The paper proposes the novel concept of anomaly-free regions (AFRs) to improve anomaly detection. An AFR is a region in the data space for which it is known that there are no anomalies inside, e.g., via domain knowledge. This region can contain any number of normal data points and can be anywhere in the data space.
The key advantage of AFRs is that they constrain the estimation of the distribution of non-anomalies: The estimated probability mass inside the AFR must be consistent with the number of normal data points inside the AFR. The authors provide a solid theoretical foundation for this concept and a reference implementation of anomaly detection using AFRs.
The empirical results confirm that anomaly detection constrained via AFRs improves upon unconstrained anomaly detection. Specifically, the authors show that, when equipped with an estimated AFR, an efficient algorithm based on random guessing becomes a strong baseline that several widely-used methods struggle to overcome. On a dataset with a ground-truth AFR available, the current state of the art is outperformed.
To Another Language
from source content
arxiv.org
Key Insights Distilled From
by Maximilian T... at arxiv.org 10-01-2024
https://arxiv.org/pdf/2409.20208.pdfDeeper Inquiries