Research map / Architectures and efficiency / CNN design
Detail-Preserving Pooling in Deep Networks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Replace VGG-16 max pooling with Lite-DPPSym | VGG-16 · ImageNet validation, single crop | Top-1 error (%): 30.40 -> 30.05 | Helped |
| Replace ResNet downsampling with Lite-DPPSym | ResNet-50 · ImageNet validation, single crop | Top-1 error (%): 24.23 -> 23.42 | Helped |
| Replace ResNet downsampling with Lite-DPPSym | ResNet-101 · ImageNet validation, ten crop | Top-5 error (%): 5.38 -> 5.20 | Helped |
| Replace both downsampling steps with DPP (ablation) | ResNet-110 · CIFAR10 | error (%): 6.89 -> 6.59 | Helped |
| Replace only the second, deeper downsampling step with DPP (ablation) | ResNet-110 · CIFAR10 | error (%): 6.89 -> 6.68 | Helped |
| Replace transition-layer average pooling with Lite-S3DPPSym (ablation) | DenseNet-BC · CIFAR10 | error (%): 3.90 -> 3.75 | Helped |
| Use symmetric rather than asymmetric reward weights (ablation) | DPP on VGG · CIFAR10 | error (%): 7.10±0.15 -> 7.19±0.10 | Mixed |
| Add uniform stochastic spatial sampling to symmetric DPP (ablation) | ResNet-110 · CIFAR10 | error (%): 5.97 -> 5.42 | Helped |
Closest papers by meaning
- S3Pool: Pooling With Stochastic Spatial Sampling (CVPR 2017 · similarity 0.90)
- Wavelet Pooling for Convolutional Neural Networks (ICLR 2018 · similarity 0.90)
- Building Detail-Sensitive Semantic Segmentation Networks With Polynomial Pooling (CVPR 2019 · similarity 0.89)
- Gaussian-Based Pooling for Convolutional Neural Networks (NeurIPS 2019 · similarity 0.88)
- TI-Pooling: Transformation-Invariant Pooling for Feature Learning in Convolutional Neural Networks (CVPR 2016 · similarity 0.88)
- Beyond Skip Connection: Pooling and Unpooling Design for Elimination Singularities (AAAI 2025 · similarity 0.87)
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.