Loading
Life sciences · Preprint
arXiv · September 3, 2026
Early or partial results. Treat as a signal, not a conclusion.
WAND is a novel unsupervised anomaly detector that integrates explainability into its scoring mechanism via witness directions in feature space. On 47 ADBench benchmark datasets, it achieves ROC-AUC parity with 16 baselines while providing native per-feature attributions at lower computational cost than post-hoc SHAP/LIME. This is an early-stage methodological contribution that demonstrates computational and interpretability advantages on synthetic benchmarks but has not been validated in clinical, operational, or peer-reviewed contexts.
Methodological benchmarking study; algorithm comparison across multiple datasets. ADBench benchmark datasets containing unlabeled, contaminated tabular data; specific dataset characteristics, sizes, and anomaly prevalence rates not provided.. Intervention: WAND: unsupervised tabular anomaly detector using witness directions on unit sphere for scoring and integrated per-feature attribution.. Compared with: 16 unsupervised baseline anomaly detectors; post-hoc SHAP, LIME, and ECOD explanations..
WAND attains best mean Friedman rank at ROC-AUC parity with 16 unsupervised baselines across 47 ADBench datasets Native explanations from WAND are more accurate and faithful than post-hoc SHAP/LIME and ECOD at a fraction of query cost Scoring is linear in sample size with a probe-efficiency bound guaranteeing every anomaly receives an explanation via witness vector
Safety was not reported in the material analysed. Check the source before drawing any conclusion about harm.
The source did not state who this applies to in practice.
A novel computational method for unsupervised anomaly detection with integrated explainability, demonstrated on 47 benchmark datasets but lacking clinical validation, real-world deployment evidence, or peer review.
As stated by the source record.
Quoted from the source exactly as published.
Graded across the dimensions that decide whether you should act, each from what the source actually supports. There is no single score, and where a dimension was not assessed it says so.
Unsupervised anomaly detection scores each point of an unlabelled, contaminated sample in a single pass, and increasingly must also explain why a point is flagged. Yet the dominant detectors give a score with no account of which features drive it, and explanations are bolted on post-hoc with SHAP or LIME, which re-query the detector thousands of times per point and only approximate it. We introduce WAND, an unsupervised tabular anomaly detector that is explainable by design. WAND organises its computation around directions on the unit sphere, scoring each point by how far its projection escapes a sub-Gaussian extreme-value baseline. The originality of our approach is that the witness directions that flag a point, being vectors in feature space, are its explanation, a per-feature attribution obtained at no cost over scoring and, since the score is differentiable, recoverable by gradients. Scoring is linear in the sample size, and a probe-efficiency bound guarantees every anomaly a witness, hence an explanation. Across 47 ADBench datasets WAND attains the best mean Friedman rank at ROC-AUC parity with 16 unsupervised baselines, so the gain is interpretability at no accuracy cost; its native explanations are more accurate and faithful than post-hoc SHAP/LIME and ECOD at a fraction of the query cost. WAND is thus a practical, interpretable solution for explainable anomaly detection.
Taken from the source record, never inferred. Follow any of these and new work involving them reaches your briefing.