Ex Parte PEDNAULT

Board of Patent Appeals and Interferences

Feb 24, 2004

09106784 (B.P.A.I. Feb. 24, 2004)

1
The opinion in support of the decision being entered today was not written for
publication and is not binding precedent of the Board.
Paper No. 17

UNITED STATES PATENT AND TRADEMARK OFFICE
_____________
BEFORE THE BOARD OF PATENT APPEALS
AND INTERFERENCES
_____________
Ex parte EDWIN PETER DAWSON PEDNAULT
_____________
Appeal No. 2002-0308
Application No. 09/106,784
______________
ON BRIEF
_______________
Before KRASS, DIXON and SAADAT, Administrative Patent Judges.
KRASS, Administrative Patent Judge.
DECISION ON APPEAL
This is a decision on appeal from the final rejection of claims 1-20.
The invention is directed to a method for constructing predictive models that can
be used to make predictions in situations where the inputs to those models can have
values that are missing or otherwise unknown.
A copy of Independent claim 1 is appended to this decision.
Appeal No. 2002-0308
Application No. 09/106,784
2
No references are relied on.
Claims 1-20 stand rejected under 35 U.S.C. §101 as being directed to non-
statutory subject matter because the claims describe an “abstract” idea.
Reference is made to the brief and answer for the respective positions of
appellant and the examiner.
OPINION
It is the examiner’s position that the instant claimed invention “lacks a practical
application providing a useful, concrete, and tangible result because it is directed to an
abstract idea. The claimed invention receives abstract input data, performs
mathematical operations, and outputs abstract data” (answer-page 11).
Appellant, naturally, takes the opposite view, contending that the instant claimed
“invention constitutes a practical application of mathematical principles to achieve a
useful, concrete, and tangible result,” that the invention “is not restricted to specific
applications,” and that predictive “modeling technology, in general, and Appellant’s
invention, in particular, are useful in a very wide range of applications” (brief-page 23)
We begin our inquiry, as we must, with the specific claims before us.
Independent claim 1 makes it very clear that we are dealing with a structure, a “program
storage device” that is readable by a machine. Thus, we are not dealing with an
“abstract” idea existing only in one’s mind. Further, the claimed subject matter is
directed to the computer program that is executed by the machine in order to perform
Appeal No. 2002-0308
Application No. 09/106,784
3
specific method steps. Accordingly, we are dealing with a computer-implemented
process that is run on a machine. Machines and processes comprise, prima facie,
statutory subject matter as they are specifically mentioned in 35 U.S.C. §101 as classes
of invention which are patentable.
The process, or method, of the instant invention is a mathematical algorithm.
While a long line of cases affirm the concept that a claim directed to a mathematical
algorithm is a nonstatutory class of invention and is not included under 35 U.S.C. §101,
a “claim drawn to subject matter otherwise statutory does not become nonstatutory
simply because it uses a mathematical formula, computer program or digital computer.”
Diamond v. Diehr, 450 U.S. 175, 187, 209 USPQ 1, 8 (1981). The question is whether
there is a practical application of this mathematical algorithm that achieves a useful,
concrete, and tangible result or whether the algorithm merely results in an abstract
number.
In the instant claimed invention, a collection of training data is collected, wherein
the training data are input values that are available to a model, together with desired
output values that the model is to predict. Then, a plurality of subordiniate models are
generated, these subordinate models together comprising an overall model. Finally, a
specification of at least one of the subordinate models generated is output and a
prediction is generated based on the at least one of the subordinate models generated.
Appeal No. 2002-0308
Application No. 09/106,784
4
While the specification addresses a specific application of this algorithm, i.e.,
used in a direct-mail targeted-marketing purpose (e.g., page 1), the instant claims are
related to a more general application of the algorithm, i.e., making a prediction with a
predictive model even when the values of some or all of the inputs are unknown. Such
a method may have application in many areas, but it does have practical application.
The mathematical algorithm manipulates numbers in order to make a predictive model
that is then used to make predictions. Reference to the instant specification, at page 1,
indicates, as an example of the useful, concrete and tangible result of the algorithm,
that a predictive model may be used “to optimize return on marketing investment by
ranking consumers according to their predicted responses to promotions, and then
mailing promotional materials only to those consumers who are most likely to respond
and generate revenue.” Thus, we cannot agree with the examiner’s assessment that
the instant claimed subject matter is, somehow, merely an “abstract idea failing to recite
a practical application having useful, concrete, and tangible result” (answer-page 3).
While a mathematical algorithm is clearly involved in the instant claimed subject
matter, we find that the claimed subject matter as a whole is directed to a combination
of computer-implemented steps which, when executed by the claimed machine, allows
for a predictive model which can make a prediction even without knowing all of the
values of the inputs. This is not a disembodied mathematical concept which may be
Appeal No. 2002-0308
Application No. 09/106,784
5
characterized as an “abstract idea,” but rather a specific machine to produce a useful,
concrete, and tangible result. In re Alappat, 33 F.3d 1526, 1544, 31 USPQ2d 1545,
1557 (Fed. Cir. 1994).
Thus, to whatever extent the instant claimed subject matter constitutes a
mathematical algorithm, it recites a practical application of a mathematical algorithm
and such subject matter falls comfortably within the broad scope of patentable subject
matter under 35 U.S.C. §101. AT&T Corp. V. Excel Communications, Inc., 172 F.3d
1352, 1355-56, 50 USPQ2d 1447,1450 (Fed. Cir. 1999).
The examiner’s decision rejecting claims 1-20 under 35 U.S.C. §101 is reversed.
REVERSED
ERROL A. KRASS )
Administrative Patent Judge )
)
)
)
) BOARD OF PATENT
JOSEPH L. DIXON ) APPEALS
Administrative Patent Judge ) AND
) INTERFERENCES
)
)
)
MAHSHID D. SAADAT )
Administrative Patent Judge )
EAK/vsh
Appeal No. 2002-0308
Application No. 09/106,784
6
FRED GIBB
McGINN & GIBB, PC
1701 CLARENDON BOULEVARD
SUITE 100
ARLINGTON, VA 22209
Appeal No. 2002-0308
Application No. 09/106,784
7
APPENDIX
Claim 1
1. A program storage device readable by a machine, tangibly
embodying a program of instructions executable by the machine to
perform method steps for constructing a predictive model that can be
used to make predictions even when the values of some or all inputs are
missing or are otherwise unknown, the method comprising:
(1) presenting a collection of training data comprising examples
of input values that are available to the model together with
corresponding desired output value(s) that the model is intended to
predict;
(2) generating a plurality of subordinate models, that together
comprise an overall model, in such a way that:
a) each subordinate model has an associated set of
application conditions that must be satisfied in order to apply
the subordinate model when making predictions, the
application conditions comprising:
i) tests for missing values for all, some, or none of
the inputs,
and
ii) tests on the values of all, some, or none of the
inputs that are applicable when the values of the inputs
mentioned in the tests have known values;
and
Appeal No. 2002-0308
Application No. 09/106,784
8
APPENDIX
cont.
b) for at least one subordinate model, the training cases
used in the construction of that subordinate model include
some cases that indirectly satisfy the application conditions
such that the application conditions are satisfied only after
replacing one or more known data values in these training
cases with missing values; and
3) outputting a specification of at least one of said subordinate
models thus generated and making a prediction based on said at
least one of said subordinate models thus-generated.