Research map / Optimization / Distributed and compressed
Communication-Efficient Distributed PCA by Riemannian 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 sign-fixed averaging (option II) or geodesic averaging (option III) to update the global variable instead of random selection (option I) (ablation) | CEDRE · a9a | convergence speed (log error vs communication rounds) | Helped |
| Sign-fixed averaging vs geodesic averaging for global update (ablation) | CEDRE · a9a | convergence speed | No clear effect |
| Increase local sample size n (with m=5n) in the synthetic setting (ablation) | CEDRE · synthetic, delta=0.2 | # communication rounds to converge: n=100 needs about 7 rounds; n=200 about 3; n>=400 about 1 | Helped |
| Decrease eigengap delta (harder problem) (ablation) | CEDRE · synthetic, n=2000 | # communication rounds to reach -25 log error: 4 rounds at delta=0.2 vs 14 rounds at delta=0.01 | Hurt |
| Compare CEDRE against distributed RGD, DSI, ILEA, DisPCA on real datasets | CEDRE · a9a, CIFAR-10, rcv1 | communication vectors to reach high accuracy: RGD needs 66%/75% more on a9a/CIFAR-10; ILEA needs 175% more on rcv1 | Helped |
| Use CEDRE over DisPCA-L at very low communication budget | CEDRE · real-world datasets | accuracy at very few communicated vectors | Hurt |
Closest papers by meaning
- Distributed Principal Component Analysis with Limited Communication (NeurIPS 2021 · similarity 0.92)
- Communication-efficient Algorithms for Distributed Stochastic Principal Component Analysis (ICML 2017 · similarity 0.88)
- An Online Riemannian PCA for Stochastic Canonical Correlation Analysis (NeurIPS 2021 · similarity 0.86)
- Decentralized Projected Riemannian Stochastic Recursive Momentum Method for Nonconvex Optimization (AAAI 2025 · similarity 0.85)
- Riemannian Diffusion Adaptation for Distributed Optimization on Manifolds (ICML 2025 · similarity 0.84)
- Nonconvex Federated Learning on Compact Smooth Submanifolds With Heterogeneous Data (NeurIPS 2024 · 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.