Research map / Large language models / Alignment and preferences
Human-LLM Collaborative Feature Engineering for Tabular Data
Zhuoyan Li, Aditya Bansal, Jinzhao Li, Shishuang He, Zhuoran Lu, Mutian Zhang, Qin Liu, Yiwei Yang, Swati Jain, Ming Yin, Yunyao Li
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 |
|---|---|---|---|
| decouple LLM proposal from utility/uncertainty-based selection (BNN+UCB) | LLM feature engineering · 13 classification datasets, MLP | error reduction vs best baseline: 7.24% | Improved |
| same, with XGBoost | LLM feature engineering · 13 classification datasets, XGBoost | error reduction vs best baseline: 9.02% | Improved |
| add selectively elicited human preference feedback (ablation) | Ours (w/o human) · 13 datasets, MLP | error reduction: 7.24% -> 8.96% | Improved |
| add selectively elicited human preference feedback (ablation) | Ours (w/o human) · 13 datasets, XGBoost | error reduction: 9.02% -> 11.23% | Improved |
| add human feedback (ablation) | Ours · diabetes and ai_usage (MLP/XGBoost) | AUROC: diabetes XGB 85.2 -> 84.8; ai_usage MLP 68.5 -> 68.3 | Mixed |
| use LLM-based methods instead of OpenFE/AutoGluon | feature engineering · 13 datasets | AUROC | Improved |
| swap backbone LLM to GPT-3.5-turbo | Ours · 13 datasets average, XGBoost | AUROC: 86.5 vs OCTree 86.0 | No clear effect |
| use the proposed method instead of OCTree on proprietary dataset | feature engineering · conversion (MLP) | AUROC: 91.1 -> 92.6 (w/o human) | Improved |
| use algorithm-triggered human feedback instead of human-led control | human-LLM collaboration · flight user study | final performance | Improved |
| use algorithm-triggered feedback instead of self-reasoning query | human-LLM collaboration · flight user study | cognitive load / completion time: p = 0.043 / p = 0.02 | Improved |
Closest papers by meaning
- Large Language Models Can Automatically Engineer Features for Few-Shot Tabular Learning (ICML 2024 · similarity 0.90)
- Optimized Feature Generation for Tabular Data via LLMs with Decision Tree Reasoning (NeurIPS 2024 · similarity 0.89)
- TabPrep: Closing the Feature Engineering Gap in Tabular Benchmarks (NeurIPS 2026 · similarity 0.89)
- Large Scale Transfer Learning for Tabular Data via Language Modeling (NeurIPS 2024 · similarity 0.88)
- A Data-Centric Perspective on Evaluating Machine Learning Models for Tabular Data (NeurIPS 2024 · similarity 0.88)
- Strengthening LLMs for Tabular Prediction with Structural Priors (NeurIPS 2026 · similarity 0.87)
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.
