Learning from Examples

To implement even the simplest minimum-distance classifier, you have to know the mean vectors (templates). To implement a minimum-Mahalanobis-distance classifier, you need to know the mean vectors and the covariance matrix. Usually, this means that you need to estimate these parameters from examples of the patterns that you want to classify. In this section, we consider the following topics:
  1. Learning the mean vector
  2. Learning the covariance matrix
  3. Regularization
  4. Validation

Left arrow Back to Mahalanobis Classifiers Up arrow Up to PR for HCI