Research map / Architectures and efficiency / CNN design
Non-Linear Convolution Filters for CNN-Based Learning
Georgios Zoumpourlis, Alexandros Doumanoglou, Nicholas Vretos, Petros Daras
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 first-layer linear filters with second-order Volterra non-linear filters | Wide ResNet-28 · CIFAR-10 | test error (%): 3.62 -> 3.51 | Improved |
| replace first-layer linear filters with second-order Volterra non-linear filters | Wide ResNet-28 · CIFAR-100 | test error (%): 18.29 -> 18.24 | No clear effect |
| replace first-layer linear filters with Volterra filters (parameter cost) | Wide ResNet-28 · CIFAR-100 | parameters: 36.6M -> 36.7M | Mixed |
Closest papers by meaning
- Learning Versatile Convolution Filters for Efficient Visual Recognition (TPAMI 2021 · similarity 0.89)
- A Tour of Convolutional Networks Guided by Linear Interpreters (ICCV 2019 · similarity 0.89)
- Involution: Inverting the Inherence of Convolution for Visual Recognition (CVPR 2021 · similarity 0.89)
- CNN Filter DB: An Empirical Investigation of Trained Convolutional Filters (CVPR 2022 · similarity 0.88)
- Convolution Goes Higher-Order: A Biologically Inspired Mechanism Empowers Image Classification (NeurIPS 2025 · similarity 0.88)
- Unveiling the Unseen: Identifiable Clusters in Trained Depthwise Convolutional Kernels (ICLR 2024 · similarity 0.88)
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.
