Research map / Probabilistic and scientific modeling / Differential equations and PDEs
Puppet-CNN: Continuous Parameter Dynamics for Input-Adaptive Convolutional Networks
Yucheng Xing, Xin Wang
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 layer-wise parameters with puppeteer ODE-generated parameters (ablation) | AlexNet · CIFAR-10 (10k train subset) | Top-1 accuracy: 65.86% -> 69.33% | Improved |
| replace layer-wise parameters with puppeteer ODE-generated parameters (ablation) | VGG · CIFAR-10 (10k train subset) | Top-1 accuracy: 67.84% -> 67.78% | No clear effect |
| replace layer-wise parameters with puppeteer ODE-generated parameters (ablation) | ResNet · CIFAR-10 (10k train subset) | Top-1 accuracy: 68.94% -> 66.69% | Hurt |
| add parameter-level adaptation (ablation) | Puppeteer-Puppet AlexNet · CIFAR-10 (10k train subset) | Top-1 accuracy: 69.33% -> 70.05% | Improved |
| add parameter-level adaptation (ablation) | Puppeteer-Puppet VGG · CIFAR-10 (10k train subset) | Top-1 accuracy: 67.78% -> 68.42% | Improved |
| add parameter-level adaptation (ablation) | Puppeteer-Puppet ResNet · CIFAR-10 (10k train subset) | Top-1 accuracy: 66.69% -> 68.25% | Improved |
| add input-dependent depth adaptation (versus fixed-depth Puppet-ResNet) (ablation) | Puppet-ResNet · CIFAR-10 (10k train subset) | Mult-Adds (G): 23.44 -> 12.34 | Improved |
| add depth adaptation (versus fixed-depth Puppet-ResNet) (ablation) | Puppet-ResNet · CIFAR-10 (10k train subset) | Speed (s/img): 0.0061 -> 0.0039 | Improved |
| train on full CIFAR-10 instead of 10k: Puppet-Puppeteer versus fixed AlexNet | AlexNet · CIFAR-10 (full train set) | Top-1 accuracy: 80.89% -> 83.75% | Improved |
| train on full CIFAR-10: Puppet-Puppeteer versus fixed ResNet | ResNet · CIFAR-10 (full train set) | Top-1 accuracy: 82.24% -> 80.47% | Hurt |
| Puppet-CNN versus ResNet backbone on harder dataset | Puppet-CNN · CIFAR-100 | Top-1 accuracy: 44.07% -> 53.26% | Improved |
| Puppet-CNN versus ResNet backbone on harder dataset | Puppet-CNN · mini-ImageNet | Top-1 accuracy: 38.41% -> 46.46% | Improved |
Closest papers by meaning
- Continuity-Preserving Convolutional Autoencoders for Learning Continuous Latent Dynamical Models from Images (ICLR 2025 · similarity 0.87)
- Recurrent neural network dynamical systems for biological vision (NeurIPS 2024 · similarity 0.86)
- Depthwise Hyperparameter Transfer in Residual Networks: Dynamics and Scaling Limit (ICLR 2024 · similarity 0.85)
- KernelWarehouse: Rethinking the Design of Dynamic Convolution (ICML 2024 · similarity 0.85)
- Evolving Parameterized Prompt Memory for Continual Learning (AAAI 2024 · similarity 0.85)
- Feature Learning Dynamics in Infinite-Depth Neural Networks (NeurIPS 2026 · 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.
