Researchers have developed LACE, a novel AutoML framework that searches for complete executable pipeline programs rather than relying on pre-defined components. This approach uses a large language model, specifically GPT-5.4 Mini, as the variation operator within an evolutionary loop. LACE was evaluated on 68 OpenML classification tasks and demonstrated superior performance compared to auto-sklearn, H2O, and XGBoost, while matching the performance of AutoGluon. A key advantage of LACE is its ability to return editable Python code, allowing practitioners to inspect and modify the generated pipelines directly. AI
IMPACT This framework offers practitioners editable, task-specific code, potentially improving the reusability and transparency of AutoML-generated pipelines.
RANK_REASON Research paper detailing a new AutoML framework. [lever_c_demoted from research: ic=1 ai=1.0]
Read on arXiv cs.NE (Neural & Evolutionary) →
- autogluon.core
- AutoML
- auto-sklearn
- GPT 5.4 Mini
- H2o Ai
- LACE
- OpenML
- scikit-learn
- Sofoklis Kitharidis
- XGBoost
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
