Ex Parte Hickey et al

Patent Trial and Appeal Board

Nov 21, 2016

12916043 (P.T.A.B. Nov. 21, 2016)

UNITED STA TES p A TENT AND TRADEMARK OFFICE
APPLICATION NO. FILING DATE FIRST NAMED INVENTOR
121916,043 10/29/2010 Marianne Hickey
56436 7590 11/23/2016
Hewlett Packard Enterprise
3404 E. Harmony Road
Mail Stop 79
Fort Collins, CO 80528
UNITED STATES DEPARTMENT OF COMMERCE
United States Patent and Trademark Office
Address: COMMISSIONER FOR PATENTS
P.O. Box 1450
Alexandria, Virginia 22313-1450
www .uspto.gov
ATTORNEY DOCKET NO. CONFIRMATION NO.
82260062 8815
EXAMINER
YOUNG, ASHLEY YA-SHEH
ART UNIT PAPER NUMBER
3623
NOTIFICATION DATE DELIVERY MODE
11/23/2016 ELECTRONIC
Please find below and/or attached an Office communication concerning this application or proceeding.
The time period for reply, if any, is set in the attached communication.
Notice of the Office communication was sent electronically on above-indicated "Notification Date" to the
following e-mail address( es):
hpe.ip.mail@hpe.com
mkraft@hpe.com
chris.mania@hpe.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
BEFORE THE PATENT TRIAL AND APPEAL BOARD
Ex parte MARIANNE HICKEY,
ANDREW BYDE,
MAHER RAHMOUNI, and
CLAUDIO BARTOLINI
Appeal2015-000550
Application 12/916,043 1
Technology Center 3600
Before HUBERT C. LORIN, ANTON W. PETTING, and
MATTHEW S. MEYERS, Administrative Patent Judges.
LORIN, Administrative Patent Judge.
DECISION ON APPEAL
STATEMENT OF THE CASE
Marianne Hickey, et al. (Appellants) seek our review under 35 U.S.C.
§ 134 of the final rejection of claims 1-8 and 20-31. We have jurisdiction
under 35 U.S.C. § 6(b) (2002).
1 Rather than identifying the real party in interest, the Appeal Brief states
that "[ t ]he real party in interest in the above-captioned application is the
assignee." Appeal Br. 2. According to current USPTO records, the assignee
is Hewlett-Packard Enterprise Development LP.
Appeal2015-000550
Application 12/916,043
We AFFIRM.
SUMMARY OF DECISION
THE INVENTION
Claim 1, reproduced below, is illustrative of the subject matter on
appeal.
1. A method for assessing health of a project, comprising:
determining, by a processor in response to computer-readable
instructions, vectors for a set of project features;
determining, by a processor in response to computer-readable
instructions, an indication of health of some aspect of the project
using a machine learning classifier that has been trained for the set of
project features using a set of training elements for one or more
projects of a data set for a plurality of time intervals, each training
element comprising a feature vector for a particular project of the data
set at a particular time interval for that particular project and a health
classification deemed to be correct for that particular project at that
particular time interval.
THE REJECTIONS
The Examiner relies upon the following as evidence of
unpatentability:
Cross US 2009/0292580 Al Nov. 26, 2009
Abe, Seiya, et al., "Estimation of Project Success Using Bayesian
Classifier," ICSE'06, pages 600-603, May 20-28, 2006, Shanghai,
China. [Abe]
The following rejections are before us for review:
1. Claims 1-8 and 20-31 are rejected under 35 U.S.C. §101 as being
directed to non-statutory subject matter.
2
Appeal2015-000550
Application 12/916,043
2. Claims 1-8 and 20-31 are rejected under 35 U.S.C. §103(a) as being
unpatentable over Cross and Abe.
ISSUES
Did the Examiner err in rejecting claims 1-8 and 20-31 under 35
U.S.C. § 101 as being directed to non-statutory subject matter?
Did the Examiner err in rejecting claims 1-8 and 20-31 under 35
U.S.C. § 103(a) as being unpatentable over Cross and Abe?
FINDINGS OF FACT
We rely on the Examiner's factual findings stated in the Answer.
Additional findings of fact may appear in the Analysis below.
ANALYSIS
The rejection of claims 1-8 and 20-31under35 US.C. §101 as being
directed to non-statutory subject matter.
The Appellants argued claims 1-8 and 20-31 as a group (Appeal Br. 9-
11 ). We select claim 1 as the representative claim for this group, and the
remaining claims 2-8 and 20-31 stand or fall with claim 1. 3 7 C.F .R.
§ 41.37(c)(l)(vii) (2007).
Alice Corp. Pty. Ltd. v. CLS Bank International, 134 S. Ct. 2347
(2014) identifies a two-step framework for determining whether claimed
subject matter is judicially-excepted from patent eligibility under § 101.
According to Alice step one, "[ w ]e must first determine whether the
claims at issue are directed to a patent-ineligible concept," such as an
abstract idea. Id. at 2355. The Examiner found that "[t]he claim(s) is/are
3
Appeal2015-000550
Application 12/916,043
directed to the abstract idea of assessing project health, which is a
fundamental economic practice and idea of itself." Ans. 11. This does not
appear to be in dispute.
Step two of Alice is "a search for an 'inventive concept'-i.e., an
element or combination of elements that is 'sufficient to ensure that the
patent in practice amounts to significantly more than a patent upon the
[ineligible concept] itself."' 134 S. Ct. at 2355 (quoting Mayo Collaborative
Servs. v. Prometheus Labs, Inc., 132 S. Ct. 1289, 1294 (2012)).
The Examiner found that
[ t ]he additional element( s) or combination of elements in the claim( s)
other than the abstract idea per se amount(s) to no more than: (i) mere
instructions to implement the idea on a computer, and/or (ii) recitation
of generic computer structure that serves to perform generic computer
functions that are well-understood, routine, and conventional activities
previously known to the pertinent industry. Viewed as a whole, these
additional claim element( s) do not provide meaningful limitation( s) to
transform the abstract idea into a patent eligible application of the
abstract idea such that the claim( s) amounts to significantly more than
the abstract idea itself.
Ans. 11.
The Appellants dispute this. According to the Appellants,
[ r ]epresentative [independent] claim 1 recites, in part,
"determining, by a processor in response to computer-readable
instructions, an indication of health of some aspect of the project
using a machine learning classifier that has been trained for the set of
project features using a set of training elements for one or more
projects of a data set for a plurality of time intervals, each training
element comprising a feature vector for a particular project of the data
set at a particular time interval for that particular project and a health
classification deemed to be correct for that particular project at that
particular time interval." Appellant contends this limitation
4
Appeal2015-000550
Application 12/916,043
meaningfully extends any abstract idea of assessing project health
beyond its mere linking to a particular technological environment.
Reply Br. 10. The other independent claims, claims 20 and 25 have similar
limitations.
A factor to consider is whether a claimed solution is necessarily
rooted in computer technology. Cf Amdocs (Israel) Limited v. Openet
Telecom, Inc, No. 2015-1180, 2016 WL 6440387, *10 (Fed. Cir. Nov. 1,
2016):
this claim entails an unconventional technological solution (enhancing
data in a distributed fashion) to a technological problem (massive
record flows which previously required massive databases). The
solution requires arguably generic components, including network
devices and "gatherers" which "gather" information. However, the
claim's enhancing limitation necessarily requires that these generic
components operate in an unconventional manner to achieve an
improvement in computer functionality. The enhancing limitation
depends not only on the invention's distributed architecture, but also
depends upon the net\vork devices and gatherers---even though these
may be generic-working together in a distributed manner.
See also DDR Holdings, LLC v. Hotels.com, L.P., 773 F.3d 1245, 1257 (Fed.
Cir. 2014):
these claims stand apart because they do not merely recite the
performance of some business practice known from the pre-Internet
world along with the requirement to perform it on the Internet.
Instead, the claimed solution is necessarily rooted in computer
technology in order to overcome a problem specifically arising in the
realm of computer networks.
The Appellants argue that, given said claim limitation, "Appellant's
claim 1 requires the use of a machine learning classifier trained in a
5
Appeal2015-000550
Application 12/916,043
specific manner that is not found in the prior art of record. Appellant
contends that it is well understood that the response of a machine learning
classifier is highly dependent upon the training data utilized." Reply Br. 10.
The evidence the Appellants provide (i.e., via the discussion of
classifiers in Specification paras. 14 and 17) appears to support that view.
Reply Br. 10. According to the Specification, "training data" are "a set of
observed examples." Specification para. 14. The Specification gives
"historical data" as an example. Specification para. 15. In other words,
classifiers will yield a result specific to and dependent on the type of
information (e.g., historical data) provided to it.
But the Appellants further argue that "a machine learning classifier
trained in a manner as recited in claim 1 would be expected to respond
differently than a machine learning classifier trained in some other manner,
such as that found in Abe et al" (Reply Br. 10, emphasis added). The
Appellants imply that machine learning classifiers are necessarily different
because they yield a result specific to and dependent on the type of
information (e.g., historical data) provided to it. To the extent the
Appellants mean to further argue that the claimed method is necessarily
rooted in computer technology, we find insufficient evidence to support it.
The record does not sufficiently show that a training scheme as
claimed - whereby a claimed machine learning classifier yields a result
specific to and dependent on the type of provided information (i.e., "a set of
training elements for one or more projects of a data set for a plurality of time
intervals, each training element ... ") - transforms a known machine learning
classifier into something structurally and/or functionally different.
6
Appeal2015-000550
Application 12/916,043
The record is more supportive of the opposite conclusion; that is, the
claimed machine learning classifier is not structurally and/ or functionally
different from what has been well understood in the art. According to the
Specification, the invention involves a common computer system "suitable
for use with various embodiments of the disclosure" (Specification para. 26).
See Specification paras. 26-28 for the common components. And the
invention is said to involve classifiers that are well understood in the art.
See Specification para. 14 (emphasis added):
Various embodiments described herein include methods of
assessing health of projects using classifiers. Machine learning
research has led to classifiers, such as statistical linear classifiers,
capable of making a classification decision on the basis of a set of
observed examples, i.e., training data. Statistical linear classifiers do
this by determining the value of a linear combination of features of
interest in relation to the training data. Examples of statistical linear
classifiers might include spam filters. Other examples of classifiers
include quadratic classifiers, kernel estimation classifiers, Bayesian
network classifiers, k-nearest neighbor (kNN) classifiers, etc.
Classifiers are well understood in the art and the embodiments are
not limited to a specific classifier.
There is insufficient discussion in the Specification to lead one of ordinary
skill in the art to understand that by training a classifier as claimed it would
be transformed from what is already well understood in the art into
something structurally and/or functionally different. Rather, one of ordinary
skill in the art would understand the change to reside in the type of
information being introduced into the classifier, not the classifier itself. See
e.g., Specification para. 17 (emphasis added): "The training data may be
randomly selected over a broad range of projects. Alternatively, the training
data may be tailored to one or more specific types of projects."
7
Appeal2015-000550
Application 12/916,043
The Appellants' remaining arguments make the point that the claim 1
machine learning classifier is "not found in the prior art." See Reply Br. 10.
However, the question here is not whether the prior art describes or discloses
using the machine learning classifier recited in claim 1 but whether there is
"an element or combination of elements [] 'sufficient to ensure that the
patent in practice amounts to significantly more than a patent upon the
[ineligible concept] itself."' Alice, 134 S. Ct. at 2355.
In summary, we find that in light of the Specification claim 1
reasonably broadly covers using conventional computer technology
comprising a common computer system and a common machine-learning
classifier operating as they are normally expected to. The Appellants have
not come forward with, and the record before us does not provide, sufficient
evidence to support, for example, finding that the claimed method for
assessing health of a project is necessarily rooted in computer technology.
Rather, the evidence weighs in favor of finding the claimed method for
assessing health of a project is instead rooted in the type of information used.
This does no more than express the assessing project health idea in more
contextual terms. Cf CyberSource Corp. v. Retail Decisions, Inc., 654 F.3d
1366, 1371 (Fed. Cir. 2011) ("The Court [Parker v. Flook, 437 U.S. 584
(1978)] rejected the notion that the recitation of a practical application for
the calculation could alone make the invention patentable."). As such, the
argued-over machine learning classifier as claimed is insufficient to ensure
that the claimed method in practice amounts to significantly more than the
"abstract idea of assessing project health" (Ans. 11) itself.
8
Appeal2015-000550
Application 12/916,043
For the foregoing reasons, the Appellants' arguments are unpersuasive
as to error in the rejection.
The rejection of claims 1-8 and 20-31under35 US.C. §103(a) as being
unpatentable over Cross and Abe.
The Examiner's position is the same for all three independent claims
1, 20, and 25. Final Act. 5-6.
Taking claim 1 as representative, the Examiner takes the position that
Cross discloses all that is claimed but for the classifier, for which Abe is
relied upon. Final Act. 5. According to the Examiner,
it would have been obvious to one of ordinary skill in the art at the
time of the invention to modify the invention of Cross to include the
classifiers of Abe for the purpose of predicting the final status of the
project (Abe, section 2.4). By doing so, one would reasonably expect
the overall appeal of the invention to increase by utilizing a common
approach to classification of categorical data to aid in predicting
project status.
Final Act. 6.
The Appellants disagree. The Appellants argue, in part,
the Examiner's foregoing assertion merely addresses what portions of
the cited references are relied upon, i.e., that Cross et al. is relied upon
to teach project health determination over various time intervals
including feature vectors and health classification at particular time
intervals, and that Abe et al. is relied upon to teach the inclusion of
machine learning classifiers, as applied to the base reference of Cross
et al. See, e.g., Examiner's Answer, page 13, last paragraph. This
assertion fails to provide any reasoned statement as to how one skilled
in the art would modify the project decay function of Cross et al. to
incorporate the Bayesian classifier of Abe et al. to arrive at the
elements of Applicant's claims.
Reply Br. 3--4. We agree.
9
Appeal2015-000550
Application 12/916,043
There is no dispute that Cross fails to disclose a machine learning
classifier. And there is no dispute that Abe discloses a machine learning
classifier. The dispute is over why one of ordinary skill in the art would be
led to include Abe's machine learning classifier in Cross so as to reach the
claimed subject matter.
The Examiner points out that "[b ]oth Cross and Abe are drawn to
project management and success determination of projects" (Final Act. 6)
and that including Abe's machine learning classifier in Cross would provide
for "predicting the final status of the project (Abe, section 2.4)" (Final Act.
6). "By doing so, one would reasonably expect the overall appeal of the
invention to increase by utilizing a common approach to classification of
categorical data to aid in predicting project status." Final Act. 6.
But the difficulty with this reasoning is that Cross and Abe are
directed to different purposes, albeit both are generally about "project
management and success determination of projects" (Final Act. 6).
Cross seeks to determine project health. This involves defining a
"project decay function", "a mathematical model or algorithm for
calculating the health of the project, referred to as the "project health." Para.
30. The process begins by defining the project. See para. 24 discussing step
105 in Fig. 1. The final status of the project is considered during the
defining of the project. "[T]he user can outline a timeline for the project,
organize a structure for development of the project, and determine a metric
for the "health" and success of the project." Para. 26. The project then
defines the project decay function to determine a project's health. See para.
30 discussing step 110 in Fig. 1.
10
Appeal2015-000550
Application 12/916,043
Abe discloses applying a Bayesian classifier to metrics to estimate the
final status (successful or unsuccessful) of a project. See Abstract. Abe
describes a study of 29 metrics. See Table 1. To determine a set of metrics
predictive of project success, three viewpoints from experts were
considered: with respect to quality, cost, and duration. Section 4.1. Table 3
shows the results when a Bayesian classifier is applied in light of said
viewpoints.
One of ordinary skill in the art with these references in hand would
possibly be led to employ Abe's Bayesian classifier in Cross's process when
the project is first defined, or maybe at the end of Cross's process. But that
would not lead one to the method as claimed wherein a classifier is used to
assess a project's health. Cross's "project decay function" and Abe's
Bayesian classifier have different purposes. Changing the purpose of Abe's
Bayesian classifier so as to direct it to assessing project health, as with
Cross's "project decay function," would amount to a reconstruction of Abe's
Bayesian classifier. "We must still be careful not to allow hindsight
reconstruction of references to reach the claimed invention without any
explanation as to how or why the references would be combined to produce
the claimed invention." Innogenetics, N. V. v. Abbott Labs., 512 F.3d 1363,
1374 n.3 (Fed. Cir. 2008). In that regard, we do not see and the Examiner
does not fully explain why one of ordinary skill would be led to substitute
Cross's "project decay function" (used for calculating a project's health)
with Abe's Bayesian classifier (used to estimate the final status (successful
or unsuccessful) of a project) and thereby reach the method as claimed
(wherein a classifier is used to assess a project's health).
11
Appeal2015-000550
Application 12/916,043
Accordingly, the rejection is not sustained.
CONCLUSIONS
The rejection of claims 1-8 and 20-31under35 U.S.C. §101 as being
directed to non-statutory subject matter is affirmed.
The rejection of claims 1-8 and 20-31 under 35 U.S.C. § 103(a) as
being unpatentable over Cross and Abe is affirmed.
DECISION
The decision of the Examiner to reject claims 1-8 and 20-31 is
affirmed.
No time period for taking any subsequent action in connection with
this appeal may be extended under 37 C.F.R. § 1.136(a). See 37
C.F.R. 1.136(a)(l )(iv).
AFFIRMED
12