Research map / Learning theory and algorithms / Prediction and decision losses
Distributed Algorithms for U-statistics-based Empirical Risk Minimization
Lanjue Chen, Alan T.K. Wan, Shuyi Zhang, Yong Zhou
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 SU-ERM/OS-ERM instead of naive averaging | distributed U-statistic ERM · Pairwise ranking, n=50, K 25 to 200 | RMSE | Improved |
| use SU-ERM instead of naive averaging for confidence intervals | SU-ERM · Pairwise ranking theta*_1, K=200 | coverage probability: 0.115 -> 0.970 | Improved |
| use OS-ERM instead of naive averaging for confidence intervals | OS-ERM · Pairwise ranking theta*_1, K=25 | coverage probability: 0.755 -> 0.970 | Improved |
| use SU-ERM instead of OS-ERM | SU-ERM · Pairwise ranking | RMSE | No clear effect |
| add iterations of DiaSER (ablation) | DiaSER · Pairwise ranking | RMSE: errors decrease rapidly within first two iterations then steady | Improved |
| increase number of machines K at fixed n=50 | naive estimator · Pairwise ranking | RMSE | No clear effect |
| increase K at fixed n | SU-ERM and OS-ERM · Pairwise ranking | computational time | No clear effect |
| use SU-ERM instead of OS-ERM | SU-ERM · Pairwise ranking and AFT | computational time | Hurt |
| use SU-ERM/OS-ERM instead of naive under AFT model | distributed U-ERM · AFT model, n=50 | coverage probability | Improved |
| decrease local sample size n at fixed N=6000 for naive | naive estimator · AFT model, N=6000 | coverage probability: naive CPs below 90% when n<60 | Hurt |
| decrease local sample size n at fixed N | SU-ERM and OS-ERM · AFT model, N=6000, K=250 n=24 | coverage probability: fall below 90% only at (K,n)=(250,24) | Hurt |
Closest papers by meaning
- Scaling-up Empirical Risk Minimization: Optimization of Incomplete $U$-statistics (JMLR 2016 · similarity 0.91)
- Tilted Empirical Risk Minimization (ICLR 2021 · similarity 0.88)
- On the Asymptotic Distribution of the Minimum Empirical Risk (ICML 2024 · similarity 0.87)
- The Common-directions Method for Regularized Empirical Risk Minimization (JMLR 2019 · similarity 0.87)
- On Medians of (Randomized) Pairwise Means (ICML 2019 · similarity 0.87)
- First-Order Adaptive Sample Size Methods to Reduce Complexity of Empirical Risk Minimization (NeurIPS 2017 · similarity 0.86)
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.
