Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2510.18813 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909868244336640 |
|---|---|
| author | Kundu, Soumyabrata Kondor, Risi |
| author_facet | Kundu, Soumyabrata Kondor, Risi |
| contents | In contrast to the somewhat abstract, group theoretical approach adopted by many papers, our work provides a new and more intuitive derivation of steerable convolutional neural networks in $d$ dimensions. This derivation is based on geometric arguments and fundamental principles of pattern matching. We offer an intuitive explanation for the appearance of the Clebsch--Gordan decomposition and spherical harmonic basis functions. Furthermore, we suggest a novel way to construct steerable convolution layers using interpolation kernels that improve upon existing implementation, and offer greater robustness to noisy data. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_18813 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | A Geometric Approach to Steerable Convolutions Kundu, Soumyabrata Kondor, Risi Computer Vision and Pattern Recognition In contrast to the somewhat abstract, group theoretical approach adopted by many papers, our work provides a new and more intuitive derivation of steerable convolutional neural networks in $d$ dimensions. This derivation is based on geometric arguments and fundamental principles of pattern matching. We offer an intuitive explanation for the appearance of the Clebsch--Gordan decomposition and spherical harmonic basis functions. Furthermore, we suggest a novel way to construct steerable convolution layers using interpolation kernels that improve upon existing implementation, and offer greater robustness to noisy data. |
| title | A Geometric Approach to Steerable Convolutions |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2510.18813 |