Research map / Federated learning and privacy / Communication-efficient FL
Leveraging Spatial and Temporal Correlations in Sparsified Mean Estimation
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 Rand-k-Temporal to fill unsampled coordinates with server-stored prior values | Rand-k mean estimation · Power Iteration (Fashion-MNIST, IID) | mean estimation error | Helped |
| Use Rand-k-Spatial(Avg) to adjust aggregation weights using the number of nodes sending each coordinate | Rand-k mean estimation · Power Iteration (Fashion-MNIST, IID) | mean estimation error | Helped |
| Use Rand-k-Temporal to fill unsampled coordinates with server-stored prior values | Rand-k mean estimation · K-Means (Fashion-MNIST, IID) | K-Means objective | Helped |
| Use Rand-k-Spatial(Avg) to adjust aggregation weights using the number of nodes sending each coordinate | Rand-k mean estimation · K-Means (Fashion-MNIST, IID) | K-Means objective | Helped |
| Use Rand-k-Temporal to fill unsampled coordinates with server-stored prior values | Rand-k mean estimation · Logistic Regression (CIFAR-10, non-IID) | training loss | Helped |
| Use Rand-k-Temporal to fill unsampled coordinates with server-stored prior values | Rand-k mean estimation · Logistic Regression (CIFAR-10, non-IID) | mean estimation error | Helped |
| Use Rand-k-Spatial(Avg) for aggregation | Rand-k mean estimation · Logistic Regression (CIFAR-10, non-IID) | mean estimation error | Hurt |
| Use Rand-k-Spatial(Avg) instead of Rand-k when spatial correlation is present (ablation) | Rand-k mean estimation · Synthetic vectors spanning the stated correlation range | mean squared error | Helped |
| Use Rand-k-Spatial(Max), corresponding to the maximum spatial-correlation setting (ablation) | Rand-k mean estimation · Synthetic vectors with R2/R1 = 9 | mean squared error | Helped |
| Use Rand-k-Spatial(Avg) rather than Rand-k-Spatial(Max) across varying spatial correlations (ablation) | Rand-k mean estimation · Synthetic vectors spanning the stated correlation range | mean squared error | Mixed |
| Set the temporal fill-in vectors to zero, reducing Rand-k-Temporal to Rand-k (ablation) | Rand-k-Temporal · Quadratic-objective gradient descent | mean estimation error | Hurt |
Closest papers by meaning
- Correlation Aware Sparsified Mean Estimation Using Random Projection (NeurIPS 2023 · similarity 0.91)
- Optimal Sparsity-Sensitive Bounds for Distributed Mean Estimation (NeurIPS 2019 · similarity 0.86)
- Fundamental Limits of Distributed Covariance Matrix Estimation Under Communication Constraints (ICML 2024 · similarity 0.84)
- Accelerating Federated Learning with Quick Distributed Mean Estimation (ICML 2024 · similarity 0.83)
- Covariances for Free: Exploiting Mean Distributions for Training-free Federated Learning (NeurIPS 2025 · similarity 0.83)
- Learning from Highly Sparse Spatio-temporal 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.