Research map / Detection and segmentation / CNN features
Auto Learning Attention
Benteng Ma, Jing Zhang, Yong Xia, Dacheng Tao
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 searched HOGA attention (AutoLA) | ResNet20 · CIFAR10 | accuracy: 91.95 -> 93.38 | Improved |
| add AutoLA instead of CBAM | ResNet50 · ImageNet | top-1 error: 22.66 -> 21.82 | Improved |
| add AutoLA | ResNet50 SSD · COCO | AP: 25.01 -> 27.78 | Improved |
| add AutoLA instead of SE | ResNet50 · COCO keypoints | AP: 73.6 -> 74.6 | Improved |
| reduce group number from K=4 to K=2 (ablation) | AutoLA · CIFAR10 ResNet20 | accuracy: 93.38 -> 93.18 | Hurt |
| use differentiable search instead of random search (ablation) | HOGA · CIFAR10 ResNet32 | accuracy: 94.15 -> 94.33 | Improved |
| search HOGA directly on target backbone instead of transferring from ResNet56 (ablation) | AutoLA · CIFAR10 ResNet20 | accuracy: 93.31 -> 93.38 | Improved |
| use searched HOGA instead of widened SE (ablation) | ResNet20 · CIFAR10 | accuracy: 93.16 -> 93.38 | Improved |
| use searched ops instead of CBAM in all HOGA edges (ablation) | HOGA k=4 · CIFAR10 ResNet20 | accuracy: 93.07 -> 93.38 | Improved |
Closest papers by meaning
- AttentionRNN: A Structured Spatial Attention Mechanism (ICCV 2019 · similarity 0.83)
- Explicitly Modeled Attention Maps for Image Classification (AAAI 2021 · similarity 0.83)
- Improving Fine-Grained Visual Recognition in Low Data Regimes via Self-Boosting Attention Mechanism (ECCV 2022 · similarity 0.83)
- MCA: Moment Channel Attention Networks (AAAI 2024 · similarity 0.83)
- SAS: Simulated Attention Score (NeurIPS 2025 · similarity 0.83)
- Attention-based PCA (NeurIPS 2026 · 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.
