CausalTAD: Injecting Causal Knowledge into Large Language Models for Tabular Anomaly Detection

TL;DR

CausalTAD improves tabular anomaly detection by integrating causal relationships into large language models, reordering and reweighting columns to enhance detection accuracy, outperforming existing methods across numerous datasets.

Contribution

The paper introduces a novel approach that injects causal knowledge into LLMs for better anomaly detection in tabular data, addressing the limitations of previous text-based conversion methods.

Findings

Outperforms state-of-the-art methods on over 30 datasets

Effective causal reordering improves anomaly detection accuracy

Reweighting strategy enhances the contribution of relevant columns

Abstract

Detecting anomalies in tabular data is critical for many real-world applications, such as credit card fraud detection. With the rapid advancements in large language models (LLMs), state-of-the-art performance in tabular anomaly detection has been achieved by converting tabular data into text and fine-tuning LLMs. However, these methods randomly order columns during conversion, without considering the causal relationships between them, which is crucial for accurately detecting anomalies. In this paper, we present CausalTaD, a method that injects causal knowledge into LLMs for tabular anomaly detection. We first identify the causal relationships between columns and reorder them to align with these causal relationships. This reordering can be modeled as a linear ordering problem. Since each column contributes differently to the causal relationships, we further propose a reweighting strategy to assign different weights to different columns to enhance this effect. Experiments across more than 30 datasets demonstrate that our method consistently outperforms the current state-of-the-art methods. The code for CausalTAD is available at https://github.com/350234/CausalTAD.