{"slug": "unsupervised-anomaly-detection-using-flow-matching-on-tabular-data", "title": "Unsupervised Anomaly Detection Using Flow Matching on Tabular Data", "summary": "A new arXiv preprint (2608.19801v1) finds that the choice of anomaly scoring function is critical for flow-matching-based unsupervised anomaly detection on tabular financial data, with trajectory-based Deviation and Reconstruction scores proving more robust to training-set contamination than the original single-step Decision score used by Time-Conditioned Contraction Matching (TCCM). The study, which compares TCCM with Forest-Flow under contaminated training data, shows that Forest-Flow becomes competitive with, and sometimes outperforms, TCCM when using these alternative scores, highlighting the importance of anomaly scoring in financial anomaly detection under severe class imbalance.", "body_md": "arXiv:2608.19801v1 Announce Type: new\nAbstract: Financial anomaly detection often relies on large unlabeled transaction logs, where anomalous samples may already be present during training. Such training-set contamination violates the clean-normal data assumption underlying many anomaly detection methods. Although flow matching has demonstrated strong performance in generative modeling, its robustness in unsupervised tabular anomaly detection remains underexplored. In this work, we study flow-matching-based anomaly detection under contaminated training data by comparing Time-Conditioned Contraction Matching (TCCM) with Forest-Flow and evaluating multiple anomaly scoring functions. Our results show that the choice of anomaly score is critical. The original single-step Decision score used by TCCM is sensitive to contamination, whereas trajectory-based Deviation and Reconstruction scores provide more stable anomaly signals. With these scores, Forest-Flow becomes competitive with, and in some cases outperforms, TCCM. These findings highlight the importance of anomaly scoring for flow-matching methods in financial anomaly detection under severe class imbalance.", "url": "https://wpnews.pro/news/unsupervised-anomaly-detection-using-flow-matching-on-tabular-data", "canonical_source": "https://www.machinebrief.com/news/unsupervised-anomaly-detection-using-flow-matching-on-tabula-cmfx", "published_at": "2026-08-21 04:00:00+00:00", "updated_at": "2026-08-21 04:14:41.497810+00:00", "lang": "en", "topics": ["machine-learning", "artificial-intelligence"], "entities": ["arXiv", "Time-Conditioned Contraction Matching", "Forest-Flow"], "alternates": {"html": "https://wpnews.pro/news/unsupervised-anomaly-detection-using-flow-matching-on-tabular-data", "markdown": "https://wpnews.pro/news/unsupervised-anomaly-detection-using-flow-matching-on-tabular-data.md", "text": "https://wpnews.pro/news/unsupervised-anomaly-detection-using-flow-matching-on-tabular-data.txt", "jsonld": "https://wpnews.pro/news/unsupervised-anomaly-detection-using-flow-matching-on-tabular-data.jsonld"}}