AutoML for Multi-Class Anomaly Compensation of Sensor Drift
Melanie Schaller, Mathis Kruse, Antonio Ortega, Marius Lindauer, Bodo Rosenhahn · 2025
Read AutoML for Multi-Class Anomaly Compensation of Sensor Drift by Melanie Schaller, Mathis Kruse, Antonio Ortega, Marius Lindauer, Bodo Rosenhahn (2025) 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-2502-19180.txt.
Description
Melanie Schaller and four co-authors use automated machine learning to fight sensor drift in industrial systems, in a 2025 paper. Sensors age, and their slow drift quietly poisons the models that read them. The team first show that the usual way of validating models, cross-validation, flatters them here: shuffling data lets future drift leak into training, hiding the problem. They propose a validation scheme that respects time and drift.
On top of that, their AutoML system searches over balancing, ensembles, features, and tuning to build drift-resistant classifiers. It also adapts as drift severity changes. The combined approach markedly improved classification under drift in their tests. The authors present the work as a practical kit for factories whose sensors cannot be recalibrated often, keeping monitoring honest as instruments age.
Catalog record
| CREATOR | Melanie Schaller, Mathis Kruse, Antonio Ortega, Marius Lindauer, Bodo Rosenhahn |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 59 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-19180