LLM-augmented feature engineering for machine learning pipelines

Short Description

In this paper, we propose a Large Language Model (LLM)-based automated feature engineering pipeline using Mistral-7B-Instruct with Retrieval-Augmented Generation and a multi-criteria validation process. It starts with prompt text and additionally integrates domain and data knowledge, to create and validate new features. We evaluate the solution in the context of an ML-based heart rate prediction process and show that the generated features achieve improvements.

Keywords

Countries

Romania