Research map / Architectures and efficiency / RNNs and normalization
Batch normalization provably avoids ranks collapse for randomly initialised deep networks
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 |
|---|---|---|---|
| add batch normalization | deep MLPs · Fashion-MNIST | training accuracy and last-layer rank | Helped |
| add rank-maximizing pre-training step (ablation) | vanilla deep MLPs · training convergence | SGD convergence speed | Helped |
| change BN weight init from uniform[-1/sqrt(d),1/sqrt(d)] to uniform[0,2/sqrt(d)] (ablation) | BN MLPs, VGG, ResNet · CIFAR-10 | learning speed | Hurt |
| remove mean centering in BN (ablation) | VGG · CIFAR-10 | training accuracy | No clear effect |
Closest papers by meaning
- Understanding Batch Normalization (NeurIPS 2018 · similarity 0.90)
- Batch Normalization Orthogonalizes Representations in Deep Random Networks (NeurIPS 2021 · similarity 0.90)
- Batch Normalization Biases Residual Blocks Towards the Identity Function in Deep Networks (NeurIPS 2020 · similarity 0.89)
- A Mean Field Theory of Batch Normalization (ICLR 2019 · similarity 0.89)
- Towards Training Without Depth Limits: Batch Normalization Without Gradient Explosion (ICLR 2024 · similarity 0.89)
- Deep Linear Network Training Dynamics from Random Initialization: Data, Width, Depth, and Hyperparameter Transfer (ICML 2025 · similarity 0.88)
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.