Research map / Learning theory and algorithms / Combinatorial and search optimization
Simultaneous Confidence Bounds for Aggregated Effects via Exact Subset Optimization
Weihang Xu, Huajie Qian, Wotao Yin, Xinshang Wang
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 SubsetMaxT bound instead of Bonferroni | Bonferroni · Synthetic dense homogeneous, m=50, k=50 | mean lower confidence bound: -1.62e+00 -> 3.10e-01 | Improved |
| use SubsetMaxT bound instead of Scheffe | Scheffe · Synthetic dense heterogeneous, m=100, k=100 | mean lower confidence bound: -8.08e-01 -> 7.31e-01 | Improved |
| use SubsetMaxT bound instead of Scheffe | Scheffe · Synthetic sparse homogeneous, m=50, k=5 | mean lower confidence bound: 2.18e+00 -> 2.25e+00 | Improved |
| use SubsetMaxT bound instead of Bonferroni | Bonferroni · Hillstrom subgroup A/B, m=32, k=32 | mean lower confidence bound: 3.72e-02 -> 7.88e-02 | Improved |
| use SubsetMaxT bound instead of Scheffe | Scheffe · Criteo subgroup A/B, m=49, k=49 | mean lower confidence bound: 1.36e+00 -> 1.40e+00 | Improved |
| use SubsetMaxT bound instead of Bonferroni | Bonferroni · OpenML-CC18 pipelines, m=50, k=50 | mean lower confidence bound: 1.91e-02 -> 3.79e-02 | Improved |
| impose the cardinality limit k=5 instead of unconstrained SubsetMaxT (ablation) | SubsetMaxT (unconstr.) · Synthetic dense homogeneous, m=50, k=5 | mean lower confidence bound: -4.80e-01 -> -2.15e-01 | Improved |
| impose the cardinality limit k=5 instead of unconstrained SubsetMaxT (FPTAS) (ablation) | SubsetMaxT (unconstr.) · Synthetic dense homogeneous, m=100, k=5 | mean lower confidence bound: -8.68e-01 -> -2.59e-01 | Improved |
| use unconstrained SubsetMaxT instead of Scheffe at large m small k | Scheffe · Synthetic dense homogeneous, m=100, k=5 | mean lower confidence bound: -3.06e-01 -> -8.68e-01 | Hurt |
| use FPTAS (eps=0.01) instead of exact MILP | SubsetMaxT · Synthetic dense homogeneous, m=50, k=5 | mean lower confidence bound: -2.15e-01 -> -2.28e-01 | Hurt |
| use FPTAS (eps=0.01) instead of exact MILP | SubsetMaxT · Synthetic sparse homogeneous, m=50, k=5 | mean lower confidence bound: 2.25e+00 -> 2.23e+00 | No clear effect |
| use SubsetMaxT instead of Bonferroni when k=1 | Bonferroni · ML pipelines simple injection, delta=0.02, k=1 | mean lower confidence bound: -0.005 -> -0.007 | Hurt |
Closest papers by meaning
- Theoretical bounds on estimation error for meta-learning (ICLR 2021 · similarity 0.85)
- Multinoulli Extension: A Lossless Yet Effective Probabilistic Framework for Subset Selection over Partition Constraints (ICML 2025 · similarity 0.84)
- PriorBoost: An Adaptive Algorithm for Learning from Aggregate Responses (ICML 2024 · similarity 0.84)
- Most Influential Subset Selection: Challenges, Promises, and Beyond (NeurIPS 2024 · similarity 0.84)
- Optimal Experiments for Partial Causal Effect Identification (NeurIPS 2026 · similarity 0.83)
- DUAL: Learning Diverse Kernels for Aggregated Two-sample and Independence Testing (NeurIPS 2025 · 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.
