Conditional anomaly detection methods for patient-management alert systems

Michal Valko, Gregory Cooper, Shyam Visweswaran, Milos Hauskrecht · 2026

Read Conditional anomaly detection methods for patient-management alert systems by Michal Valko, Gregory Cooper, Shyam Visweswaran, Milos Hauskrecht (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-00490.txt.

Description

Michal Valko and Milos Hauskrecht improve conditional anomaly detection, the search for data points that are odd relative to their context. In the conditional version, a pattern counts as anomalous only on a subset of attributes, judged against the values of the rest. Instance-based methods for this task lean hard on a distance metric, since the nearest examples decide what counts as normal here.

The authors study and devise a metric learning method that tunes the distance itself to reflect the conditional anomaly pattern as well as possible. Learning the metric, rather than accepting a default one, targets exactly the comparisons the detector will make. The paper's contribution is methodological: it shows how supervised metric learning can be adapted to conditional anomalies, where what 'near' should mean changes with context. That adaptation aims to lift the accuracy of a whole family of instance-based detectors.

Catalog record

CREATORMichal Valko, Gregory Cooper, Shyam Visweswaran, Milos Hauskrecht
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE11 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.00490

Browse the full library · About the Tower of Babel