Research map / Architectures and efficiency / CNN design
Polar Transformer Networks
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 polar origin augmentation (ablation) | PTN-B · Rotated MNIST | error [%]: 1.33 (0.12) -> 1.12 (0.03) | Helped |
| Add training-time rotation augmentation (ablation) | PTN-B · Rotated MNIST | error [%]: 1.46 (0.11) -> 1.12 (0.03) | Helped |
| Add wrap-around padding (ablation) | PTN-B · Rotated MNIST | error [%]: 1.31 (0.06) -> 1.12 (0.03) | Helped |
| Use PTN-ResNet32 instead of ResNet32 | SVHN · SVHN | error: 2.25 (0.15) -> 2.82 (0.07) | Hurt |
| Use PTN-ResNet32 instead of ResNet32 | Rotated SVHN · ROTSVHN | error: 9.83 (0.29) -> 7.90 (0.14) | Helped |
| Use PTN-ResNet32 instead of ResNet32 | SVHN- · SVHN- | error: 2.09 (0.06) -> 2.85 (0.07) | Hurt |
| Use PTN-ResNet32 instead of ResNet32 | ROTSVHN- · ROTSVHN- | error: 5.39 (0.09) -> 3.96 (0.04) | Helped |
| Combine PTN-B with CCNN and training/test rotation augmentation | Rotated MNIST · Rotated MNIST | error [%]: 1.14 (0.08) -> 0.89 (0.06) | Helped |
Closest papers by meaning
- Empowering Networks With Scale and Rotation Equivariance Using A Similarity Convolution (ICLR 2023 · similarity 0.90)
- Harmonic Networks: Deep Translation and Rotation Equivariance (CVPR 2017 · similarity 0.89)
- A Probabilistic Approach to Learning the Degree of Equivariance in Steerable CNNs (ICML 2024 · similarity 0.86)
- On Making $SE(2)$-Invariant Networks Optimal (NeurIPS 2026 · similarity 0.86)
- A Characterization Theorem for Equivariant Networks with Point-wise Activations (ICLR 2024 · similarity 0.86)
- Affine Equivariant Networks Based on Differential Invariants (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.