Research map / Optimization / Distributed and compressed
Double Quantization for Communication-Efficient Distributed Optimization
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 double quantization instead of full precision | AsyLPG on MNIST · MNIST, training loss below 0.05 | transmitted bits: 2.42$e$9 -> 3.33$e$8 | Helped |
| Add gradient sparsification to double quantization | AsyLPG on MNIST · MNIST, training loss below 0.05 | transmitted bits: 3.33$e$8 -> 2.73$e$8 | Helped |
| Add momentum acceleration to double quantization | AsyLPG on MNIST · MNIST, training loss below 0.05 | transmitted bits: 3.33$e$8 -> 1.26$e$8 | Helped |
| Increase μ and reduce b_x across the tested settings (ablation) | AsyLPG on MNIST · MNIST, training loss below 0.5 | transmitted bits: 2.42$e$7; 2.33$e$7; 1.52$e$7; 1.07$e$7; 1.12$e$7; 1.24$e$7; 1.14$e$7; 1.08$e$7; 1.44$e$7; 2.16$e$8 | Mixed |
Closest papers by meaning
- A Linear Speedup Analysis of Distributed Deep Learning with Sparse and Quantized Communication (NeurIPS 2018 · similarity 0.91)
- Communication-Efficient Distributed Learning via Lazily Aggregated Quantized Gradients (NeurIPS 2019 · similarity 0.90)
- Theoretically Better and Numerically Faster Distributed Optimization with Smoothness-Aware Quantization Techniques (NeurIPS 2022 · similarity 0.89)
- DoubleSqueeze: Parallel Stochastic Gradient Descent with Double-pass Error-Compensated Compression (ICML 2019 · similarity 0.89)
- JointSQ: Joint Sparsification-Quantization for Distributed Learning (CVPR 2024 · similarity 0.89)
- Error Compensated Quantized SGD and its Applications to Large-scale Distributed Optimization (ICML 2018 · similarity 0.89)
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.