Research map / Architectures and efficiency / RNNs and normalization
Towards Understanding the Condensation of Neural Networks at Initial Training
Hanxu Zhou, Qixuan Zhou, Tao Luo, Yaoyu Zhang, Zhi-Qin John Xu
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 |
|---|---|---|---|
| use condensed Gaussian init N(0,(1/m_bar)^2) instead of common Glorot/N(0,1/m_bar) | Resnet18-like FC layers · CIFAR10 | test accuracy: 88.07% (best common) vs 88.26% (best condensed) | No clear effect |
| condensed init vs common init, lowest accuracy | Resnet18-like FC layers · CIFAR10 | test accuracy: 86.73% vs 86.88% | No clear effect |
| N(0,1/m_out^4) init vs Glorot_uniform, trial 3 | Resnet18-like FC layers · CIFAR10 | test accuracy: 0.8772 -> 0.8688 | Hurt |
Closest papers by meaning
- Empirical Phase Diagram for Three-layer Neural Networks with Infinite Width (NeurIPS 2022 · similarity 0.91)
- Compelling ReLU Networks to Exhibit Exponentially Many Linear Regions at Initialization and During Training (ICML 2025 · similarity 0.87)
- Better NTK Conditioning: A Free Lunch from (ReLU) Nonlinear Activation in Wide Neural Networks (NeurIPS 2025 · similarity 0.87)
- Neural Characteristic Activation Analysis and Geometric Parameterization for ReLU Networks (NeurIPS 2024 · similarity 0.86)
- Neural Collapse under Gradient Flow on Shallow ReLU Networks for Orthogonally Separable Data (NeurIPS 2025 · similarity 0.86)
- Early Neuron Alignment in Two-layer ReLU Networks with Small Initialization (ICLR 2024 · 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.
