Research map / Architectures and efficiency / CNN design
G2DeNet: Global Gaussian Distribution Embedding Network and Its Application to Visual Recognition
Check 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 |
|---|---|---|---|
| insert a trainable global Gaussian embedding layer into AlexNet | AlexNet · MS-COCO 2014 validation | classification error (%): 24.6 -> 21.5 | Helped |
| use an end-to-end global Gaussian embedding layer | VGG-VD16 on Birds-200-2011 · Birds-200-2011 | accuracy (%) | Helped |
| use a global Gaussian embedding layer instead of bilinear pooling | VGG-VD16 on FGVC-Aircraft · FGVC-Aircraft | accuracy (%): 84.1 -> 89.0 | Helped |
| use a global Gaussian embedding layer instead of bilinear pooling | VGG-VD16 on FGVC-Cars · FGVC-Cars | accuracy (%): +1.9% | Helped |
| add bounding boxes during training and testing | G2DeNet on Birds-200-2011 · Birds-200-2011 | accuracy (%): 87.1 -> 87.6 | Helped |
| train G2DeNet end-to-end rather than without training (ablation) | G2DeNet with VGG-VD16 · Birds-200-2011 | accuracy (%): +5.9% | Helped |
| train G2DeNet end-to-end rather than fine-tuning the CNN separately (ablation) | G2DeNet with VGG-VD16 · Birds-200-2011 | accuracy (%): +3.6% | Helped |
| use square-root Gaussian embedding instead of Nakayama embedding (ablation) | G2DeNet · Birds-200-2011 | accuracy (%): 83.5 -> 87.1 | Helped |
| use square-root Gaussian embedding instead of Calvo/Lovrić embedding (ablation) | G2DeNet · Birds-200-2011 | accuracy (%): 84.1 -> 87.1 | Helped |
| use square-root Gaussian embedding instead of Log-Euclidean Gaussian embedding (ablation) | G2DeNet · Birds-200-2011 | accuracy (%): 83.8 -> 87.1 | Helped |
Closest papers by meaning
- Deep Global Generalized Gaussian Networks (CVPR 2019 · similarity 0.88)
- Global Second-Order Pooling Convolutional Networks (CVPR 2019 · similarity 0.88)
- Gaussian Graph Network: Learning Efficient and Generalizable Gaussian Representations from Multi-view Images (NeurIPS 2024 · similarity 0.86)
- Efficient Equivariant Network (NeurIPS 2021 · similarity 0.86)
- Asymptotic Analysis of Two-Layer Neural Networks after One Gradient Step under Gaussian Mixtures Data with Structure (ICLR 2025 · similarity 0.84)
- Improving Visual Prompt Tuning by Gaussian Neighborhood Minimization for Long-Tailed Visual Recognition (NeurIPS 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.