Anomaly Detection in Machine Learning

Status: public · Confidence: medium (0.8) · Basis: verified_sources

## TL;DR

Anomaly detection identifies observations that depart from expected behavior, using survey-backed statistical, isolation-based, and deep-learning methods.

## Core Explanation

The public evidence here focuses on a general survey definition, the Isolation Forest algorithm, and a modern deep-learning taxonomy rather than unsupported tool claims.

## Source-Mapped Facts

- Chandola, Banerjee, and Kumar define anomaly detection as finding patterns in data that do not conform to expected behavior. ([source](https://doi.org/10.1145/1541880.1541882))
- Isolation Forest isolates anomalies by random partitioning; anomalies tend to require fewer splits than normal points. ([source](https://doi.org/10.1109/ICDM.2008.17))
- The ACM Computing Surveys review on deep anomaly detection organizes deep methods into a taxonomy of high-level and fine-grained method categories. ([source](https://doi.org/10.1145/3439950))

## Further Reading

- [Anomaly Detection: A Survey](https://doi.org/10.1145/1541880.1541882)
- [Isolation Forest](https://doi.org/10.1109/ICDM.2008.17)
- [Deep Learning for Anomaly Detection: A Review](https://doi.org/10.1145/3439950)