Research map / Architectures and efficiency / CNN design
PDO-eConvs: Partial Differential Operator Based Equivariant Convolutions
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 conventional CNN convolutions with PDO-eConvs equivariant to p8 | CNN · MNIST-rot-12k, without augmentation | test error (%): 5.03 -> 1.87 | Helped |
| Use PDO-eConvs instead of p4 G-CNN convolutions | G-CNN · MNIST-rot-12k, without augmentation | test error (%): 2.28 -> 1.87 | Helped |
| Use PDO-eConvs rather than SFCNN | SFCNN comparison · MNIST-rot-12k, with augmentation | test error (%): 0.714 -> 0.709 | Helped |
| Use PDO-eConvs rather than HexaConv with p6m | ResNet-26 · CIFAR-10 | test error (%): 8.64 -> 5.38 | Helped |
| Use PDO-eConvs equivariant to p8 rather than G-CNN equivariant to p4m | ResNet-44 · CIFAR-10 | test error (%): 4.94 -> 3.68 | Helped |
| Use PDO-eConvs equivariant to p8 rather than G-CNN equivariant to p4m | ResNet-44 · CIFAR-100 | test error (%): 23.19 -> 20.01 | Helped |
| Use PDO-eConvs rather than G-CNN | ResNet-26 · CIFAR-10 | test error (%): 4.17 -> 3.50 | Helped |
| Use PDO-eConvs rather than Wide ResNet | ResNet-26 · CIFAR-10 | test error (%): 4.00 -> 3.50 | Helped |
| Use PDO-eConvs rather than Wide ResNet | ResNet-26 · CIFAR-100 | test error (%): 19.25 -> 18.40 | Helped |
Closest papers by meaning
- Steerable Partial Differential Operators for Equivariant Neural Networks (ICLR 2022 · similarity 0.91)
- Affine Equivariant Networks Based on Differential Invariants (CVPR 2024 · similarity 0.89)
- Variational Partial Group Convolutions for Input-Aware Partial Equivariance of Rotations and Color-Shifts (ICML 2024 · similarity 0.88)
- Equivariant Neural Tangent Kernels (ICML 2025 · similarity 0.87)
- Approximately Equivariant Neural Processes (NeurIPS 2024 · similarity 0.86)
- A Probabilistic Approach to Learning the Degree of Equivariance in Steerable CNNs (ICML 2024 · similarity 0.86)
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.