Research map / Label-efficient and robust learning / Recognition and long-tail
CapProNet: Deep Feature Learning via Orthogonal Projections onto Capsule Subspaces
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 orthogonal capsule projection (c=8) | ResNet-110 · CIFAR10 | error rate: 6.41 -> 5.19 | Helped |
| Add orthogonal capsule projection (c=8) | ResNet-110 · CIFAR100 | error rate: 27.22 -> 21.93 | Helped |
| Add orthogonal capsule projection (c=2 or c=8) | ResNet-110 · SVHN | error rate: 2.01 -> 1.79 | Helped |
| Increase capsule dimension from c=4 to c=8 (ablation) | CapProNet+ResNet-110 · CIFAR100 | error rate: 22.45 -> 21.93 | Helped |
| Increase capsule dimension from c=4 to c=8 (ablation) | CapProNet+ResNet-110 · SVHN | error rate: 1.82 -> 1.79 | No clear effect |
| Increase capsule dimension from c=4 to c=8 (ablation) | CapProNet+WideResNet-28 · CIFAR100 | error rate: 19.98 -> 19.83 | Helped |
| Add capsule projection (c=4) | WideResNet-28 · CIFAR10 | error rate: 4.17 -> 3.64 | Helped |
| Add capsule projection (c=8) | Densenet-BC k=12 · CIFAR100 | error rate: 22.27 -> 21.19 | Helped |
| Add capsule projection (c=8) | Resnet50 · ImageNet validation, single crop 224x224 | Top-1 error: 24.09 -> 23.203 | Helped |
| Add capsule projection (c=4) | Resnet101 · ImageNet validation, single crop 224x224 | Top-1 error: 22.81 -> 21.89 | Helped |
| Replace orthogonal capsule projection with grouped output neurons (ablation) | ResNet-110 · CIFAR10 | error rate: 5.24 -> 6.26 | Hurt |
| Increase capsule dimension from c=4 to c=8 (ablation) | CapProNet+WideResNet-28 · CIFAR10 | error rate: 3.64 -> 3.85 | Hurt |
Closest papers by meaning
- SubSpace Capsule Network (AAAI 2020 · similarity 0.91)
- Matrix capsules with EM routing (ICLR 2018 · similarity 0.89)
- DeepCaps: Going Deeper With Capsule Networks (CVPR 2019 · similarity 0.88)
- OrthCaps: An Orthogonal CapsNet with Sparse Attention Routing and Pruning (CVPR 2024 · similarity 0.87)
- Building Deep Equivariant Capsule Networks (ICLR 2020 · similarity 0.87)
- ParseCaps: An Interpretable Parsing Capsule Network for Medical Image Diagnosis (AAAI 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.