Conformal prediction for anomaly detection.
hardAnswer
- Given calibration set of normal points, use non-conformity score to produce p-values with guaranteed (1-α) coverage on inliers under exchangeability. p-value < α → outlier at level α.
- Distribution-free, model-agnostic → wraps any anomaly detector for calibrated alarms.
- Standard in modern high-assurance production monitoring; foundational in the CP framework (Vovk, Shafer).
Check yourself — multiple choice
- Random
- Non-conformity score + calibration set → p-values with guaranteed coverage under exchangeability; model-agnostic; calibrated anomaly alarms
- Same as z-score
- Not real
Conformal AD: p-values with guaranteed inlier coverage; model-agnostic.
#anomaly-detection#theory
Practise Unsupervised Learning
214 interview questions in this topic.