Research map / Architectures and efficiency / RNNs and normalization
Batch Renormalization: Towards Reducing Minibatch Dependence in Batch-Normalized Models
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 BatchNorm with Batch Renorm | Inception-v3 · ImageNet validation, minibatch size 32 | top-1 accuracy: 78.3% -> 78.5% | Helped |
| Replace BatchNorm with Batch Renorm | Inception-v3 · ImageNet validation, microbatch size 4 | accuracy: 74.2% -> 76.5% | Helped |
| Replace BatchNorm with Batch Renorm | Inception-v3 · ImageNet test, non-i.i.d. minibatches | accuracy: 67% -> 78.5% | Helped |
| Use minibatch statistics during inference rather than moving averages | BatchNorm · ImageNet test, non-i.i.d. minibatches | accuracy: 67% -> 76.5% | Helped |
| Split each minibatch into halves and separate same-label image pairs | BatchNorm · ImageNet test, non-i.i.d. minibatches | accuracy: 67% -> 77.4% | Helped |
| Normalize groups of 4 examples instead of 32 | BatchNorm · ImageNet validation | accuracy: 78.3% -> 74.2% | Hurt |
| Train with non-i.i.d. minibatches containing 2 images per sampled label | BatchNorm · ImageNet test | accuracy: 78.3% -> 67% | Hurt |
Closest papers by meaning
- Four Things Everyone Should Know to Improve Batch Normalization (ICLR 2020 · similarity 0.90)
- Understanding Batch Normalization (NeurIPS 2018 · similarity 0.89)
- Small Batch Size Training for Language Models: When Vanilla SGD Works, and Why Gradient Accumulation is Wasteful (NeurIPS 2025 · similarity 0.87)
- Towards Training Without Depth Limits: Batch Normalization Without Gradient Explosion (ICLR 2024 · similarity 0.87)
- Reducing Divergence in Batch Normalization for Domain Adaptation (AAAI 2025 · similarity 0.86)
- Batch Normalization Is Blind to the First and Second Derivatives of the Loss (AAAI 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.