Research map / Architectures and efficiency / CNN design
SubSpace Capsule Network
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 |
|---|---|---|---|
| Replace the GAN discriminator layers with SCN layers | Semi-supervised CIFAR10 classification · CIFAR10, 4000 labeled samples | classification error rate: 14.32±0.21 | Helped |
| Replace the GAN discriminator layers with SCN layers | Semi-supervised SVHN classification · SVHN, 1000 labeled samples | classification error rate: 8.11 -> 4.58 | Helped |
| Replace the final ResNet-34 block layers with SCN layers | ImageNet classification · ImageNet | Top-1 error rate: 27.13 -> 25.64 | Helped |
| Replace the final ResNet-110 bottleneck block with SCN layers | Supervised CIFAR10 classification · CIFAR10 | classification error rate: 6.41% -> 5.15% | Helped |
| Use SCN convolution layers in the generator | CelebA image generation · CelebA 128×128 | FID: 9.67^* -> 6.23 | Helped |
| Use SCN convolution layers in the generator | LSUN bedroom image generation · LSUN bedroom 128×128 | FID: 21.1^* -> 9.94 | Helped |
| Use SCN convolution layers in the generator | LSUN horse image generation · LSUN horse 256×256 | FID: 16.11 -> 12.83 | Helped |
| Use SCN convolution layers in the generator | LSUN cat image generation · LSUN cat 256×256 | FID: 37.52 -> 29.20 | Helped |
| Add an SCN fully connected layer with 10 capsule types of dimension 2 (ablation) | SVHN semi-supervised classifier · SVHN, 1000 labeled samples | classification error rate: 8.11 -> 5.8 | Helped |
| Add an SCN fully connected layer with 10 capsule types of dimension 4 (ablation) | SVHN semi-supervised classifier · SVHN, 1000 labeled samples | classification error rate: 8.11 -> 5.12 | Helped |
| Add an SCN fully connected layer with 10 capsule types of dimension 8 (ablation) | SVHN semi-supervised classifier · SVHN, 1000 labeled samples | classification error rate: 8.11 -> 5.2 | Helped |
| Add SCN convolution with 64 capsule types of dimension 2 (ablation) | SVHN semi-supervised classifier · SVHN, 1000 labeled samples | classification error rate: 8.11 -> 5.26 | Helped |
Closest papers by meaning
- CapProNet: Deep Feature Learning via Orthogonal Projections onto Capsule Subspaces (NeurIPS 2018 · similarity 0.91)
- Matrix capsules with EM routing (ICLR 2018 · similarity 0.90)
- ParseCaps: An Interpretable Parsing Capsule Network for Medical Image Diagnosis (AAAI 2025 · similarity 0.83)
- OrthCaps: An Orthogonal CapsNet with Sparse Attention Routing and Pruning (CVPR 2024 · similarity 0.83)
- Exploring the Precise Dynamics of Single-Layer GAN Models: Leveraging Multi-Feature Discriminators for High-Dimensional Subspace Learning (NeurIPS 2024 · similarity 0.83)
- From Layers to States: A State Space Model Perspective to Deep Neural Network Layer Dynamics (ICLR 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.