Research map / Architectures and efficiency / RNNs and normalization
Deep Learning without Weight Transport
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 weight mirroring to learn feedback weights without weight transport | ResNet-18 · ImageNet | top-1 test error: 97.4(2)% -> 30.2(7)% | Helped |
| Use reciprocal Kolen-Pollack learning to learn feedback weights | ResNet-18 · ImageNet | top-1 test error: 97.4(2)% -> 29.2(4)% | Helped |
| Use weight mirroring to learn feedback weights without weight transport | ResNet-50 · ImageNet | top-1 test error: 98.9(1)% -> 23.4(5)% | Helped |
| Use reciprocal Kolen-Pollack learning to learn feedback weights | ResNet-50 · ImageNet | top-1 test error: 98.9(1)% -> 23.9(7)% | Helped |
| Apply BatchNorm before ReLU rather than after it | Sign-symmetry · ImageNet, ResNet-18 | top-1 test error: 39.2(4)% -> 37.8(4)% | Helped |
| Apply BatchNorm before ReLU rather than after it | Sign-symmetry · ImageNet, ResNet-50 | top-1 test error: 33.8(3)% -> 32.6(6)% | Helped |
| Apply BatchNorm before ReLU rather than after it | Backprop, feedback alignment, Kolen-Pollack, and weight mirror · ImageNet | top-1 test error | No clear effect |
Closest papers by meaning
- One-Time Soft Alignment Enables Resilient Learning without Weight Transport (NeurIPS 2026 · similarity 0.89)
- Pretraining with Random Noise for Fast and Robust Learning without Weight Transport (NeurIPS 2024 · similarity 0.86)
- Improving equilibrium propagation without weight symmetry through Jacobian homeostasis (ICLR 2024 · similarity 0.84)
- The Effect of Weight Precision on the Neuron Count in Deep ReLU Networks (ICML 2024 · similarity 0.84)
- Improving Deep Learning Speed and Performance Through Synaptic Neural Balance (AAAI 2025 · similarity 0.84)
- Rotational Equilibrium: How Weight Decay Balances Learning Across Neural Networks (ICML 2024 · similarity 0.84)
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.