One-class SVM — mechanism and pitfalls.
hardAnswer
- Learns a decision boundary enclosing 'normal' data by maximizing margin from origin in RKHS (with RBF kernel typically).
- Parameter ν upper-bounds fraction of outliers.
- Sensitive to hyperparameters (ν, γ).
- Doesn't scale beyond ~10k.
- Modern alternative: Isolation Forest (faster, more robust).
- Still useful when a semantic 'normal' class is well-defined and features are dense.
Check yourself — multiple choice
- Random
- Boundary enclosing 'normal' in RKHS with RBF kernel; ν upper-bounds outlier fraction; hyperparameter-sensitive; poor scaling — Isolation Forest often preferred
- Same as SVM classifier
- Not real
One-class SVM: RKHS boundary; ν bounds outlier fraction; sensitive to params.
#anomaly-detection
Practise Unsupervised Learning
214 interview questions in this topic.
Related questions
- What are the main approaches to anomaly detection?
- How do outliers affect k-means and how do you handle them?
- Isolation Forest — how does it detect anomalies?
- Local Outlier Factor (LOF) — how does it work?
- Elliptic envelope / robust Mahalanobis distance — when to use?
- Autoencoder for anomaly detection — how?