arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2501.02191cs.LGcs.SI
收藏

基于高阶消息传递的 LLM 增强混合型数据插补

On LLM-Enhanced Mixed-Type Data Imputation with High-Order Message Passing

The University of New South Wales(新南威尔士大学);Antai College of Economics & Management, Shanghai Jiao Tong University(上海交通大学安泰经济与管理学院);Zhejiang Gongshang University(浙江工商大学);Data61, CSIRO(联邦科学与工业研究组织Data61研究院)

Jianwei Wang, Kai Wang, Ying Zhang, Wenjie Zhang, Xiwei Xu, Xuemin Lin

AI总结:

针对数值、分类与文本混合表格的缺失值问题,提出 UnIMP 框架,通过超图、BiHMP 双向高阶消息传递、Xfusion 融合 LLM 及两阶段训练提升插补性能。

摘要:

缺失数据插补旨在填补原始数据集中的缺失值,以实现数据集的完整性;它对于大型语言模型(LLM)等现代数据驱动模型至关重要,并在过去数十年中吸引了越来越多的关注。尽管其十分重要,现有的缺失数据插补方案要么:1)仅支持数值型和分类型数据;要么 2)由于其设计优先考虑文本数据,且缺乏表格数据插补所需的关键属性,表现出不能令人满意的性能。本文提出 UnIMP,这是一个统一插补(Unified IMPutation)框架,利用 LLM 和高阶消息传递来增强对混合型数据(包括数值型、分类型和文本数据)的插补。具体而言,我们首先引入一种面向单元格的超图来对表格进行建模。随后,我们提出 BiHMP,这是一种高效的双向高阶消息传递(Bidirectional High-order Message-Passing)网络,用于在所构建的超图上聚合全局—局部信息和高阶关系,同时捕获列间异质性和列内同质性。为了有效且高效地使 LLM 的能力与 BiHMP 聚合的信息相匹配,我们引入 Xfusion;它与 BiHMP 一起充当 LLM 的适配器。我们遵循预训练和微调流程来训练 UnIMP,并集成两项优化:分块技术,即将表格划分为更小的块以提升效率;以及渐进式掩码技术,即逐步使模型适应并学习更复杂的数据模式。理论证明和在 10 个真实世界数据集上的实证实验均突显了 UnIMP 相对于现有技术的优越性。

英文摘要:

Missing data imputation, which aims to impute the missing values in the raw datasets to achieve the completeness of datasets, is crucial for modern data-driven models like large language models (LLMs) and has attracted increasing interest over the past decades. Despite its importance, existing solutions for missing data imputation either 1) only support numerical and categorical data or 2) show an unsatisfactory performance due to their design prioritizing text data and the lack of key properties for tabular data imputation. In this paper, we propose UnIMP, a Unified IMPutation framework that leverages LLM and high-order message passing to enhance the imputation of mixed-type data including numerical, categorical, and text data. Specifically, we first introduce a cell-oriented hypergraph to model the table. We then propose BiHMP, an efficient Bidirectional High-order Message-Passing network to aggregate global-local information and high-order relationships on the constructed hypergraph while capturing the inter-column heterogeneity and intra-column homogeneity. To effectively and efficiently align the capacity of the LLM with the information aggregated by BiHMP, we introduce Xfusion, which, together with BiHMP, acts as adapters for the LLM. We follow a pre-training and fine-tuning pipeline to train UnIMP, integrating two optimizations: chunking technique, which divides tables into smaller chunks to enhance efficiency; and progressive masking technique, which gradually adapts the model to learn more complex data patterns. Both theoretical proofs and empirical experiments on 10 real world datasets highlight the superiority of UnIMP over existing techniques.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑