Research map / Optimization / Distributed and compressed
Decentralized Stochastic Optimization and Gossip Algorithms with Compressed Communication
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 error-feedback compressed gossip with stored estimates | Choco-Gossip · Average consensus with arbitrary compression quality | linear convergence rate: (1 - \dfrac{\delta^2 \omega}{82})^t | Helped |
| Use qsgd_256 compressed communication (ablation) | Choco-Gossip · Average consensus, ring topology | consensus error | No clear effect |
| Use rand_1% sparsification (ablation) | Choco-Gossip · Average consensus, ring topology | iterations to converge: 100× slower | Hurt |
| Use rand_1% sparsification (ablation) | Choco-Gossip · Average consensus, ring topology | communication cost to converge | Helped |
| Use rand_1% sparsification (ablation) | Choco-SGD · epsilon and rcv1, sorted data, ring, n=9 | communication cost: 100× less communication | Helped |
| Use qsgd_4 quantization (ablation) | Choco-SGD · epsilon and rcv1, sorted data, ring, n=9 | communication cost: approximately 15× less communication | Helped |
| Replace rand_1% sparsification with top_1% sparsification (ablation) | Choco-SGD · epsilon and rcv1, sorted data, ring, n=9 | suboptimality | Helped |
| Use Choco-SGD instead of DCD-SGD | Decentralized SGD · All reported decentralized SGD settings | suboptimality | Helped |
| Increase the quantization levels s (ablation) | DCD-SGD · Decentralized SGD, qsgd_s | suboptimality | Helped |
| Use ECD-SGD instead of DCD-SGD | Decentralized SGD · Reported decentralized SGD settings | suboptimality | Hurt |
| Use exact averaging with a fully connected graph (ablation) | Plain decentralized SGD · Decentralized SGD, fully connected topology | suboptimality | Helped |
Closest papers by meaning
- Towards Faster Decentralized Stochastic Optimization with Communication Compression (ICLR 2025 · similarity 0.93)
- Optimal Algorithms for Smooth and Strongly Convex Distributed Optimization in Networks (ICML 2017 · similarity 0.91)
- Scalable Distributed Stochastic Optimization via Bidirectional Compression: Beyond Pessimistic Limits (NeurIPS 2026 · similarity 0.90)
- Revisiting Decentralized Online Convex Optimization with Compressed Communication (NeurIPS 2026 · similarity 0.89)
- Distributed Online Convex Optimization with Compressed Communication: Optimal Regret and Applications (NeurIPS 2026 · similarity 0.89)
- Accelerated Methods with Compressed Communications for Distributed Optimization Problems Under Data Similarity (AAAI 2025 · 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.