Research map / Probabilistic and scientific modeling / Molecules and proteins
DMCO: Budget-Aware Co-Optimization of Data Cleaning and AutoML
Xiaoou Ding, Zekai Qian, Siying Chen, Hongbin Hu, Chen Wang, Hongzhi Wang, Jianmin 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 |
|---|---|---|---|
| replace gradient-based cleaning sampling with random sampling (add gradient sampling) (ablation) | DMCO · six datasets (Fig. 4) | F1 / MSE | Improved |
| replace loss-based AutoML sampling with random sampling (add loss sampling) (ablation) | DMCO · six datasets (Fig. 4) | F1 / MSE | Improved |
| replace two-stage pipeline with time-slice MAB allocation (ablation) | DMCO · six datasets (Fig. 4) | F1 / MSE | Improved |
| add data cleaning on top of dirty-data AutoML via DMCO | DA · Smartfactory (slice 10) | F1: 0.7427 -> 0.9488 | Improved |
| add AutoML on top of rule-based cleaning via DMCO | CR · Nasa (slice 10) | MSE: 13.3376 -> 7.6229 | Improved |
| use DMCO instead of DA on Adult | DA · Adult (slice 10) | F1: 0.6248 -> 0.6049 | Hurt |
| use DMCO instead of full cleaning plus AutoML | CA · Nasa (slice 10) | MSE: 1.8630 -> 7.6229 | Hurt |
| swap Auto-sklearn for Auto-Gluon inside DMCO (ablation) | DMCO · Nasa (slice 5) | MSE: 7.6229 -> 3.9906 | Improved |
| swap Auto-sklearn for FLAML inside DMCO (ablation) | DMCO · Nasa (slice 5) | MSE: 7.6229 -> 15.5613 | Hurt |
| swap Auto-sklearn for Auto-Gluon inside DMCO (ablation) | DMCO · Skin (slice 5) | F1: 0.9495 -> 0.9989 | Improved |
| swap Auto-sklearn for FLAML inside DMCO (ablation) | DMCO · Soilmoisture (slice 5) | MSE: 0.1038 -> 0.1687 | Hurt |
| use DMCO under limited budget versus full cleaning and exhaustive AutoML | full optimization · six real datasets | performance: 82.19% | Mixed |
Closest papers by meaning
- Clean First, Align Later: Benchmarking Preference Data Cleaning for Reliable LLM Alignment (NeurIPS 2025 · similarity 0.84)
- Adaptive Data Optimization: Dynamic Sample Selection with Scaling Laws (ICLR 2025 · similarity 0.83)
- Domain2Vec: Vectorizing Datasets to Find the Optimal Data Mixture without Training (ICML 2025 · similarity 0.83)
- AutoMMLab: Automatically Generating Deployable Models from Language Instructions for Computer Vision Tasks (AAAI 2025 · similarity 0.83)
- Datasets, Documents, and Repetitions: The Practicalities of Unequal Data Quality (NeurIPS 2025 · similarity 0.83)
- Data Mixture Optimization: A Multi-fidelity Multi-scale Bayesian Framework (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.
