Research map / Optimization / Optimizers
Decoupled Parallel Backpropagation with Convergence Guarantee
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 DDG instead of BP | ResNet-56 · CIFAR-10 test | Top 1 classification accuracy (%): 93.12 -> 93.11 | No clear effect |
| Use DDG instead of BP | ResNet-56 · CIFAR-100 test | Top 1 classification accuracy (%): 69.79 -> 70.17 | Helped |
| Use DDG instead of BP | ResNet-110 · CIFAR-10 test | Top 1 classification accuracy (%): 93.53 -> 93.41 | No clear effect |
| Use DDG instead of BP | ResNet-110 · CIFAR-100 test | Top 1 classification accuracy (%): 71.90 -> 71.39 | No clear effect |
| Distribute ResNet-110 across 4 GPUs using DDG | ResNet-110 · CIFAR-10 | training speedup: about 2 times | Helped |
| Increase the DDG module count from 2 to 4 (ablation) | ResNet-8 · CIFAR-10 | Top 1 classification accuracy | No clear effect |
| Place the single split point deeper, at layers 3, 5, or 7 rather than layer 1 (ablation) | ResNet-8 DDG · CIFAR-10 | Top 1 classification accuracy | No clear effect |
Closest papers by meaning
- Accelerated training through iterative gradient propagation along the residual path (ICLR 2025 · similarity 0.87)
- One-Time Soft Alignment Enables Resilient Learning without Weight Transport (NeurIPS 2026 · similarity 0.84)
- Forward Learning with Top-Down Feedback: Empirical and Analytical Characterization (ICLR 2024 · similarity 0.84)
- Reducing Fine-Tuning Memory Overhead by Approximate and Memory-Sharing Backpropagation (ICML 2024 · similarity 0.84)
- Overcoming Rank Collapse in Feedback Alignment (NeurIPS 2026 · similarity 0.83)
- PETRA: Parallel End-to-end Training with Reversible Architectures (ICLR 2025 · similarity 0.83)
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.