Can you help me find the name of this classification method:

Assume we have $n$ dimensional feature vectors we want to classify in two classes.

  1. We model the classes as two $n$ dimensional gaussian distributions estimated from the data.
  2. We classify a new vector to the class that maximizes the PDF (probability density function) at that point.
  • 1
    $\begingroup$ Two component Gaussian mixture model? $\endgroup$
    – Bey
    Jan 14 '19 at 3:44

Probably Quadratic Discriminant Analysis.

There are also names for different constraints you could make:

  1. Covariance matrices of both classes are equal - Linear Discriminant Analysis.

  2. Only diagonal elements of the covariance matrix are non-zero - Naive Bayes Classifier

  3. Covariance matrix is identity (diagonals = 1, non-diagonals = 0) - Nearest Centroid Classifier


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.