Research map / Architectures and efficiency / RNNs and normalization
Dynamic Routing Between Capsules
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 |
|---|---|---|---|
| Use CapsNet with 3 routing iterations and reconstruction instead of the baseline CNN | MNIST · MNIST test | test error (%): $0.39$ -> $\bm{0.25}_{\pm 0.005}$ | Helped |
| Use CapsNet with 3 routing iterations and reconstruction instead of the baseline CNN | MultiMNIST · MultiMNIST test | classification error (%): $8.1$ -> $\bm{5.2}$ | Helped |
| Use CapsNet rather than a traditional convolutional model | MNIST-trained models evaluated on affNIST · affNIST test | accuracy (%): 66% -> 79% | Helped |
| Use 3 routing iterations with reconstruction rather than 1 routing iteration without reconstruction (ablation) | CapsNet · MNIST test | test error (%): $0.34_{\pm 0.032}$ -> $\bm{0.25}_{\pm 0.005}$ | Helped |
| Add reconstruction loss (ablation) | CapsNet with 1 routing iteration · MNIST test | test error (%): $0.34_{\pm 0.032}$ -> $0.29_{\pm 0.011}$ | Helped |
| Add reconstruction loss (ablation) | CapsNet with 3 routing iterations · MNIST test | test error (%): $0.35_{\pm 0.036}$ -> $\bm{0.25}_{\pm 0.005}$ | Helped |
| Increase routing iterations from 1 to 3 (ablation) | CapsNet with reconstruction · MNIST test | test error (%): $0.29_{\pm 0.011}$ -> $\bm{0.25}_{\pm 0.005}$ | Helped |
| Increase routing iterations from 1 to 3 (ablation) | CapsNet with reconstruction · MultiMNIST test | classification error (%): $7.5$ -> $\bm{5.2}$ | Helped |
| Increase routing iterations (ablation) | CapsNet · CIFAR10 training | training loss | Helped |
Closest papers by meaning
- Self-Routing Capsule Networks (NeurIPS 2019 · similarity 0.90)
- Matrix capsules with EM routing (ICLR 2018 · similarity 0.89)
- Ladder Capsule Network (ICML 2019 · similarity 0.89)
- Capsules with Inverted Dot-Product Attention Routing (ICLR 2020 · similarity 0.89)
- Introducing Routing Uncertainty in Capsule Networks (NeurIPS 2020 · similarity 0.88)
- OrthCaps: An Orthogonal CapsNet with Sparse Attention Routing and Pruning (CVPR 2024 · similarity 0.85)
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.