REFEAT guides an LLM with six reasoning prompts, selected by a bandit algorithm, to generate tabular features that improve average accuracy over baselines on 59 OpenML datasets.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Tabular Feature Discovery With Reasoning Type Exploration
REFEAT guides an LLM with six reasoning prompts, selected by a bandit algorithm, to generate tabular features that improve average accuracy over baselines on 59 OpenML datasets.