SubStrat: A Subset-Based Optimization Strategy for Faster AutoML
Summary: SubStrat accelerates black-box AutoML by genetically selecting a small, representative data subset, rather than pruning the pipeline/configuration space. It then refines the discovered pipeline on full data, cutting runtime 76.3% on Auto-Sklearn, TPOT, and H2O for only 4.15% accuracy loss. (summarized by gpt-5.6-luna on Jul 24 2026)
Incoming Non-self Citations Over Time
Authors
- 1. Teddy Lazebnik (University College London)
- 2. Amit Somech (Bar-Ilan University)
- 3. Abraham Itzhak Weinberg (Bar-Ilan University)
BibTeX Citation
@article{lazebnik_vldb23,
title = {{SubStrat: A Subset-Based Optimization Strategy for Faster AutoML}},
author = {Lazebnik, Teddy and Somech, Amit and Weinberg, Abraham Itzhak},
journal = {PVLDB},
series = {{VLDB} '23},
volume = {16},
number = {4},
pages = {772--780},
doi = {10.14778/3574245.3574261},
url = {https://doi.org/10.14778/3574245.3574261},
year = {2023}
}Incoming Citations (Sorted by Pagerank)
Showing 3 of 3 citing papers.
| Rank | Citing Paper | Year | Venue | Pagerank |
|---|---|---|---|---|
| 10,466 | HAMMER: An Automatic RAG Tuning System via Hierarchical Memory-Guided Monte Carlo Tree Search | 2026 | SIGMOD | 4.9793485e-05 |
| 10,722 | CAPS: Cost-Aware ML Pipeline Selection | 2026 | VLDB | 4.9793485e-05 |
| 11,454 | Datamap-Driven Tabular Coreset Selection for Classifier Training | 2025 | VLDB | 4.9793485e-05 |
Previous
Page 1 / 1
Next
Outgoing Citations (Sorted by Pagerank)
Showing 7 of 7 cited papers.
Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.
| Rank | Cited Paper | Year | Venue | Pagerank |
|---|---|---|---|---|
| 1,668 | SystemDS: A Declarative Machine Learning System for the End-to-End Data Science Lifecycle | 2020 | CIDR | 9.9371612e-05 |
| 1,927 | Elastic Machine Learning Algorithms in Amazon SageMaker | 2020 | SIGMOD | 9.3607648e-05 |
| 2,036 | Ease.ml: Towards Multi-tenant Resource Sharing for Machine Learning Workloads | 2018 | VLDB | 9.1520279e-05 |
| 3,067 | Oracle AutoML: A Fast and Predictive AutoML Pipeline | 2020 | VLDB | 7.6876681e-05 |
| 3,331 | VolcanoML: Speeding up End-to-End AutoML via Scalable Search Space Decomposition | 2021 | VLDB | 7.4166095e-05 |
| 6,028 | A Scalable AutoML Approach Based on Graph Neural Networks | 2022 | VLDB | 5.9110992e-05 |
| 8,741 | Assassin: an Automatic claSSificAtion system baSed on algorithm SelectIoN | 2021 | VLDB | 5.2886806e-05 |
Previous
Page 1 / 1
Next
Semantically Similar Papers
| # | Overall Rank | Paper | Year | Venue |
|---|---|---|---|---|
| 1 | 8,807 | ApproxML: Efficient Approximate Ad-Hoc ML Models Through Materialization and Reuse | 2019 | VLDB |
| 2 | 6,156 | Automatic Data Acquisition for Deep Learning | 2021 | VLDB |
| 3 | 11,196 | Subgroup Discovery with Small and Alternative Feature Sets | 2025 | SIGMOD |
| 4 | 10,722 | CAPS: Cost-Aware ML Pipeline Selection | 2026 | VLDB |
| 5 | 5,100 | Towards Benchmarking Feature Type Inference for AutoML Platforms | 2021 | SIGMOD |
| 6 | 7,538 | Automating and Optimizing Data-Centric What-If Analyses on Native Machine Learning Pipelines | 2023 | SIGMOD |
| 7 | 3,331 | VolcanoML: Speeding up End-to-End AutoML via Scalable Search Space Decomposition | 2021 | VLDB |
| 8 | 6,028 | A Scalable AutoML Approach Based on Graph Neural Networks | 2022 | VLDB |
| 9 | 3,067 | Oracle AutoML: A Fast and Predictive AutoML Pipeline | 2020 | VLDB |
| 10 | 5,933 | Doing More with Less: Characterizing Dataset Downsampling for AutoML | 2021 | VLDB |