MLOps Glossary

Outlier detection

Outlier detection:

(aka anomaly detection) A machine learning task for identifying instances which are “unusual” in some sense to what they should be. Typically this is an unsupervised task as we don’t always know what “unusual” looks like beforehand. This is important for production as most machine learning models will give untrustworthy or plain wrong predictions on outliers. There is typically a separate model for outlier detection alongside the original model.

SEE HOW SELDON WORKS FOR YOU

Serve, monitor, explain, and manage your models today.

© 2023 Seldon Technologies. All Rights Reserved.

Rise London
41 Luke Street
Shoreditch
EC2A 4DP

UK: +44 (20) 7193-6752
US. +1 (646) 397-9911

Email: hello@seldon.io