Research map / Architectures and efficiency / CNN design
Greedy Layerwise Learning Can Scale To ImageNet
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 2-hidden-layer rather than 1-hidden-layer auxiliary problems | SimCNN layerwise training · ImageNet-1k validation | Top-1 accuracy: 58.1 -> 65.7 | Helped |
| Use 3-hidden-layer rather than 1-hidden-layer auxiliary problems | SimCNN layerwise training · ImageNet-1k validation | Top-5 accuracy: 79.7 -> 88.7 | Helped |
| Use 3-hidden-layer rather than 1-hidden-layer auxiliary problems | SimCNN layerwise training · CIFAR-10 | accuracy: 88.3 -> 91.7 | Helped |
| Replace invertible downsampling with strided convolution (ablation) | SimCNN (k=1) · CIFAR-10 | accuracy: 88.3 -> 87.8 | Hurt |
| Replace invertible downsampling with average pooling (ablation) | SimCNN (k=1) · CIFAR-10 | accuracy: 88.3 -> 87.6 | Hurt |
| Replace invertible downsampling with max pooling (ablation) | SimCNN (k=1) · CIFAR-10 | accuracy: 88.3 -> 88.0 | Hurt |
| Reduce the final auxiliary network width from 2048 to 512 (ablation) | SimCNN (k=3) · ImageNet-1k validation | accuracy | Hurt |
| Prune each newly trained layer from 128 to 64 filters and fine-tune its auxiliary network (ablation) | SimCNN (k=3, J=3) · CIFAR-10 | accuracy: 87.5% -> 87.6% | No clear effect |
Closest papers by meaning
- Efficient Learning of CNNs using Patch Based Features (ICML 2022 · similarity 0.88)
- Learning Across Scales---Multiscale Methods for Convolution Neural Networks (AAAI 2018 · similarity 0.87)
- Gradient Descent Learns One-hidden-layer CNN: Don’t be Afraid of Spurious Local Minima (ICML 2018 · similarity 0.87)
- PICNN: A Pathway towards Interpretable Convolutional Neural Networks (AAAI 2024 · similarity 0.86)
- $\mu$PC: Scaling Predictive Coding to 100+ Layer Networks (NeurIPS 2025 · similarity 0.86)
- Feature Learning Dynamics in Infinite-Depth Neural Networks (NeurIPS 2026 · similarity 0.86)
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.