As shown above, PDL2's inductive learning ability
allows it to naturally handle generalization. It is
capable of learning new rules from examples, as well
as modifying existing rules. PDL2 also always
extends the partial information it is given to a total
function from the input space to the output space.
Every point of the input space is matched to some
point of the output space, either via a rule that covers
it or via its "proximity" to known points.