Question 1

What is the main goal of anomaly detection in this survey?

Accepted Answer

The survey focuses on identifying unexpected or uncommon occurrences in data, especially outliers that may indicate irregular events such as fraud in healthcare contexts.

Question 2

How do unsupervised anomaly detection methods work when labels are unavailable?

Accepted Answer

They rely on the assumption that anomalies are rare and different from normal data, learning the structure or distribution through methods such as clustering, density-based approaches, autoencoders, or PCA and then flagging deviations.

Question 3

Which techniques are highlighted as showing superior performance?

Accepted Answer

The review emphasizes Isolation Forest, Bayesian hierarchical models, and deep autoencoders as demonstrating stronger results compared with traditional approaches on relevant healthcare datasets.