Research map / Architectures and efficiency / CNN design
Channel Equilibrium Networks for Learning Deep Representation
Wenqi Shao, Shitao Tang, Xingang Pan, Ping Tan, Xiaogang Wang, Ping Luo
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 CE block | ResNet50 · ImageNet val | top-1 accuracy: 76.6 -> 78.3 | Improved |
| Add CE block | ResNet18 · ImageNet val | top-1 accuracy: 70.4 -> 71.9 | Improved |
| Add CE block | ResNet101 · ImageNet val | top-1 accuracy: 78.0 -> 79.0 | Improved |
| Add CE block | MobileNetv2 1× · ImageNet val | top-1 accuracy: 72.5 -> 74.6 | Improved |
| Add CE block | ShuffleNetv2 0.5× · ImageNet val | top-1 accuracy: 59.2 -> 60.5 | Improved |
| Add CE block | ShuffleNetv2 1× · ImageNet val | top-1 accuracy: 69.0 -> 71.2 | Improved |
| Add CE block | Mask R-CNN ResNet50 · COCO2017 mini-val | AP^b: 38.6 -> 40.8 | Improved |
| Add CE block | Mask R-CNN ResNet50 · COCO2017 mini-val | AP^m: 34.2 -> 36.9 | Improved |
| Use synchronized CE statistics | Mask R-CNN ResNet50 · COCO2017 mini-val | AP^b: 40.8 -> 42.0 | Improved |
| Use the BD branch without IR (ablation) | ResNet50 · ImageNet val | top-1 accuracy: 76.6 -> 77.0 | Improved |
| Use the IR branch without BD (ablation) | ResNet50 · ImageNet val | top-1 accuracy: 76.6 -> 77.3 | Improved |
| Combine BD and IR branches (ablation) | ResNet50 · ImageNet val | top-1 accuracy: 76.6 -> 78.3 (+1.7) | Improved |
Closest papers by meaning
- Cross-channel Communication Networks (NeurIPS 2019 · similarity 0.85)
- Training a Predictive Coding Network on ImageNet using Equilibrium Propagation (NeurIPS 2026 · similarity 0.83)
- Separation and Bias of Deep Equilibrium Models on Expressivity and Learning Dynamics (NeurIPS 2024 · similarity 0.82)
- Positive Concave Deep Equilibrium Models (ICML 2024 · similarity 0.81)
- Unveiling The Matthew Effect Across Channels: Assessing Layer Width Sufficiency via Weight Norm Variance (NeurIPS 2024 · similarity 0.81)
- Convolutional Channel-Wise Competitive Learning for the Forward-Forward Algorithm (AAAI 2024 · similarity 0.81)
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.
