu, Junhe Zhao, Xiaolong Jiang, Baochang Zhang, Jianzhuang Liu, Guodong Guo, Rongrong Ji
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 Bayesian kernel loss (ablation) | BONN ResNet18 · ImageNet val | top-1 accuracy (%): 56.3 -> 58.3 | Improved |
| add Bayesian feature loss (ablation) | BONN ResNet18 · ImageNet val | top-1 accuracy (%): 56.3 -> 58.4 | Improved |
| add both Bayesian losses (ablation) | BONN ResNet18 · ImageNet val | top-1 accuracy (%): 56.3 -> 59.3 | Improved |
| add both Bayesian losses (ablation) | BONN ResNet18 · ImageNet val | top-5 accuracy (%): 79.8 -> 81.6 | Improved |
| add feature loss on top of kernel loss (ablation) | BONN ResNet18 · ImageNet val | top-1 accuracy (%): 58.3 -> 59.3 | Improved |
| add kernel loss on top of feature loss (ablation) | BONN ResNet18 · ImageNet val | top-1 accuracy (%): 58.4 -> 59.3 | Improved |
| replace XNOR-Net with BONN | WRN22 16-16-32-64 · CIFAR-10 | test accuracy (%): 81.90 -> 87.34 | Improved |
| replace XNOR-Net with BONN | WRN22 16-16-32-64 · CIFAR-100 | test accuracy (%): 53.17 -> 60.91 | Improved |
| replace XNOR-Net with BONN | WRN22 64-64-128-256 · CIFAR-10 | test accuracy (%): 88.52 -> 92.36 | Improved |
| replace XNOR-Net with BONN | ResNet18 · ImageNet val | top-1 accuracy (%): 51.2 -> 59.3 | Improved |
| replace Bi-Real Net with BONN | ResNet18 · ImageNet val | top-1 accuracy (%): 56.4 -> 59.3 | Improved |
| replace PCNN with BONN | ResNet18 · ImageNet val | top-1 accuracy (%): 57.3 -> 59.3 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- XNOR-Net — XNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networks (ECCV 2016) · better on 5 of 5 shared results
- Bi-Real Net — Bi-Real Net: Enhancing the Performance of 1-bit CNNs with Improved Representational Capability and Advanced Training Algorithm (ECCV 2018) · better on 2 of 2 shared results
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
- Towards Compact 1-bit CNNs via Bayesian Learning (IJCV 2021 · similarity 0.91)
- Bi-Real Net: Enhancing the Performance of 1-bit CNNs with Improved Representational Capability and Advanced Training Algorithm (ECCV 2018 · similarity 0.87)
- TBN: Convolutional Neural Network with Ternary Inputs and Binary Weights (ECCV 2018 · similarity 0.86)
- Modulated Convolutional Networks (CVPR 2018 · similarity 0.86)
- A Bayesian Approach to Data Point Selection (NeurIPS 2024 · similarity 0.85)
- Position: Bayesian Deep Learning is Needed in the Age of Large-Scale AI (ICML 2024 · similarity 0.84)
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.
