Research map / Architectures and efficiency / CNN design
On the Relationship between Self-Attention and Convolutional Layers
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 |
|---|---|---|---|
| Use non-isotropic Gaussian positional encoding instead of quadratic positional encoding (ablation) | Six-layer self-attention image classifier · CIFAR-10 | accuracy | No clear effect |
| Prune degenerated non-isotropic attention heads and continue training (ablation) | Six-layer, nine-head self-attention model · CIFAR-10 | accuracy | No clear effect |
| Learn relative positional encodings from data | Self-attention image classifier · Image classification attention patterns | local pixel attention | Helped |
| Increase the number of heads from 9 to 16 (ablation) | Six-layer self-attention image classifier · Image classification attention patterns | localized attention in early layers | Helped |
Closest papers by meaning
- Stand-Alone Self-Attention in Vision Models (NeurIPS 2019 · similarity 0.89)
- Scaling Local Self-Attention for Parameter Efficient Visual Backbones (CVPR 2021 · similarity 0.89)
- Self-Attention-Based Contextual Modulation Improves Neural System Identification (ICLR 2025 · similarity 0.87)
- Skip-Attention: Improving Vision Transformers by Paying Less Attention (ICLR 2024 · similarity 0.86)
- Bridging the Divide: Reconsidering Softmax and Linear Attention (NeurIPS 2024 · similarity 0.85)
- Revisiting the Integration of Convolution and Attention for Vision Backbone (NeurIPS 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.