Research map / Detection and segmentation / CNN features
AutoBSS: An Efficient Algorithm for Block Stacking Style Search
Yikang Zhang, Jian Zhang, Zhao Zhong
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 |
|---|---|---|---|
| search BSS with AutoBSS | ResNet18 · ImageNet | top-1 accuracy (350 ep): 72.19 -> 73.91 | Improved |
| search BSS with AutoBSS | ResNet50 · ImageNet | top-1 accuracy (350 ep): 77.69 -> 79.29 | Improved |
| search BSS with AutoBSS | MobileNetV2 · ImageNet | top-1 accuracy (350 ep): 73.90 -> 74.50 | Improved |
| search BSS with AutoBSS | EfficientNet-B0 · ImageNet | top-1 accuracy: 77.12 -> 77.79 | Improved |
| search BSS with AutoBSS | EfficientNet-B1 · ImageNet | top-1 accuracy: 79.19 -> 79.48 | Improved |
| use AutoBSS instead of random search of 64 BSSC | ResNet18 · ImageNet | top-1 accuracy (120 ep): 72.34 -> 73.22 | Improved |
| use AutoBSS instead of random search of 64 BSSC | ResNet50 · ImageNet | top-1 accuracy (120 ep): 77.48 -> 78.17 | Improved |
| use AutoBSS instead of random search of 64 BSSC | MobileNetV2 · ImageNet | top-1 accuracy (120 ep): 72.13 -> 72.96 | Improved |
| use AutoBSS instead of random search of 64 BSSC | EfficientNet-B0 · ImageNet | top-1 accuracy: 76.73 -> 77.79 | Improved |
| add BSS Clustering to the search (ablation) | AutoBSS on ResNet18 · ImageNet | mean top-1 of best 5 BSS: 72.62 -> 73.08 | Improved |
| use AutoBSS instead of uniform rescaling or pruning for compression | MobileNetV2 at 130M FLOPs · ImageNet | top-1 accuracy: 68.05 -> 69.65 | Improved |
| search BSS with AutoBSS | Mask R-CNN-R50 · COCO | AP bbox: 39.24 -> 40.15 | 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.
- MobileNetV2 — MobileNetV2: Inverted Residuals and Linear Bottlenecks (CVPR 2018) · better on 2 of 2 shared results
- PSANet50 — PSANet: Point-wise Spatial Attention Network for Scene Parsing (ECCV 2018) · 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
- BossNAS Family: Block-Wisely Self-Supervised Neural Architecture Search (TPAMI 2025 · similarity 0.84)
- BlockQNN: Efficient Block-Wise Neural Network Architecture Generation (TPAMI 2020 · similarity 0.84)
- Practical Block-Wise Neural Network Architecture Generation (CVPR 2018 · similarity 0.83)
- BlockSwap: Fisher-guided Block Substitution for Network Compression on a Budget (ICLR 2020 · similarity 0.83)
- Block-Wisely Supervised Neural Architecture Search With Knowledge Distillation (CVPR 2020 · similarity 0.83)
- Efficient Few-Shot Neural Architecture Search by Counting the Number of Nonlinear Functions (AAAI 2025 · 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.
