Cisco Technology, Inc.

Patent Trials and Appeals Board

Sep 10, 2020

2019001276 (P.T.A.B. Sep. 10, 2020)

UNITED STATES PATENT AND TRADEMARK OFFICE
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
APPLICATION NO. FILING DATE FIRST NAMED INVENTOR ATTORNEY DOCKET NO. CONFIRMATION NO.
14/275,344 05/12/2014 Javier Cruz Mota 0140368.U 2965
126982 7590 09/10/2020
Behmke Innovation Group LLC
One Marina Park Drive
Suite 1410
Boston, MA 02210
EXAMINER
VANG, MENG
ART UNIT PAPER NUMBER
2457
NOTIFICATION DATE DELIVERY MODE
09/10/2020 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):
patent@bigiplaw.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
________________

BEFORE THE PATENT TRIAL AND APPEAL BOARD
________________

Ex parte JAVIER CRUZ MOTA,
JEAN-PHILIPPE VASSEUR, and ANDREA DI PIETRO
________________

Appeal 2019-001276
Application 14/275,344
Technology Center 2400
________________

Before BRADLEY W. BAUMEISTER, ERIC B. CHEN, and
JOHN F. HORVATH, Administrative Patent Judges.

BAUMEISTER, Administrative Patent Judge.

DECISION ON APPEAL
Appellant appeals under 35 U.S.C. § 134(a) from the Examiner’s final
rejection of claims 1–3, 5–13 and 15–23, which constitute all of the pending
claims.1 Appeal Br. 6–34. We have jurisdiction under 35 U.S.C. § 6(b).
The Board conducts a limited de novo review of the appealed
rejections for error based upon the issues identified by Appellant, and in
light of the arguments and evidence produced thereon. Ex parte Frye,
94 USPQ2d 1072, 1075 (BPAI 2010) (precedential).
We affirm.

1 We use the word “Appellant” to refer to “applicant” as defined in
37 C.F.R. § 1.42. Appellant identifies the real party in interest as Cisco
Technology, Inc. Appeal Brief filed May 21, 2018 (“Appeal Br.”) 1.
Appeal 2019-001276
Application 14/275,344

2
STATEMENT OF THE REJECTIONS
Claims 1–3, 5–13, and 15–23 stand rejected under 35 U.S.C. § 101 as
being directed to patent ineligible subject matter. Final Action mailed
Oct. 16, 2017 (“Final Act.”), 7–10.
Claims 1–3, 6–13, and 16–23 stand rejected under 35 U.S.C. § 103 as
being unpatentable over Baker (US 2007/0277242 Al; published Nov. 29,
2007) and Radinsky (US 2012/0084859 Al; published Apr. 5, 2012). Final
Act. 11–18.
Claims 5 and 15 stand rejected under 35 U.S.C. § 103 as being
unpatentable over Baker, Radinsky, and Hudis (US 2008/0229422 Al;
published Sept. 18, 2008). Final Act. 18–19.

THE SECTION 101 REJECTION
The Claimed Invention and Rejection
Independent claim 1 represents the appealed claims.2 Claim 1 is
reproduced below with paragraph numbering added for clarity and emphasis
added to the claim language that recites an abstract idea:
1. A method, comprising:
[(a)] sending, by a device, voting optimization requests to a
plurality of network nodes that identify a validation data set,
wherein the plurality of network nodes are nodes that host
classifiers that are associated with a given classification problem
between classifiers distributed across a network, and wherein the
classifiers are machine learning processes that associate a label
from among a set of labels with an input set of data;
[(b)] receiving, at the device, voting optimization data from the
plurality of network nodes, wherein the network nodes generate

2 Appellant argues the section 101 rejection of all of the appealed claims
together as a group, selecting claim 1 as representative. Appeal Br. 5–13.
Appeal 2019-001276
Application 14/275,344

3
the voting optimization data by executing classifiers using the
validation data set, and wherein the voting optimization data
includes a classifier identifier for a particular classifier and an
output of the particular classifier based on the validation data set;
[(c)] selecting a set of one or more voting classifiers from
among the classifiers based on the voting optimization data,
wherein the selection results in an optimum voting strategy for
the given classification problem between the classifiers
distributed across the network and wherein the optimum voting
strategy includes the set of one or more selected voting classifiers
and a minimum number of votes required from the selected one
or more voting classifiers to establish an agreement between the
set of one or more selected voting classifiers during a vote; and
[(d)] notifying, by the device, one or more network nodes of the
plurality of network nodes of the selection, wherein each of the
notified network nodes hosts a voting classifier in the set of one
or more selected voting classifiers.
The Examiner determines that claims 1–3, 5–13, and 15–23 are
directed to a judicial exception to patent eligible subject matter without
reciting significantly more. Final Act. 7–10. More specifically, the
Examiner finds that “the claims fall under certain methods of organizing
human activity and are directed to the abstract idea of collecting
information, analyzing it, and displaying certain results of the collection and
analysis. Id. at 8 (citing Electric Power Group, LLC, v. Alstom, 830 F.3d
1350 (Fed. Cir. 2016)).
Principles of Law
A. SECTION 101:
Inventions for a “new and useful process, machine, manufacture, or
composition of matter” generally constitute patent-eligible subject matter.
35 U.S.C. § 101. However, the U.S. Supreme Court has long interpreted
35 U.S.C. § 101 to include implicit exceptions: “[l]aws of nature, natural
Appeal 2019-001276
Application 14/275,344

4
phenomena, and abstract ideas” are not patentable. Alice Corp. v. CLS Bank
Int’l, 573 U.S. 208, 216 (2014).
In determining whether a claim falls within an excluded category, we
are guided by the Court’s two-step framework, described in Mayo
Collaborative Services v. Prometheus Laboratories, Inc., 566 U.S. 66
(2012), and Alice. Alice, 573 U.S. at 217–18 (citing Mayo, 566 U.S. at 75–
77). In accordance with that framework, we first determine what concept
the claim is “directed to.” See Alice, 573 U.S. at 219 (“On their face, the
claims before us are drawn to the concept of intermediated settlement, i.e.,
the use of a third party to mitigate settlement risk.”); see also Bilski v.
Kappos, 561 U.S. 593, 611 (2010) (“Claims 1 and 4 in petitioners’
application explain the basic concept of hedging, or protecting against
risk.”).
Concepts determined to be abstract ideas, and thus patent ineligible,
include certain methods of organizing human activity, such as fundamental
economic practices (Alice, 573 U.S. at 219–20; Bilski, 561 U.S. at 611);
mathematical formulas (Parker v. Flook, 437 U.S. 584, 594–95 (1978)); and
mental processes (Gottschalk v. Benson, 409 U.S. 63, 67 (1972)). Concepts
determined to be patent eligible include physical and chemical processes,
such as “molding rubber products” (Diamond v. Diehr, 450 U.S. 175, 191
(1981)); “tanning, dyeing, making water-proof cloth, vulcanizing India
rubber, smelting ores” (id. at 182 n.7 (quoting Corning v. Burden, 56 U.S.
252, 267–68 (1854))); and manufacturing flour (Benson, 409 U.S. at 69
(citing Cochrane v. Deener, 94 U.S. 780, 785 (1876))).
In Diehr, the claim at issue recited a mathematical formula, but the
Court held that “a claim drawn to subject matter otherwise statutory does not
Appeal 2019-001276
Application 14/275,344

5
become nonstatutory simply because it uses a mathematical formula.”
Diehr, 450 U.S. at 187; see also id. at 191 (“We view respondents’ claims as
nothing more than a process for molding rubber products and not as an
attempt to patent a mathematical formula.”). Having said that, the Court
also indicated that a claim “seeking patent protection for that formula in the
abstract . . . is not accorded the protection of our patent laws, and this
principle cannot be circumvented by attempting to limit the use of the
formula to a particular technological environment.” Id. at 191 (citing
Benson and Flook) (citation omitted); see also, e.g., id. at 187 (“It is now
commonplace that an application of a law of nature or mathematical formula
to a known structure or process may well be deserving of patent
protection.”).
If the claim is “directed to” an abstract idea, we turn to the second
step of the Alice and Mayo framework, where “we must examine the
elements of the claim to determine whether it contains an ‘inventive
concept’ sufficient to ‘transform’ the claimed abstract idea into a patent-
eligible application.” Alice, 573 U.S. at 221 (internal quotation marks
omitted). “A claim that recites an abstract idea must include ‘additional
features’ to ensure ‘that the [claim] is more than a drafting effort designed to
monopolize the [abstract idea].’” Id. (alterations in original) (quoting Mayo,
566 U.S. at 77). “[M]erely requir[ing] generic computer implementation[]
fail[s] to transform that abstract idea into a patent-eligible invention.” Id.
B. USPTO SECTION 101 GUIDANCE:
In January 2019, the U. S. Patent and Trademark Office (“USPTO”)
published revised guidance on the application of 35 U.S.C. § 101.
2019 Revised Patent Subject Matter Eligibility Guidance, 84 Fed. Reg. 50
Appeal 2019-001276
Application 14/275,344

6
(Jan. 7, 2019) (“2019 Guidance”), updated by USPTO, October 2019
Update: Subject Matter Eligibility (available at https://www.uspto.gov/sites/
default/files/documents/peg_oct_2019_update.pdf) (“October 2019
Guidance Update”); see also October 2019 Patent Eligibility Guidance
Update, 84 Fed. Reg. 55942 (Oct. 18, 2019) (notifying the public of the
availability of the October 2019 Guidance Update). “All USPTO personnel
are, as a matter of internal agency management, expected to follow the
guidance.” 2019 Guidance, 84 Fed. Reg. at 51; see also October 2019
Guidance Update at 1.
Under the 2019 Guidance, we first look to whether the claim recites
the following:
(1) any judicial exceptions, including certain groupings of abstract
ideas (i.e., mathematical concepts, certain methods of organizing
human activities such as a fundamental economic practice, or mental
processes); and
(2) additional elements that integrate the judicial exception into a
practical application (see MPEP §§ 2106.05(a)–(c), (e)–(h)).
2019 Guidance, 84 Fed. Reg. at 52–55.
Only if a claim (1) recites a judicial exception and (2) does not
integrate that exception into a practical application, do we then look to
whether the claim:
(3) adds a specific limitation beyond the judicial exception that is not
“well-understood, routine, [and] conventional” in the field (see MPEP
§ 2106.05(d)); or
Appeal 2019-001276
Application 14/275,344

7
(4) simply appends well-understood, routine, conventional activities
previously known to the industry, specified at a high level of
generality, to the judicial exception.
2019 Guidance, 84 Fed. Reg. at 56.
Analysis
STEP 2A, PRONG 1:
Under step 2A, prong 1, of the 2019 Guidance, we first look to
whether the claim recites any judicial exceptions, including certain
groupings of abstract ideas (i.e., mathematical concepts, certain methods of
organizing human activities such as a fundamental economic practice, or
mental processes). 84 Fed. Reg. at 52–54.
Limitation (c) recites,
selecting a set of one or more voting classifiers from
among the classifiers based on the voting optimization data,
wherein the selection results in an optimum voting strategy for
the given classification problem between the classifiers
distributed across the network and wherein the optimum voting
strategy includes the set of one or more selected voting classifiers
and a minimum number of votes required from the selected one
or more voting classifiers to establish an agreement between the
set of one or more selected voting classifiers during a vote.
Appeal Br. 22.
This limitation is broad enough to read on a human selecting the one
or more voting classifiers after reading a computer display of the voting
optimization data. Accordingly, limitation (c)’s step of selecting a set of one
or more voting classifiers, as claimed, constitutes a mental process, such as
an evaluation or judgment that can be performed in the human mind.
The 2019 Guidance expressly recognizes such mental processes as
constituting patent-ineligible abstract ideas. 2019 Guidance, 84 Fed. Reg.
Appeal 2019-001276
Application 14/275,344

8
at 52. Because limitation (c) recites a patent-ineligible abstract idea,
limitations (c) recites a judicial exception to patent-eligible subject matter
under step 2A, prong 1, of the 2019 Guidance.
STEP 2A, PRONG 2:
Under step 2A, prong 2, of the 2019 Guidance, we next analyze
whether claim 1 recites additional elements that individually or in
combination integrate the judicial exception into a practical application.
2019 Guidance, 84 Fed. Reg. at 53–55. The 2019 Guidance provides
exemplary considerations that are indicative of an additional element or
combination of elements integrating the judicial exception into a practical
application, such as an additional element reflecting an improvement in the
functioning of a computer or an improvement to other technology or a
technical field. Id. at 55; MPEP § 2106.05(a).
Claim 1 recites the “additional elements,” “a device,” “a plurality of
network nodes,” and one or more “classifiers” on the network nodes.
Appeal Br. 22. The term “device” is a nonce word. See Welker Bearing Co.
v PHD, Inc., 550 F.3d 1090, 1096 (Fed Cir. 2008) (holding that an
unadorned term that is simply a nonce word, or a verbal construct that is not
recognized as the name of structure, is simply a substitute for the term
“means for”). Also, nothing in claim 1 reasonably indicates that the network
nodes and their classifiers are other than generic computer components and
processes that are used to carry out the abstract idea. Indeed, Appellant’s
Specification indicates that the network nodes and classifiers may be
realized by generic networked computers:
Learning machine process 248 contains computer
executable instructions executed by the processor 220 to perform
Appeal 2019-001276
Application 14/275,344

9
various functions, such as attack detection and reporting. In
general, machine learning is concerned with the design and the
development of techniques that take as input empirical data (such
as network statistics and performance indicators), and recognize
complex patterns in these data. One very common pattern among
machine learning techniques is the use of an underlying model
M, whose parameters are optimized for minimizing the cost
function associated to M, given the input data.
Spec. 7:5–12.
Use of generic computer components to perform generic machine
learning classification is insufficient to integrate a judicial exception into a
practical application. See Alice, 573 U.S. at 226 (determining that the claim
limitations “data processing system,” “communications controller,” and
“data storage unit” were generic computer components that amounted to
mere instructions to implement the abstract idea on a computer);
October 2019 Guidance Update at 11–12 (holding that a recitation of
generic-computer limitations for implementing the abstract idea “would not
be sufficient to demonstrate integration of a judicial exception into a
practical application”).
Aside from limitation (c), claim 1’s three remaining limitations are
directed to insignificant extra-solution activity. More specifically, limitation
(a) recites, “sending, by a device, voting optimization requests to a plurality
of network nodes that [perform a recited function].” Limitation (b) recites,
“receiving, at the device, voting optimization data from the plurality of
nodes, wherein the network nodes generate the voting optimization data [in a
specified manner].” Limitation (d) recites, “notifying, by the device, [those
nodes hosting the] selected voting classifiers.”
Appeal 2019-001276
Application 14/275,344

10
Limitations (a) and (b) do not add any meaningful limitations beyond
the abstract idea because they merely recite insignificant pre-solution
activity:
An example of pre-solution activity is a step of gathering data for
use in a claimed process, e.g., a step of obtaining information
about credit card transactions, which is recited as part of a
claimed process of analyzing and manipulating the gathered
information by a series of steps in order to detect whether the
transactions were fraudulent.
MPEP § 2106.05(g).
Similarly, limitation (d) does not add any meaningful limitations to
the abstract idea because it is directed to the insignificant post-solution
activity of transmitting data. See, e.g., Apple, Inc. v. Ameranth, Inc., 842
F.3d 1229, 1241–42 (Fed. Cir. 2016) (holding that printing or downloading
generated menus constituted insignificant extra-solution activity).
Appellant argues that the present claims are directed toward the
technical solution of optimizing a voting strategy for classifiers in a network.
Appeal Br. 9. More specifically, Appellant argues, “the presently claimed
invention allows a voting strategy to be computed between different
classifiers, which is optimized for a particular classification problem that an
element of a Self-Learning Network is facing, such as for instance the
classification of traffic between normal and attack.” Id. Appellant
continues,
[t]he obtained meta-classifier (voting between several
different classifiers) improves the classification performance
obtained by a single classifier and allows to combine classifiers
that consider the same output label but that have been computed
using, potentially, different machine learning techniques, such as
. . . [artificial neural networks (ANN)], [support-vector machines
(SVM)], naive Bayes, etc., and even different input features.
Appeal 2019-001276
Application 14/275,344

11
Accordingly, the amount of data exchanged between the
classification entities is reduced significantly, making this
invention well suited to [Internet of Things (IoT)] networks.
Id.
The Examiner answers, “claim 1 fails to recite any specific steps,
techniques or criteria for selecting a set of voting classifiers that would result
in an optimum voting strategy.” Ans. 5. “Instead, the ‘selecting’ step is
recited at a high level of generality and fails to require any steps that explain
how an ‘optimum voting strategy’ is achieved.” Id. We agree with the
Examiner.
Claim 1 does not recite a method for optimizing a voting strategy.
Rather, it recites sending validation data to the classifiers, receiving
optimization data identifying the classifiers and their outputs, and “selecting
a set of one or more voting classifiers . . . based on the optimization data,
wherein the selection results in an optimum voting strategy.” Appeal Br. 22
(App’x A). Nothing in claim 1 indicates how the optimum voting strategy is
determined or how the set of classifiers is selected based on the optimization
data.
Therefore, the claim is directed to an abstract idea. See Intellectual
Ventures I LLC v. Symantec Corp., 838 F.3d 1307, 1316 (Fed. Cir. 2016)
(explaining that “when a claim directed to an abstract idea ‘contains no
restriction on how the result is accomplished . . . [and] [t]he mechanism . . .
is not described, although this is stated to be the essential innovation[,]’ then
the claim is not patent-eligible”) (internal citation omitted); see also Internet
Patents Corp. v. Active Network, Inc., 790 F.3d 1343, 1348 (Fed. Cir. 2015)
(a claim that “contains no restriction on how the result is accomplished”
Appeal 2019-001276
Application 14/275,344

12
even though the result “is stated to be the essential innovation” is not patent
eligible).
Further, previous cases have found that the steps recited in claim 1
amount to conventional computer activities or routine data-gathering and
data-transmission steps. See, e.g. Elec. Power Grp., LLC v. Alstom S.A., 830
F.3d 1350, 1355 (Fed. Cir. 2016) (“Nothing in the claims, understood in
light of the specification, requires anything other than off-the-shelf,
conventional computer, network, and display technology for gathering,
sending, and presenting the desired information.”); OIP Techs. v.
Amazon.com, Inc., 788 F.3d 1359, 1363, (Fed. Cir. 2015) (claims reciting,
inter alia, sending messages over a network, gathering statistics, using a
computerized system to automatically determine an estimated outcome, and
presenting offers to potential customers found to merely recite “well-
understood, routine conventional activities” by either requiring conventional
computer activities or routine data-gathering steps); see also buySAFE, Inc.
v. Google, Inc., 765 F.3d 1350, 1355 (Fed. Cir. 2014) (“That a computer
receives and sends the information over a network—with no further
specification—is not even arguably inventive.”); Bancorp Servs. v. Sun Life
Assurance Co., of Can. (U.S.), 687 F.3d 1266, 1278 (Fed. Cir. 2012) (“The
computer required by some of Bancorp’s claims is employed only for its
most basic function, the performance of repetitive calculations, and as such
does not impose meaningful limits on the scope of those claims.”).
For these reasons, Appellant does not persuade us that claim 1 is
directed to an improvement in the function of a computer or to any other
technology or technical field. MPEP § 2106.05(a). Nor has Appellant
persuasively demonstrated that claim 1 is directed to a particular machine or
Appeal 2019-001276
Application 14/275,344

13
transformation. MPEP §§ 2106.05(b), (c). Nor has Appellant persuasively
demonstrated that claim 1 adds any other meaningful limitations for the
purposes of the analysis under Section 101. MPEP § 2106.05(e).
Accordingly, Appellant has not persuaded us that claim 1 integrates the
recited abstract ideas into a practical application within the meaning of the
2019 Guidance. See 2019 Guidance, 84 Fed. Reg. at 52–55.
STEP 2B:
Under step 2B of the 2019 Guidance, we next analyze whether
claim 1 adds any specific limitations beyond the judicial exception that,
either alone or as an ordered combination, amount to more than “well-
understood, routine, conventional” activity in the field. 84 Fed. Reg. at 56;
MPEP § 2106.05(d).
Appellant’s Specification describes the claimed device and network
nodes and their classifiers generically, indicating that these additional
elements were well-understood, routine, and conventional:
According to various embodiments, lightweight learning
machine classifiers may be distributed to network nodes for
purposes of attack detection. In general, a classifier refers to a
machine learning process that is operable to associate a label
from among a set of labels with to an input set of data. For
example, a classifier may apply a label (e.g., “Attack” or
“No Attack”) to a given set of network metrics (e.g., traffic rate,
etc.). The distributed classifiers may be considered
“lightweight” in that they may have lower computational
requirements than a full-fledged classifier, at the tradeoff of
lower performance. To improve attack detection, a central
computing device (e.g., a FAR, NMS, etc.) that has greater
resources may execute a more computationally intensive
classifier in comparison to the distributed lightweight classifier.
In cases in which a distributed classifier detects an attack, it may
provide data to the central device to validate the results and/or to
Appeal 2019-001276
Application 14/275,344

14
initiate countermeasures. However, since the performance of a
distributed classifier may be relatively low, this also means that
there may be a greater amount of false positives reported to the
central classifier.
Spec. 14; see also, e.g., Spec. 3 (describing conventional computer networks
as being a distributed collection of nodes interconnected by communication
links); Spec. 28 (“while the techniques herein are described primarily in the
context of an LLN [Low power and Lossy Network], the techniques herein
may be applied more generally to any form of computer network, such as an
enterprise network.”).
Furthermore, Appellant’s Specification does not indicate that
consideration of these conventional elements as an ordered combination
adds any significance beyond the additional elements, as considered
individually. Rather, Appellant’s Specification indicates that the invention
is directed to an abstract idea that is made more efficient with generic
computer components—“the computation of an optimized voting strategy
for a particular classification problem.” Spec. 27.
For these reasons, we determine that claim 1 does not recite additional
elements that, either individually or as an ordered combination, amount to
significantly more than the judicial exception within the meaning of the
2019 Guidance. 84 Fed. Reg. at 52–55; MPEP § 2106.05(d).
Accordingly, we sustain the Examiner’s rejection of claim 1 under
35 U.S.C. § 101 as being directed to patent-ineligible subject matter. We,
likewise, sustain the 101 rejection of claims 2, 3, 5–13, and 15–23, which
Appellant does not argue separately. Appeal Br. 13.

Appeal 2019-001276
Application 14/275,344

15
THE SECTION 103 REJECTION
The Examiner finds the combination of Baker and Radinsky teaches
all elements of claim 1. See Final Act. 11–15. Regarding the step of
“selecting a set of one or more voting classifiers from among the classifiers
based on the voting optimization data, wherein the selection results in an
optimum voting strategy for the given classification problem,” the Examiner
finds Radinsky teaches this limitation. Id. at 14 (citing Radinsky ¶¶ 25, 30,
36, Figs. 1, 2).
Appellant disagrees for several reasons. See Appeal Br. 13–20. First,
Appellant argues “nothing in Baker optimizes the voting strategy of the
nodes in the network.” Id. at 15. This argument is unpersuasive because the
Examiner relies on Radinsky, not Baker, for teaching this limitation. See
Final Act. 14; see also In re Keller, 642 F.2d 413, 425 (CCPA 1981); In re
Merck & Co., 800 F.2d 1091, 1097 (Fed. Cir. 1986).
Appellant next argues, “Radinsky does not discuss a process for
optimizing a voting strategy in the network. That is, Radinsky does not say
how it optimizes its selection process.” Appeal Br. 15 (emphasis added).
This argument is unpersuasive because, as discussed above, claim 1 does not
recite how the classification engines are selected to generate an optimum
voting strategy. Rather, claim 1 recites “selecting a set of one or more
voting classifiers . . . based on the voting optimization data, wherein the
selection results in an optimum voting strategy.” Id. at 22 (App’x A).
Claim 1 does not recite how the optimization data is used to select the
classifiers that form an optimum voting strategy.
Moreover, we agree with the Examiner that Radinsky teaches
optimizing a voting strategy, Radinsky teaches selection component 114
Appeal 2019-001276
Application 14/275,344

16
contains one or more classifiers 116 that select one or more engines 104 to
generate a candidate engine 118. See Radinsky ¶¶ 25–26, Fig. 1. The
engines 104 are selected because they are known to be good at detecting
certain types of files. Id. ¶ 30. That is, engines 104 are selected based on
optimization data as required by claim 1 (i.e., engine identifiers and outputs).
Id. The outputs of selected engines 104 (i.e., candidate engine 118) are
provided to aggregation component 128, which determines an overall output
130. Id. ¶ 26, Fig. 1. Thus, Radinsky generates an optimum voting strategy
comprising selected engines 104 and aggregation component 128.
Appellant next argues, the “combination of Baker and Radinsky does
not provide a process for optimizing a voting strategy” because it does not
teach “wherein the optimum voting strategy includes: (1) the set of one or
more selected voting classifiers and (2) a minimum number of votes required
from the selected one or more voting classifiers.” Appeal Br. 16 (emphasis
omitted). The Examiner relies on Baker to teach an optimum voting strategy
that includes a set of classifiers and requires a minimum number of classifier
votes. See Final Act. 13 (citing Baker ¶¶ 31, 33, 40). Appellant argues this
is error because “the voting process is static in Baker,” which “initiates a
non-optimized voting process using the strategy already implemented in the
network.” Appeal Br. 18.
We are not persuaded by Appellant’s arguments, which fail to
consider what the combined teachings of Baker and Radinsky would have
suggested to a person skilled in the art. See Keller, 642 F.2d at 425; Merck,
800 F.2d at 1091. As explained above, Radinsky selects one or more
engines 104 to generate candidate engine 118 and aggregation component
128 to aggregate the outputs of engines 104 in candidate engine 118 (i.e., an
Appeal 2019-001276
Application 14/275,344

17
optimum voting strategy). See Radinsky ¶¶ 25–26, 30, Fig. 1. Baker
teaches an optimized voting strategy could consist of a threshold number of
votes from a selected set of classifiers. See Baker ¶¶ 29–31 (teaching an
agent sends an attack petition to a plurality of agents, and a leader monitors
the votes from the plurality of agents for a threshold number of votes).3
Finally, Appellant argues, “Radinsky does not discuss a process for
optimizing a voting strategy” because it only “makes the selection of
classification engines based on information learned off-line. Thus, the
selection process does not utilize machine learning.” Appeal Br. 19. We
disagree.
Although Radinsky selects engines 104 because “[o]ff-line, it can be
learned that a subset of the expensive engines 104 are good at detecting
BOTs,” this does not mean that the selection process does not use machine
learning. Radinsky ¶ 30. Indeed, Radinsky provides “a description for
training machine learning classifiers to be applied for expert selection.
Specifically, the algorithm is divided into the classifier’s training and the
classifier’s application on unknown and possibly realtime data.” Id. ¶ 65
(emphasis added). Thus, each “classifier can be provided per anti-malware

3 We note that Radinsky also teaches an optimum voting strategy consisting
of a selected set of classifiers and a minimum number of classifier votes.
Radinsky’s candidate engine 118 is a set of selected classifiers and
Radinsky’s aggregation component 128 “process[es] the correspond outputs
of classification information 122 and output[s] an overall detection output
130.” Radinsky ¶ 26. Radinsky further teaches the classifier outputs “can
be combined in one of several ways to make the final determination if the
unknown file is malware,” including by determining “if at least one anti-
malware engine classifies the unknown file as malware” or “if two or more
engines detect the file is malware.” Id. ¶¶ 85–86.
Appeal 2019-001276
Application 14/275,344

18
engine (that can be trained offline in a controlled environment). However,
this is not to be construed as so limited as training can also be performed on
real examples (e.g., online or offline . . .).” Id. ¶ 66.
Accordingly, for the reasons explained above, Appellant has failed to
demonstrate the Examiner erred in rejecting claim 1 as obvious over Baker
and Radinsky.
Appellant does not separately argue for the patentability of claims 2,
3, 6–13, and 16–23 over the combination of Baker and Radinsky. See
Appeal Br. 13–20. Accordingly, we affirm the rejection of claims 2, 3, 6–
13, and 16–23 for the same reasons as claim 1.
Finally, Appellant argues claims 5 and 15 are patentable over the
combination of Baker, Radinsky, and Hudis for the same reasons as claims 1
and 11. Id. at 20. Accordingly, we affirm the rejection of claims 5 and 15
for the same reasons as claim 1.

DECISION SUMMARY
In summary:

Claims Rejected 35 U.S.C. § Reference(s)/
Basis
Affirmed Reversed
1–3, 5–13, 15–23 101 Eligibility 1–3, 5–13,
15–23

1–3, 6–13, 16–23 103 Baker,
Radinsky
1–3, 6–13,
16–23

5, 15 103 Baker,
Radinsky,
Hudis
5, 15
Overall
Outcome
1–3, 5–13,
15–23

Appeal 2019-001276
Application 14/275,344

19
TIME PERIOD FOR RESPONSE
No time period for taking any subsequent action in connection with
this appeal may be extended under 37 C.F.R. § 1.136(a)(1). See 37 C.F.R.
§ 1.136(a)(1)(iv).
AFFIRMED