Research map / Probabilistic and scientific modeling / Bayesian optimization
An ADMM Based Framework for AutoML Pipeline Configuration
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 |
|---|---|---|---|
| Decompose joint pipeline configuration into alternating hyperparameter and algorithm-selection subproblems | AutoML pipeline configuration · fri-c2 and 7 other datasets | speedup to reach JOPT best objective: 10 \times; 153\times; 42 \times; 25 \times; 11 \times; 9 \times; 40 \times | Helped |
| Use BO for hyperparameter optimization and CMAB for algorithm selection instead of BO for both subproblems (ablation) | AdADMM · 8 datasets | speedup to reach JOPT best objective: 10 \times vs 2 \times; 153\times vs 25\times; 42 \times vs 5 \times; 25 \times vs 7 \times; 11 \times vs 4 \times; 9 \ti | Helped |
| Use BO for both hyperparameter optimization and algorithm selection rather than CMAB for algorithm selection (ablation) | AdADMM · 8 datasets | final objective improvement over JOPT: 5 \% vs 0 \%; 5 \% vs 0 \%; 64\% vs 56\%; 13\% vs 8 \%; 3 \% vs 4 \%; 1 \% vs 1 \%; 26\% vs 9 \%; 5 \% vs 0 \% | Mixed |
| Use adaptive rather than fixed BO iteration budgets for ADMM subproblems (ablation) | ADMM on fri-c2 · fri-c2 | validation performance across optimization time | Helped |
| Split the joint optimization with ADMM instead of optimizing the complete joint problem with BO (ablation) | AutoML pipeline configuration · fri-c2 and 8-dataset comparison | optimization speed and final objective: over $10\times$; over $10\% | Helped |
| Incorporate black-box constraints during optimization rather than optimize unconstrained and filter afterward | ADMM on Home Credit Default Risk · Home Credit Default Risk | objective among constraint-satisfying pipelines | Helped |
| Incorporate black-box constraints during optimization rather than optimize unconstrained and filter afterward | ADMM on Home Credit Default Risk · Home Credit Default Risk | fraction of pipelines satisfying constraints | Helped |
| Relax latency or disparate-impact constraint thresholds (ablation) | Constrained ADMM · Home Credit Default Risk | objective among constraint-satisfying pipelines | Helped |
Closest papers by meaning
- Searching for Machine Learning Pipelines Using a Context-Free Grammar (AAAI 2021 · similarity 0.89)
- Pipeline Combinators for Gradual AutoML (NeurIPS 2021 · similarity 0.89)
- ADELA: Accelerating Evolutionary Design of Machine Learning Pipelines with the Accompanying Surrogate Model (AAAI 2025 · similarity 0.85)
- AutoMMLab: Automatically Generating Deployable Models from Language Instructions for Computer Vision Tasks (AAAI 2025 · similarity 0.85)
- AutoML-Agent: A Multi-Agent LLM Framework for Full-Pipeline AutoML (ICML 2025 · similarity 0.84)
- BBOWP-Bench: Evaluating LLMs on Black-Box Optimization Word Problems (NeurIPS 2026 · 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.