One-Class SVM
A boundary learned from only normal training examples, flagging anything outside it as novel, without ever seeing an example of what abnormal looks like.
Think of It Like This
Like a security guard who memorizes every authorized employee's face, challenging anyone who walks through the lobby without a familiar gait.
The nu parameter bounds the fraction of training points allowed to fall outside the boundary, trading a wide, forgiving fit against a tight, strict one — every setting trades detection rate against false-positive rate, with no choice improving both.
The method assumes the training data is clean, uncontaminated normal data; feeding it data with anomalies already mixed in teaches the boundary to consider those anomalies normal too.