Research map / Optimization / Distributed and compressed
DSGD-CECA: Decentralized SGD with Communication-Optimal Exact Consensus Algorithm
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 |
|---|---|---|---|
| Replace dynamic exp-2 / other topologies with CECA exact-consensus message passing in DSGD (DSGD-CECA-2P) | DSGD for CNN · MNIST | test accuracy (%): 98.50 vs 98.36 best baseline (OD-EquiDyn) | Helped |
| Use CECA exact-consensus message passing in DSGD (DSGD-CECA-2P) | DSGD for CNN · MNIST | train loss: 2.083 vs 2.079 centralized SGD; vs 2.090 ring | Mixed |
| Use 1-port CECA (CECA-1P) for finite-time exact consensus on undirected graphs | average consensus · finite-time consensus, n=130 and n=1026 | consensus achieved in ceil(log2 n) iterations | Helped |
| Use DSGD-CECA on least-squares problem vs other stochastic decentralized algorithms | DSGD least squares · distributed least squares, n=258 | convergence performance | Helped |
| Restrict to power-of-2 agent count (dynamic exp-2) vs CECA for arbitrary n | DSGD topology choice · theory (transient iterations) | transient iteration complexity | No clear effect |
Closest papers by meaning
- Topology-aware Generalization of Decentralized SGD (ICML 2022 · similarity 0.90)
- Communication-Efficient Topologies for Decentralized Learning with $O(1)$ Consensus Rate (NeurIPS 2022 · similarity 0.89)
- Towards Stability and Generalization Bounds in Decentralized Minibatch Stochastic Gradient Descent (AAAI 2024 · similarity 0.89)
- Stability and Generalization of Zeroth-Order Decentralized Stochastic Gradient Descent with Changing Topology (AAAI 2025 · similarity 0.87)
- Improved Stability and Generalization Guarantees of the Decentralized SGD Algorithm (ICML 2024 · similarity 0.87)
- Jointly Improving the Sample and Communication Complexities in Decentralized Stochastic Minimax Optimization (AAAI 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.