|
Abstract:
We explore methods for incorporating prior knowledge about a
problem at hand in Support Vector learning machines. We show that
both invariances under group transformations and prior knowledge
about locality in images can be incorporated by constructing
appropriate kernel functions.
|
