Research map / Architectures and efficiency / RNNs and normalization
Path-conditioned training: a principled way to rescale ReLU neural networks
Arthur Lebeurrier, Titouan Vayer, Rémi Gribonval
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 |
|---|---|---|---|
| rescale ReLU network at initialization with PathCond | SGD, Kaiming-initialized MLP · CIFAR-10 MLP | epochs to 99% training accuracy: up to 1.5x faster | Improved |
| rescale at initialization with PathCond | SGD, CIFAR-NV · CIFAR-10 | best test accuracy: 77 -> 80 | Improved |
| replace Equinormalization rescaling with PathCond | SGD MLP · CIFAR-10 MLP | epochs to 99% training accuracy | Improved |
| add PathCond with increasing network depth (more hidden layers) (ablation) | SGD MLP · CIFAR-10 MLP | training speedup | Mixed |
| increase learning rate (ablation) | PathCond · CIFAR-10 | speedup over baseline | Mixed |
| increase compression factor (more unbalanced autoencoder) (ablation) | PathCond · MNIST autoencoder | final training loss | Improved |
| use PathCond on autoencoder at all compression factors | SGD · MNIST autoencoder | final training loss | Improved |
| vary layer width irregularity and init variance (ablation) | PathCond · random ReLU nets | log rescaling magnitude | Mixed |
Closest papers by meaning
- Optimal Sets and Solution Paths of ReLU Networks (ICML 2023 · similarity 0.89)
- Neural Characteristic Activation Analysis and Geometric Parameterization for ReLU Networks (NeurIPS 2024 · similarity 0.89)
- Compelling ReLU Networks to Exhibit Exponentially Many Linear Regions at Initialization and During Training (ICML 2025 · similarity 0.89)
- Simplicity Bias and Optimization Threshold in Two-Layer ReLU Networks (ICML 2025 · similarity 0.87)
- The Symmetries of Three-Layer ReLU Networks (NeurIPS 2026 · similarity 0.87)
- Activation-Descent Regularization for Input Optimization of ReLU Networks (ICML 2024 · similarity 0.87)
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.
