Causal-aware Anomaly Detection for Tabular Data

Anomaly detection (AD) methods often ignore causal dependencies and treat anomalies as outliers, which is brittle when anomalies are primarily mechanism violations rather than extreme values. We propose CausalAno, a causal-aware detector that trains a causal GAN on normal data and leverages its discriminator to learn mechanism-consistent representations. We score test samples by fitting a Gaussian model in this feature space and computing the Mahalanobis distance, measuring deviation from the normal causal manifold. We demonstrate the effectiveness of CausalAno with extensive experiments on 28 tabular datasets (18 continuous-only and 10 mixed-type), comparing against 16 SOTA baselines. Our results show consistent improvements across both mixed-type and numerical-only settings. Our ablation studies further confirm that the gains come from the causal factorization in the causal GAN rather than a generic GAN backbone. Our CausalAno offers a practical and effective solution for real-world applications where anomalies often arise from mechanism-violating behaviors.

Traditional anomaly detection tools are great at catching extreme outliers, but they often miss subtle anomalies where the hidden rules or relationships between variables are broken. A data point might look normal on its own, but its combination with other factors violates real-world mechanisms. We developed CausalAno, an AI system that learns the deep causal relationships within normal data using a specialized generative model. When analyzing new data, it mathematically measures how far a sample deviates from this learned "normal" structure. This allows it to successfully flag instances that break the established causal rules. After testing CausalAno across 28 different tabular datasets, it consistently outperformed 16 state-of-the-art baselines. This research matters because real-world anomalies—like financial fraud or medical errors—usually involve these complex, mechanism-violating behaviors. CausalAno offers a highly practical and accurate solution for detecting these sophisticated threats.