Research map / Architectures and efficiency / CNN design
Fast and Compact Bilinear Pooling by Shifted Random Maclaurin
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 |
|---|---|---|---|
| add shifted projections (SRM+ with K=8 instead of RM+ with K=1) (ablation) | FCBN · CUB | accuracy: 83.5 -> 85.6 | Helped |
| add shifted projections (SRM+ with K=8 instead of RM+ with K=1) (ablation) | FCBN · MIT | accuracy: 78.3 -> 80.2 | Helped |
| increase K from 4 to 8 (ablation) | FCBN · Aircraft | accuracy | Hurt |
| increase projection dimension D from 250 to 1000 (ablation) | SRM+ · CUB | accuracy: 85.0 -> 85.8 | Helped |
| add matrix normalization (ablation) | FCBN · CUB | accuracy: 83.7 -> 85.6 | Helped |
| add matrix normalization (ablation) | FCBN · MIT | accuracy: 78.1 -> 80.2 | Helped |
| add matrix normalization (ablation) | FCBN · DTD | accuracy: 66.1 -> 66.8 | Helped |
| add matrix normalization (ablation) | FCBN · Aircraft | accuracy: 87.7 -> 90.5 | Helped |
| replace RM+ with shifted projections (SRM+) (ablation) | FCBN, 4K compact feature · CUB | accuracy: 85.2 -> 85.6 | Helped |
| replace RM+ with shifted projections (SRM+) (ablation) | FCBN, 4K compact feature · MIT | accuracy: 80.5 -> 80.2 | Hurt |
| replace RM+ with shifted projections (SRM+) (ablation) | FCBN, 8K compact feature · DTD | accuracy: 67.5 -> 67.8 | Helped |
| replace RM+ with shifted projections (SRM+) (ablation) | FCBN, 8K compact feature · Aircraft | accuracy: 90.2 -> 90.5 | Helped |
Closest papers by meaning
- Efficient Compact Bilinear Pooling via Kronecker Product (AAAI 2022 · similarity 0.90)
- Compact Bilinear Pooling (CVPR 2016 · similarity 0.90)
- Low-Rank Bilinear Pooling for Fine-Grained Classification (CVPR 2017 · similarity 0.88)
- Revisiting Bilinear Pooling: A Coding Perspective (AAAI 2020 · similarity 0.86)
- Random features for Grassmannian kernel approximation with bounded rank-one projections (NeurIPS 2026 · similarity 0.82)
- Stochastic Layer-Wise Shuffle for Improving Vision Mamba Training (ICML 2025 · similarity 0.82)
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.