Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Xu, Tommy, Zhang, Zhitian, Sun, Xiangyu, Zung, Lauren Kelly, Hajimirsadeghi, Hossein, Mori, Greg
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2505.21807
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866913918773886976
author Xu, Tommy
Zhang, Zhitian
Sun, Xiangyu
Zung, Lauren Kelly
Hajimirsadeghi, Hossein
Mori, Greg
author_facet Xu, Tommy
Zhang, Zhitian
Sun, Xiangyu
Zung, Lauren Kelly
Hajimirsadeghi, Hossein
Mori, Greg
contents Predictive modeling on tabular data is the cornerstone of many real-world applications. Although gradient boosting machines and some recent deep models achieve strong performance on tabular data, they often lack interpretability. On the other hand, large language models (LLMs) have demonstrated powerful capabilities to generate human-like reasoning and explanations, but remain under-performed for tabular data prediction. In this paper, we propose a new approach that leverages reasoning-based LLMs, trained using reinforcement learning, to perform more accurate and explainable predictions on tabular data. Our method introduces custom reward functions that guide the model not only toward better prediction accuracy but also toward human-understandable reasons for its predictions. The proposed method is evaluated on financial benchmark datasets and compared against established LLMs.
format Preprint
id arxiv_https___arxiv_org_abs_2505_21807
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle TabReason: A Reinforcement Learning-Enhanced Reasoning LLM for Explainable Tabular Data Prediction
Xu, Tommy
Zhang, Zhitian
Sun, Xiangyu
Zung, Lauren Kelly
Hajimirsadeghi, Hossein
Mori, Greg
Machine Learning
Predictive modeling on tabular data is the cornerstone of many real-world applications. Although gradient boosting machines and some recent deep models achieve strong performance on tabular data, they often lack interpretability. On the other hand, large language models (LLMs) have demonstrated powerful capabilities to generate human-like reasoning and explanations, but remain under-performed for tabular data prediction. In this paper, we propose a new approach that leverages reasoning-based LLMs, trained using reinforcement learning, to perform more accurate and explainable predictions on tabular data. Our method introduces custom reward functions that guide the model not only toward better prediction accuracy but also toward human-understandable reasons for its predictions. The proposed method is evaluated on financial benchmark datasets and compared against established LLMs.
title TabReason: A Reinforcement Learning-Enhanced Reasoning LLM for Explainable Tabular Data Prediction
topic Machine Learning
url https://arxiv.org/abs/2505.21807