Adapting prior-data fitted networks for tabular anomaly detection
Researchers adapt TabPFN features for tabular anomaly detection, with ZEN and FOCUS beating ADBench baselines.
The paper adapts prior-data fitted networks for unsupervised tabular anomaly detection, where no labeled anomalies exist before deployment and the reference set may itself be contaminated. Frozen TabPFN nearest-neighbor scoring, called ZEN, already exceeds every ADBench baseline in mean AUROC, and reference-set fine-tuning, called FOCUS, improves further. The method generalizes across PFN models.
- Frozen TabPFN nearest-neighbor scoring already beats every ADBench baseline
- FOCUS fine-tunes features on a possibly contaminated reference set without anomaly labels
- The approach generalizes across prior-data fitted network models
From the original article
1 min read at huggingface.co
While deep features have transformed anomaly detection in images and video, their impact on tabular data has been less substantial, partly due to the limited availability of strong deep representations. Recently, prior-data fitted networks (PFNs) have emerged as a promising source of such representations for tabular data.
By Maximilian Bershtman, Niv Cohen · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.