Deep SVDD — the core idea.
hardAnswer
- Deep Support Vector Data Description (Ruff et al. 2018): train a neural encoder φ_θ to map all normal training data into a small hypersphere in feature space (center c, minimum radius).
- Anomaly score: distance ||φ_θ(x) - c||.
- End-to-end representation-plus-boundary learning.
- Modern default for deep anomaly detection on images / time series.
Check yourself — multiple choice
- Random
- Train NN encoder to map normals into a hypersphere in feature space; anomaly score = distance to center; deep AD default for images
- Same as one-class SVM
- Not real
Deep SVDD: NN maps normals to hypersphere; distance = anomaly score.
#anomaly-detection#deep-learning
Practise Unsupervised Learning
214 interview questions in this topic.