Research map / Image generation / Autoregressive and latent generation
Nonlocal Neural Networks, Nonlocal Diffusion and Nonlocal Modeling
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 original blocks with proposed nonlocal-stage blocks at the same location, using 2 blocks | PreResNet-20 · CIFAR-10 validation | error (%): 7.83 -> 7.74 | Helped |
| replace original blocks with proposed nonlocal-stage blocks at the same location, using 3 blocks | PreResNet-20 · CIFAR-10 validation | error (%): 8.28 -> 7.62 | Helped |
| replace original blocks with proposed nonlocal-stage blocks at the same location, using 4 blocks | PreResNet-20 · CIFAR-10 validation | error (%): 15.02 -> 7.37 | Helped |
| replace original blocks with proposed blocks, using 3 blocks at different network places | PreResNet-20 · CIFAR-10 validation, different places | error (%): 8.07 -> 7.33 | Helped |
| replace original blocks with proposed nonlocal-stage blocks, using 2 blocks | PreResNet-56 · CIFAR-100 validation | error (%): 26.13 -> 26.04 | Helped |
| replace original blocks with proposed nonlocal-stage blocks, using 3 blocks | PreResNet-56 · CIFAR-100 validation | error (%): 26.26 -> 25.57 | Helped |
| replace original blocks with proposed nonlocal-stage blocks, using 4 blocks | PreResNet-56 · CIFAR-100 validation | error (%): 34.89 -> 25.43 | Helped |
| increase proposed nonlocal blocks from 4 to 5 (ablation) | PreResNet-20 · CIFAR-10 validation | error (%): 7.37 -> 7.29 | Helped |
| increase proposed nonlocal blocks from 5 to 6 (ablation) | PreResNet-20 · CIFAR-10 validation | error (%): 7.29 -> 7.55 | Hurt |
| increase proposed nonlocal blocks from 4 to 5 (ablation) | PreResNet-56 · CIFAR-100 validation | error (%): 25.43 -> 25.29 | Helped |
| increase proposed nonlocal blocks from 5 to 6 (ablation) | PreResNet-56 · CIFAR-100 validation | error (%): 25.29 -> 25.49 | Hurt |
| add 4 original nonlocal blocks instead of 3 (ablation) | PreResNet-20 · CIFAR-10 validation | error (%): 8.28 -> 15.02 | Hurt |
Closest papers by meaning
- Non-Local Neural Networks (CVPR 2018 · similarity 0.83)
- An Analytical Theory of Spectral Bias in the Learning Dynamics of Diffusion Models (NeurIPS 2025 · similarity 0.82)
- Neural Diffusion Models (ICML 2024 · similarity 0.81)
- Locality in Image Diffusion Models Emerges from Data Statistics (NeurIPS 2025 · similarity 0.81)
- Shallow diffusion networks provably learn hidden low-dimensional structure (ICLR 2025 · similarity 0.81)
- Hyperbolic-PDE GNN: Spectral Graph Neural Networks in the Perspective of A System of Hyperbolic Partial Differential Equations (ICML 2025 · similarity 0.81)
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.