Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Capsule Routing via Variational Bayes
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 VB routing instead of EM routing with the same network | Capsule network · SVHN | error rate: 5.17% -> 3.9% | Helped |
| Use VB routing instead of EM routing with the same network | Capsule network · CIFAR-10 | error rate: 12.26% -> 11.2% | Helped |
| Use VB routing instead of EM routing | Capsule network · smallNORB novel azimuth | error rate: 12.67 -> 11.33 | Helped |
| Use VB routing instead of EM routing | Capsule network · smallNORB novel elevation | error rate: 12.04 -> 11.59 | Helped |
| Use VB routing instead of EM routing | Capsule network · smallNORB | error rate: 1.8% -> 1.6% | Helped |
| Use VB routing instead of EM routing | Capsule network · Fashion-MNIST | error rate: 5.36% -> 5.2% | Hurt |
| Replace the convolutional stem with 4 residual blocks (8 layers) (ablation) | VB-Routing · CIFAR-10 | error rate: 11.2% -> 7.8% | Helped |
| Learn the Dirichlet alpha prior from data (ablation) | VB-Routing · Fashion-MNIST | error rate: 5.2% -> 5.19% | No clear effect |
| Vary the Wishart and Dirichlet prior initialisation coefficients (ablation) | VB-Routing · Fashion-MNIST | error rate | No clear effect |
| Reduce the number of routing iterations (ablation) | VB-Routing · Image classification benchmarks | test error | No clear effect |
| Use multiple test-time crops (ablation) | smallNORB capsule network · smallNORB test | error rate: 1.4% -> 1.29% | Helped |
Closest papers by meaning
- Introducing Routing Uncertainty in Capsule Networks (NeurIPS 2020 · similarity 0.90)
- Matrix capsules with EM routing (ICLR 2018 · similarity 0.89)
- Building Deep Equivariant Capsule Networks (ICLR 2020 · similarity 0.88)
- Ladder Capsule Network (ICML 2019 · similarity 0.88)
- OrthCaps: An Orthogonal CapsNet with Sparse Attention Routing and Pruning (CVPR 2024 · similarity 0.85)
- VIKING: Deep variational inference with stochastic projections (NeurIPS 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.