5.6. Class Separability Measures
The emphasis in the previous section was on techniques referring to the discrimination properties of individual features. However, such methods neglect to take into account the correlation that unavoidably exists among the various features and influences the classification capabilities of the feature vectors that are formed. Measuring the discrimination effectiveness of feature vectors will now become our major concern. This information will then be used in two ways. The first is to allow us to combine features appropriately and end up with the “best” feature vector for a given dimension l. The second is to transform the original data on the basis of an optimality criterion in order to come up with features ...
Get Pattern Recognition, 4th Edition now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
