Research map / Federated learning and privacy / Communication-efficient FL
Sketching for First Order Method: Efficient Algorithm for Low-Bandwidth Channel and Vulnerability
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 |
|---|---|---|---|
| Apply iterative sketching and de-sketching to local gradient updates | Federated learning with strongly-convex smooth objectives · Strongly-convex smooth federated learning | expected suboptimality: \frac{L}{2}\E[\|w^0-w^*\|_2^2]e^{-\mu\eta_{\mathrm{local}} T} + 4\eta_{\mathrm{local}}^2 L^2 K^3\sigma^2/\mu | Helped |
| Communicate sketched local updates instead of full-dimensional updates | IterativeSketchingFL · Strongly-convex smooth federated learning | communication cost: O( ({LN}/{\mu}) \max\{d, \sqrt{{\sigma^2}/{(\mu\epsilon)}}\}\log({L\E[\|w^0 - w^*\|_2^2]}/{\epsilon})) bits of communica | Helped |
| Apply iterative sketching and de-sketching to local gradient updates | Federated learning with convex smooth objectives · Convex smooth federated learning | expected suboptimality: \frac{4\E[\|w^0 - w^*\|_2^2]}{\eta_{\mathrm{local}} K T } + 32\eta_{\mathrm{local}}^2 L K^2 \sigma^2 | Helped |
| Communicate sketched local updates instead of full-dimensional updates | IterativeSketchingFL · Convex smooth federated learning | communication cost: O(\E[\|w^0-w^*\|_2^2]N \max\{{Ld}/{\epsilon}, {\sigma\sqrt{ L}}/{\epsilon^{3/2}}\}) vs O\left(\E[\|w^0-w^*\|_2^2]Nd \cdo | Helped |
| Add Gaussian noise to local stochastic gradients | IterativeSketchingFL · Private federated learning | differential privacy: \epsilon_{\mathrm{DP}} = \sqrt{TK}\cdot \widehat \epsilon, ~~~\delta_{\mathrm{DP}} = TK\cdot \widehat \delta | Helped |
| Use sketching without additional privacy noise | Federated learning gradient communication · Sketched gradient leakage attack | attack feasibility | No clear effect |
Closest papers by meaning
- Sketched Gaussian Mechanism for Private Federated Learning (NeurIPS 2025 · similarity 0.89)
- Sketching for Distributed Deep Learning: A Sharper Analysis (NeurIPS 2024 · similarity 0.88)
- Sketched Adaptive Distributed Deep Learning: A Sharp Convergence Analysis (NeurIPS 2025 · similarity 0.87)
- FedNS: A Fast Sketching Newton-Type Algorithm for Federated Learning (AAAI 2024 · similarity 0.87)
- z-SignFedAvg: A Unified Stochastic Sign-Based Compression for Federated Learning (AAAI 2024 · similarity 0.85)
- Noisy SIGNSGD Is More Differentially Private Than You (Might) Think (ICML 2025 · similarity 0.85)
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.