One-Class SVM: Unsupervised Anomaly Detection via Support Vector Machines

One-Class SVM (OCSVM) is a variant of Support Vector Machine designed for anomaly detection in an unsupervised setting. Unlike traditional supervised SVMs that require both positive and negative examples, OCSVM learns a decision boundary using only data from a single class (typically the normal class). Its primary goal is to identify novel or a ...

Posted on Fri, 04 Sep 2026 16:50:59 +0000 by tex1820