Research map / Federated learning and privacy / Communication-efficient FL
COLA: Decentralized Linear Learning
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 CoLa instead of DIGing or decentralized ADMM | Decentralized Lasso on Webspam and ridge regression on URL · Webspam and URL reputation | communication rounds | Helped |
| Use CoLa instead of DIGing or decentralized ADMM | Decentralized Lasso on Webspam and ridge regression on URL · Webspam and URL reputation | computation time | Helped |
| Increase local data passes per communication round (ablation) | CoLa · Experiments varying subproblem quality | communication rounds | Helped |
| Increase local data passes per communication round (ablation) | CoLa · Experiments varying subproblem quality | wall-clock time | Mixed |
| Use a more connected topology with smaller β (ablation) | CoLa · Lasso on RCV1, 16 nodes | convergence rate | Helped |
| Increase the probability that nodes stay in the network (ablation) | CoLa · URL reputation | convergence rate | Helped |
| Allow nodes to drop out while freezing their local variables (ablation) | CoLa · URL reputation, unreliable networks | suboptimality | Helped |
Closest papers by meaning
- $D^2$: Decentralized Training over Decentralized Data (ICML 2018 · similarity 0.86)
- Decentralized Sporadic Federated Learning: A Unified Algorithmic Framework with Convergence Guarantees (ICLR 2025 · similarity 0.84)
- Byzantine-Robust Decentralized Learning via Remove-then-Clip Aggregation (AAAI 2024 · similarity 0.84)
- DePRL: Achieving Linear Convergence Speedup in Personalized Decentralized Learning with Shared Representations (AAAI 2024 · similarity 0.84)
- RESIST: Resilient Decentralized Learning Using Consensus Gradient Descent (NeurIPS 2026 · similarity 0.84)
- Revisiting Differentially Private Algorithms for Decentralized Online Learning (ICML 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.