Research map / Architectures and efficiency / CNN design
Co-Occurrence Neural Network
Check 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 third-stack conv2 layers with CoL | ResNet · CIFAR-10 | test error (%): 8.11 -> 8.13 | No clear effect |
| Replace third-stack conv2 layers with CoL rather than magnitude-based pruning | ResNet · CIFAR-10 | test error (%): 9.57 -> 8.13 | Helped |
| Replace third-stack conv2 layers with CoL rather than SVD-based pruning | ResNet · CIFAR-10 | test error (%): 8.75 -> 8.13 | Helped |
| Reduce first convolution support from 7×7 to 3×3 (ablation) | FCN · ADE20k | Accuracy: 62.69 -> 62.02 | Hurt |
| Add CoL after the reduced 3×3 first convolution | FCN · ADE20k | Accuracy: 62.02 -> 64.41 | Helped |
| Remove Conv2 (ablation) | FCN · ADE20k | Accuracy: 64.41 -> 62.09 | Hurt |
| Replace Conv2 with CoL (ablation) | FCN · ADE20k | Accuracy: 62.09 -> 63.27 | Helped |
| Remove the spatial filter w[3×3×3] from CoL (ablation) | ResNet with CoL in third stack · CIFAR-10 | test error (%): 8.13 -> 8.53 | Hurt |
| Remove the co-occurrence term L[5×5] from CoL (ablation) | ResNet with CoL in third stack · CIFAR-10 | test error (%): 8.13 -> 8.53 | Hurt |
Closest papers by meaning
- Deep Co-Occurrence Feature Learning for Visual Object Recognition (CVPR 2017 · similarity 0.91)
- CNN Filter DB: An Empirical Investigation of Trained Convolutional Filters (CVPR 2022 · similarity 0.87)
- The Master Key Filters Hypothesis: Deep Filters Are General (AAAI 2025 · similarity 0.85)
- PICNN: A Pathway towards Interpretable Convolutional Neural Networks (AAAI 2024 · similarity 0.85)
- B-cosification: Transforming Deep Neural Networks to be Inherently Interpretable (NeurIPS 2024 · similarity 0.84)
- Unveiling the Unseen: Identifiable Clusters in Trained Depthwise Convolutional Kernels (ICLR 2024 · similarity 0.84)
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.