Appeal No. 95-0633
Application 07/728,426
a manner which was consistent with the law at that time, the
most recent decisions of the Court of Appeals for the Federal
Circuit cast substantial doubt on the propriety of this test.
It is the current view of the court that unpatentable
mathematical algorithms are identifiable by showing that they
are merely abstract ideas constituting disembodied concepts or
truths that are not “useful.” From a practical standpoint,
this means that to be patentable an algorithm must be applied
in a “useful” way. See State Street Bank & Trust Co. v.
Signature Financial Group, Inc., 149 F.3d 1368, 47 USPQ2d 1596
(Fed. Cir. 1998).
Independent claim 22 is directed to a method for training
a neural network classifier. Pairs of data points are
determined and used as input/output pairs to train a neural
network classifier. We are of the view that the training of a
neural network clearly has practical utility. Even if the
mathematical algorithm by which the data pairs are determined
can be considered an abstract idea, that abstract idea is
clearly employed in a useful way. The transformation of data
through a series of mathematical calculations to produce
input/output training pairs for a neural network classifier
5
Page: Previous 1 2 3 4 5 6 7 8 Next
Last modified: November 3, 2007