> ML_ALGORITHM // ONE-CLASS-SVM_v1.0
One-Class Support Vector Machine (OC-SVM)
Support vector method for novelty detection that learns a decision boundary encompassing normal data while treating everything outside as anomalous.
Support Vector Methodsclassical-unsupervisedblack-boxmedium (1k-100k)
Back to All AlgorithmsComputational Complexity
Training Complexity:O(n^2 * p) to O(n^3)
Inference Complexity:O(support_vectors * p)
Hardware Profile
CPU Friendly:Yes
Requires GPU:No
Memory Footprint:moderate
Interpretability & Data
Interpretability Tier:black-box
Training Data Needs:medium (1k-100k)
Interpretability Assessment
Decisions are driven by high-dimensional kernel distances from an origin hyperplane.
Suitable Tasks & Supported Modalities
Suitable Tasks:
anomaly detectionnovelty detection
Supported Modalities:
tabularembeddings
Implementing Libraries
Foundational Literature
Common Pitfalls & Warnings
- Setting RBF gamma too high creates an overfitted boundary that encloses each training point in an isolated bubble