You plan to use unsupervised anomaly detection to classify tissue samples as normal or mutated. Which two conditions support using an unsupervised anomaly detection method?

Unsupervised anomaly detection is ideal when anomalies (mutations) are rare compared to normal data, as it focuses on identifying data points that deviate significantly from the majority. If anomalies were equally common, supervised methods would be more appropriate. Additionally, unsupervised methods work best when future anomalies are expected to share similar characteristics with existing anomalies, allowing the model to learn a consistent definition of "normal" and flag deviations. If future anomalies had entirely different features, a model trained on past data might fail to detect them effectively. Having existing labels would make supervised learning the preferred approach, not unsupervised.