Research map / Architectures and efficiency / RNNs and normalization
Toward Faster and Simpler Matrix Normalization via Rank-1 Update
Tan Yu, Yunfeng Cai, Ping Li
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 RUN normalization (eta=0.6 vs eta=0.0) (ablation) | bilinear pooling · CUB | accuracy (%): 84.1 -> 86.3 | Improved |
| add RUN normalization (eta=0.6 vs eta=0.0) (ablation) | bilinear pooling · DTD | accuracy (%): 65.6 -> 68.4 | Improved |
| add RUN normalization (eta=0.6 vs eta=0.0) (ablation) | bilinear pooling · MIT | accuracy (%): 79.8 -> 80.8 | Improved |
| add RUN normalization (eta=0.6 vs eta=0.0) (ablation) | bilinear pooling · Aircraft | accuracy (%): 88.9 -> 89.8 | No clear effect |
| set eta=2.0 instead of 0.6 (ablation) | RUN · CUB | accuracy (%): 86.3 -> 83.9 | Hurt |
| vary eta within [0.4, 1.5] (ablation) | RUN · CUB | accuracy (%): 86.0 -> 86.4 | No clear effect |
| use K=2 power iterations instead of 5 (ablation) | RUN · Aircraft | accuracy (%) | No clear effect |
| use Tensor Sketch instead of Random Maclaurin for CBP (ablation) | RUN + CBP · CUB | accuracy (%) | No clear effect |
| increase compact feature dimension from 1K to 10K (ablation) | RUN + CBP · CUB | accuracy (%) | Improved |
| replace NS iteration with RUN for normalization | bilinear pooling · GPU time per batch | GPU time (ms): 833ms -> 2.5ms | Improved |
| replace SVD in MoNet-2 with RUN | compact bilinear pooling · GPU time per batch | GPU time (ms): 13850ms -> 2.5ms | Improved |
| replace SVD in MoNet-2 with RUN | compact bilinear pooling · Aircraft | accuracy (%): 86.7 -> 91.0 | Improved |
Closest papers by meaning
- Fast and Compact Bilinear Pooling by Shifted Random Maclaurin (AAAI 2021 · similarity 0.90)
- Towards Faster Training of Global Covariance Pooling Networks by Iterative Matrix Square Root Normalization (CVPR 2018 · similarity 0.87)
- Efficient Compact Bilinear Pooling via Kronecker Product (AAAI 2022 · similarity 0.86)
- Grassmann Pooling as Compact Homogeneous Bilinear Pooling for Fine-Grained Visual Classification (ECCV 2018 · similarity 0.86)
- Large-Scale Low-Rank Matrix Learning with Nonconvex Regularizers (TPAMI 2018 · similarity 0.86)
- Bilinear Factor Matrix Norm Minimization for Robust PCA: Algorithms and Applications (TPAMI 2017 · similarity 0.86)
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.
