Research map / Learning theory and algorithms / Prediction and decision losses
Simultaneous Inference for Massive Data: Distributed Bootstrap
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 n+k-1-grad instead of k-grad (ablation) | Distributed bootstrap · Simultaneous confidence intervals, linear and logistic regression | average width relative to oracle width | Helped |
| Use k-grad with only k worker-level multipliers (ablation) | Distributed bootstrap · Simultaneous confidence intervals when k is small | coverage | Hurt |
| Use n+k-1-grad, assigning separate multipliers to master-node data (ablation) | Distributed bootstrap · Simultaneous confidence intervals when k is small | coverage | Helped |
| Increase the number of communication rounds (ablation) | k-grad and n+k-1-grad · Simultaneous confidence intervals, including k≥n | coverage | Helped |
| Increase the number of machines, reducing local sample size (ablation) | k-grad and n+k-1-grad · Simultaneous confidence intervals | coverage | Hurt |
| Increase model dimension (ablation) | k-grad and n+k-1-grad · Simultaneous confidence intervals | coverage | Hurt |
| Use n+k-1-grad rather than BLB (ablation) | Distributed bootstrap · Logistic regression, large k | average width relative to oracle width | Helped |
| Increase the number of machines (ablation) | k-grad and n+k-1-grad · Linear regression, Toeplitz design, d=2^3; runtime | runtime (seconds): 0.29 -> 0.30 | No clear effect |
| Increase the number of machines (ablation) | BLB · Linear regression, Toeplitz design, d=2^3; runtime | runtime (seconds): 22.66 -> 159.88 | Hurt |
| Increase the number of machines (ablation) | SDB · Linear regression, Toeplitz design, d=2^3; runtime | runtime (seconds): 0.08 -> 5.39 | Hurt |
| Increase the number of machines (ablation) | BLB · Linear regression, Toeplitz design, d=2^7; runtime | runtime (seconds): 981.17 -> 1950.91 | Mixed |
| Increase the number of machines (ablation) | SDB · Linear regression, Toeplitz design, d=2^7; runtime | runtime (seconds): 3.44 -> 12.66 | Hurt |
Closest papers by meaning
- Bootstrap Model Aggregation for Distributed Statistical Learning (NeurIPS 2016 · similarity 0.88)
- Bootstrap in High Dimension with Low Computation (ICML 2023 · similarity 0.87)
- Spend Wisely: Maximizing Post-Training Gains in Iterative Synthetic Data Bootstrapping (NeurIPS 2025 · similarity 0.84)
- Turbocharging Gaussian Process Inference with Approximate Sketch-and-Project (NeurIPS 2025 · similarity 0.82)
- Dimension-free Score Matching and Time Bootstrapping for Diffusion Models (NeurIPS 2025 · similarity 0.82)
- Orthogonal Bootstrap: Efficient Simulation of Input Uncertainty (ICML 2024 · similarity 0.81)
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.