Adobe Inc.

Patent Trials and Appeals Board

Feb 25, 2021

2020005411 (P.T.A.B. Feb. 25, 2021)

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.
14/542,112 11/14/2014 Bruno Castro da Silva P5021-US 5279
108982 7590 02/25/2021
FIG. 1 Patents
116 W. Pacific Avenue
Suite 200
Spokane, WA 99201
EXAMINER
DAGNEW, SABA
ART UNIT PAPER NUMBER
3682
NOTIFICATION DATE DELIVERY MODE
02/25/2021 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):
Fig1Docket@fig1patents.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
____________________

BEFORE THE PATENT TRIAL AND APPEAL BOARD
____________________

Ex parte BRUNO CASTRO DA SILVA
and TRUNG H. BUI
__________________

Appeal 2020-005411
Application 14/542,112
Technology Center 3600
____________________

Before MURRIEL E. CRAWFORD, JAMES P. CALVE, and
BRADLEY B. BAYAT, Administrative Patent Judges.

CALVE, Administrative Patent Judge.

DECISION ON APPEAL

STATEMENT OF THE CASE
Pursuant to 35 U.S.C. § 134(a), Appellant1 appeals from the decision
of the Examiner to reject claims 1–18, 20, and 21, which are all of the
pending claims.2 We have jurisdiction under 35 U.S.C. § 6(b).
We AFFIRM.

1 “Appellant” refers to “applicant” as defined in 37 C.F.R. § 1.42. Appellant
identifies Adobe Inc. as the real party in interest. Appeal Br. 2.
2 Claim 19 is cancelled. Appeal Br. 67 (Claims App.).
Appeal 2020-005411
Application 14/542,112

2
CLAIMED SUBJECT MATTER
Claim 1, 10, and 16 are independent. Representative claim 1 recites:
1. A method implemented by a feature discovery
module implemented at least partially in hardware of one or
more computing devices, the method comprising:
generating a random projection matrix that is usable to
compress a dataset representing a plurality of features
associated with one or more customers;
creating a simulator to model customer behavior based
on a first subset of the plurality of features;
training a policy to determine which advertisements to
present to a new customer based on a second subset of the
plurality of features, the first subset representing a first group of
multiple customers and the second subset representing a second
group of multiple customers that is different than the first
group, the policy being trained by at least using the random
projection matrix to compress the second subset of the plurality
of features including removing at least one feature from the
second subset, in which:
an additional feature in the second subset includes
information that matches information in the at least one
feature; and
the information of the at least one feature is
represented by one or more data points than the
information of the additional feature; and evaluating a
level of performance of the policy using the simulator.

Appeal Br. 59 (Claims App.).
REJECTION3
Claims 1–18, 20, and 21 are rejected under 35 U.S.C. § 101 as
directed to a judicial exception without significantly more.

3 The Examiner withdrew the rejections of claims 1–18, 20, and 21 under
35 U.S.C. § 103. See Non-Final Act. 6–26; Ans. 3.
Appeal 2020-005411
Application 14/542,112

3
ANALYSIS
Patent Eligibility of Claims 1–18, 20, and 21
Appellant largely argues the claims as a group. See Appeal Br. 11–
20. We select claim 1 as representative. See 37 C.F.R. § 41.37(c)(1)(iv).
The Examiner analyzes the limitations of claim 1 and determines that
they recite certain methods of organizing human activity to manage personal
behavior, relationships, or interactions between people (e.g., social activities,
teaching, and following rules or instructions) as well as commercial or legal
interactions (including agreements in the form of contracts, legal obligations,
advertising, marketing, or sales activities or behaviors, business relations).
Non-Final Act. 3–4. The Examiner also determines the claims recite mental
processes (including observation, evaluation, judgment, opinion) because a
human can present advertisements based on behavior or interests. Id. at 4.
The Examiner determines that the additional element of “hardware of
one or more computing devices” is recited at a high level of generality that
amounts to no more than adding the words “apply it” to implement a judicial
exception using the computer as a tool, which does not integrate the abstract
idea into a practical application. Id. The Examiner determines “compress a
dataset” is recited without any technical features as extra-solution activity.
Id. at 4–5. Regarding Step 2B, the Examiner determines that the additional
elements merely apply the abstract idea of determining an advertisement on
the system and presenting the results of such abstract processes of analyzing
information, without more, such as using a particular tool for determination
and presentation, is abstract as an ancillary part of the abstract determination
and presentation. Id. at 5.
Appeal 2020-005411
Application 14/542,112

4
Principles of Law
Section 101 of the Patent Act states:
Whoever invents or discovers any new and useful
process, machine, manufacture, or composition of matter, or
any new and useful improvement thereof, may obtain a patent
therefor, subject to the conditions and requirements of this title.
35 U.S.C. § 101. This provision contains an implicit exception that laws of
nature, natural phenomena, and abstract ideas are not patentable subject
matter. See Alice Corp. v. CLS Bank Int’l, 573 U.S. 208, 216 (2014).
To distinguish patents that claim laws of nature, natural phenomena,
and abstract ideas from those that claim patent-eligible applications, we first
determine whether the claims are directed to a patent-ineligible concept. Id.
at 217. If they are, we consider the elements of each claim, individually and
as an ordered combination, to determine if the additional elements transform
the nature of the claim into a patent-eligible application by providing an
“inventive concept” sufficient to ensure the claims amount to significantly
more than a patent on the ineligible concept itself. Id. at 217–18.
The USPTO has issued guidance about this framework. 2019 Revised
Patent Subject Matter Eligibility Guidance, 84 Fed. Reg. 50 (Jan. 7, 2019)
(“Revised Guidance”). Under the Revised Guidance, to determine whether a
claim is “directed to” an abstract idea, we evaluate whether the claim recites:
(1) any judicial exceptions, including certain groupings of abstract ideas
listed in the Revised Guidance (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)
(9th ed. Rev. 10.2019 June 2020) (“MPEP”)). Id. at 52–55.
Appeal 2020-005411
Application 14/542,112

5
Only if a claim (1) recites a judicial exception and also (2) does not
integrate that exception into a practical application, do we then consider
whether the claim either (3) adds a specific limitation beyond a judicial
exception that is not “well-understood, routine, conventional” in the field
(see MPEP § 2106.05(d)) or (4) simply appends well-understood, routine,
conventional activities previously known to the industry, specified at a high
level of generality, to the judicial exception. Id. at 56.
Step 1: Are the Claims Within a Statutory Category?
Claim 1 recites a method, which is a statutory category of invention
under 35 U.S.C. § 101, namely, a process. Non-Final Act. 3; Appeal Br. 11.
Step 2A, Prong One: Do the Claims Recite a Judicial Exception?
We agree that claim 1 recites certain methods of organizing human
activity and mental processes. Non-Final Act. 3–4; Ans. 3–4. The Revised
Guidance enumerates this abstract idea as certain methods of organizing
human activity––fundamental economic principles or practices, commercial
or legal interactions such as advertising, marketing, and sales activities or
behaviors, and managing personal behavior, relationships, and interactions
between people and mental processes of observation, evaluation, judgment,
and opinion. See Revised Guidance, 84 Fed. Reg. at 52.
The claims relate to personalized marketing wherein companies target
website visitors with personalized content with a high probability to induce
the visitors to buy products. Spec. ¶ 1. An optimization algorithm analyzes
past customer sales interactions to model customers using a large number of
features that are mapped to optimal actions of customized offers or content
to present to users. Id. The method automates feature selection that experts
conventionally perform to identify smaller sets of features. Id. ¶¶ 2, 3, 11.
Appeal 2020-005411
Application 14/542,112

6
The preamble of claim 1 recites this purpose as “a feature discovery
module implemented at least partially in hardware of one or more computing
devices.” Appeal Br. 59 (Claims App.). The method recites techniques for
automatic discovery of high-performance features used to model customers
and generate policies that can map the customer features to optimal actions
designed to optimize the effectiveness of targeted marketing strategies that
target advertising to customers visiting a website. See Spec. ¶¶ 3, 11, 12.
Targeting personalized advertising to website visitors to optimize the
likelihood they will make a purchase is a fundamental practice. See Affinity
Labs of Tex., LLC v. Amazon.com Inc., 838 F.3d 1266, 1271 (Fed. Cir. 2016)
(providing customized information and targeted advertising to users based
on user demographic information is a fundamental practice); Intellectual
Ventures I LLC v. Capital One Bank (USA), 792 F.3d 1363, 1369–70 (Fed.
Cir. 2015) (tailoring advertising content based on information known about a
user, e.g., where a user lives or the time of day a user views content, is a
fundamental practice even when performed on the Internet); In re Morsa,
809 F. App’x 913, 917 (Fed. Cir. 2020) (targeting advertising to users based
on user demographic and psychographic information is an abstract idea).
The method collects and analyzes past interactions of customers who
purchased products on websites to identify features and generate policies to
optimize sales and marketing activities to consumers at the website. See id.
¶¶ 1, 3, 12, 13, 21. Customer “features” include personal information such
as a location, age, income, height, marital status, education level, hair color,
interests, hobbies, home ownership status, and commercial activities such as
logins to a website, length of visits, regularity of visits, and a variety of other
information associated with a customer. Id. ¶ 15.
Appeal 2020-005411
Application 14/542,112

7
The first limitation of “generating a random projection matrix that is
usable to compress a dataset representing a plurality of features associated
with one or more customers” (Appeal Br. 59) recites this abstract idea. A
“random projection” is “a matrix, array, or other data structure having a set
of random numbers independently drawn from a probability density
function” such as a Gaussian distribution with a mean of zero. Spec. ¶ 16.
It collects and organizes features that describe customers. See id. ¶ 25.
Organizing customer features into a random projection matrix is an
abstraction. See Elec. Power Grp., LLC v. Alstom S.A., 830 F.3d 1350, 1355
(Fed. Cir. 2016) (“But merely selecting information, by content or source,
for collection, analysis, and display does nothing significant to differentiate
a process from ordinary mental processes, whose implicit exclusion from
§ 101 undergirds the information-based category of abstract ideas.”); In re
TLI Commc’ns LLC Patent Litig., 823 F.3d 607, 613 (Fed. Cir. 2016)
(“[T]he claims, as noted, are simply directed to the abstract idea of classify-
ing and storing digital images in an organized manner. . . . [W]e have
applied the ‘abstract idea’ exception to encompass inventions pertaining to
[such] methods of organizing human activity.”); Content Extraction and
Transmission LLC v. Wells Fargo Bank, N.A., 776 F.3d 1343, 1347 (Fed.
Cir. 2014) (collecting data, recognizing certain data in the collected data,
and storing recognized data recite mental steps that humans perform such as
banks reviewing checks, recognizing data in checks (e.g., amount, account
number, account holder), and storing data); Berkheimer v. HP Inc., 881 F.3d
1360, 1367 (Fed. Cir. 2018) (parsing to extract parts that are reassembled
into composite files is similar to the collecting and recognizing of Content
Extraction and the classifying in an organized manner of TLI).
Appeal 2020-005411
Application 14/542,112

8
The Specification confirms that “a random projection matrix” is a data
structure used to organize customer features at a high level of generality
without technical details to take it outside the abstract realm. Spec. ¶ 25. It
organizes activities of customers engaging in commercial interactions on the
Internet by compressing (reducing) a large number of features into a dataset
with less features without technical details of the compression process. Id.
The Specification indicates a random projection matrix can represent
a baseline set of values determined by an expert. Id. ¶ 16. Thus, generating
a random projection matrix replicates the mental processes that humans can
and do perform. See id. ¶¶ 2 (experts can design sets of feature), 11 (same);
CyberSource Corp. v. Retail Decisions, Inc., 654 F.3d 1366, 1372–73 (Fed.
Cir. 2011) (collecting Internet credit card transaction data and making a map
(similar to a matrix) of credit card numbers used at each IP address can be
performed entirely in the human mind as can using the map to detect credit
card fraud by identifying different card transactions at the same IP address).
As claimed, a random projection matrix is a basic data structure that
does not take claim 1 outside the abstract realm or impart patent eligibility.
See BSG Tech LLC v. BuySeasons, Inc., 899 F.3d 1281, 1287 (Fed. Cir.
2018) (“[T]he recitation of a database structure slightly more detailed than a
generic database does not save the asserted claims at step one.”); Digitech
Image Techs., LLC v. Elecs. for Imaging, Inc., 758 F.3d 1344, 1350 (Fed.
Cir. 2014) (organizing information through mathematical correlations not
tied to a specific structure or machine is abstract). Essentially, it indexes
historical customer activity and personal information to identify optimal
features for advertising. As such, it is part of the abstract idea. See BSG,
899 F.3d at 1286 (fundamental practice or method of organizing activity).
Appeal 2020-005411
Application 14/542,112

9
The next step of “creating a simulator to model customer behavior
based on a first subset of the plurality of features” recites the same abstract
idea. See Appeal Br. 59 (Claims App.). The Specification indicates that a
simulator simulates user behavior or user responses to a marketing strategy.
Spec. ¶¶ 45, 47. Thus, it also replicates human mental processes.
As claimed, the simulator models customer behavior based on a first
subset of customer features. No technical details or functions are claimed.
The description of simulator modeling is equally broad. Spec. ¶¶ 13, 22, 30,
34, 43, 45, 47, 50–52. It simulates (replicates) mental processes that experts
perform when they analyze customer features and behaviors to design a set
of features to represent customer purchase behavior. Id. ¶¶ 2, 11, 13, 16, 30.
Similar claims to modeling human activity and mental processes have
been held to recite abstract ideas. See FairWarning IP, LLC v. Iatric Sys.,
Inc., 839 F.3d 1089, 1096–97 (Fed. Cir. 2016) (compiling and combining
disparate data sources to generate a picture of a user’s activity, identity, and
frequency of activity by selecting and applying rules to audit log data recited
ordinary mental processes rather than a technological advance in accessing
and combining disparate information sources); SmartGene, Inc. v. Advanced
Biological Labs., SA, 555 F. App’x 950, 954–55 (Fed. Cir. 2014) (claims to
a set of “expert rules” for evaluating and selecting treatment regimes from a
knowledge base modeled mental processes that doctors routinely perform of
comparing new and stored information and using rules to identify medical
options); CyberSource, 654 F.3d at 1372–73 (making a map (model) of
Internet credit card numbers and transactions from particular IP addresses
and using the map to determine if a particular transaction is valid can be
performed entirely in the human mind including the logical reasoning).
Appeal 2020-005411
Application 14/542,112

10
Nor does creating a simulator to model customer activity take claim 1
out of the abstract realm when recited at this level of generality. See In re
Jobin, 811 F. App’x 633, 637 (Fed. Cir. 2020) (data structure model used to
collect, organize, group, and store data by techniques such as conducting a
survey or crowdsourcing recited a method of organizing human activity not
an improvement to computer technology); In re Downing, 754 F. App’x 988,
990, 993 (Fed. Cir. 2018) (integrated planning model recited the concept of
personal management, resource planning, and forecasting by collecting and
analyzing business/non-business information without inventive technology);
see also Simio, LLC v. FlexSim Software Prods., Inc., 983 F.3d 1353, 1359–
60 (Fed. Cir. 2020) (making simulation models by graphical modeling rather
than programming was abstract). Here, claim 1 recites no technical details.
The next step of
training a policy to determine which advertisement to present to
a new customer based on a second subset of the plurality of
features . . . the policy being trained by at least using the
random projection matrix to compress the second subset of the
plurality of features including removing at least one feature
from the second subset
involves the abstract idea identified above. Appeal Br. 59 (Claims App.).
A policy is trained to discover high-performance features to optimize
the effectiveness of personalized marketing strategies. Spec. ¶¶ 12, 25. It is
a compressed set of features that describe customers. It is used to generate a
strategy to determine advertisements to show to customer to encourage them
to purchase products. Id. ¶¶ 22–25, 33, 44. This description of training a
policy to generate a strategy confirms the abstract nature of this feature. As
claimed, it is trained by using the random projection matrix to compress a
subset of features without any technical details recited for this process.
Appeal 2020-005411
Application 14/542,112

11
Claims to optimization of commercial interactions and product prices
via a product survey (similar to a model) were abstract in OIP. OIP Techs.,
Inc. v. Amazon.com, Inc., 788 F.3d 1359, 1362 (Fed. Cir. 2015) (automatic
pricing method for electronic commerce involved the concept of offer-based
price optimization, which is a fundamental economic concept similar to the
concepts held to be abstract ideas in Alice and other cases). The price testing
method presented different prices to potential customers, analyzed the data
using statistics to automatically determine, based on the statistics, estimated
outcomes of using each price for the product, and selected a price at which
to sell the product based on this training model to make offers to customers
at the optimum price. Id. at 1361.
Here, claim 1 simply generates a random projection matrix, which is
based on surveys of customer interactions on websites (Spec. ¶ 21), and uses
it to train a policy to determine which advertisements to present to customers
to optimize sales (id. ¶ 44) without any technical details. A “policy can be
used to select automobile advertisements for customers between the ages of
20–30, collect prep courses for lower-income customers, dolls for customers
with young daughters, and so on.” Id. ¶ 33. Thus, it represents the abstract
idea identified above. See Bridge & Post, Inc. v. Verizon Commc’ns, Inc.,
778 F. App’x 882, 887 (Fed. Cir. 2019) (targeted marketing that serves
tailored information to users based on provided data is an abstract idea).
Training a policy by using a random projection matrix to remove at
least one feature also recites a mental process. As an example, a customer’s
location represented by two data points (latitude, longitude) can be reduced
to a single data point (zip code). Spec. ¶ 26. Or, if a customer’s income is
not relevant to a purchase decision, it can be removed. Id. ¶ 27.
Appeal 2020-005411
Application 14/542,112

12
Although the Specification provides examples of formulas that can
compress customer features (Spec. ¶¶ 55–56), claim 1 does not recite any
formulas or particulars of the compression process. Thus, as claimed, the
compression recites a mental process. See Elec. Power, 830 F. 3d at 1354
(“[A]nalyzing information by steps people go through in their minds, or by
mathematical algorithms, without more, [are] essentially mental processes
within the abstract-idea category.”). Experts remove features to generate
subsets of customer features as a mental process. See Spec. ¶¶ 2, 11, 16.
The final limitations recite the compression. An additional feature in
the second subset of compressed data includes information (e.g., a zip code)
that matches information in at least one feature removed from the second
subset and represented by one or more data points (e.g., latitude, longitude
of customer’s location) other than the information of the additional feature
(e.g., zip code). Spec. ¶ 26. Using a generic policy trained by a random
projection matrix by compressing (reducing) a dataset recites the abstract
idea identified above as the Specification confirms. Id. ¶¶ 16, 25–27, 44.
The generic description of compression in the Specification confirms
its abstract nature. “By using a reduced number of features describing the
customer, policies can be trained faster and possibly more efficiently than
traditional techniques.” Id. ¶ 49. Features can be combined, removed, or
modified to reduce a dataset and simplify data used to describe customers.
Id. ¶¶ 28, 32, 49. No technical details are recited for the compression. Thus,
the compression can be performed as a mental process as discussed above.
See Revised Guidance, 84 Fed. Reg. at 52.
Accordingly, we determine claim 1 recites the abstract idea identified
above.
Appeal 2020-005411
Application 14/542,112

13
Step 2A, Prong Two: Integration into a Practical Application
We next consider whether claim 1 recites any additional elements that
integrate the abstract idea into a practical application. Revised Guidance,
84 Fed. Reg. at 54. We determine that claim 1 lacks additional elements that
improve a computer or other technology or implement the abstract idea in
conjunction with a particular machine or manufacture that is integral to the
claim. Nor does it include an additional element that transforms or reduces a
particular article to a different state or thing or applies the abstract idea in a
meaningful way beyond linking it to a particular technological environment.
See Revised Guidance, 84 Fed. Reg. at 55; Non-Final Act. 4–5; Ans. 3–7.
Appellant argues it is “not practical for a human to train a policy to
determine which advertisements to present to a new customer because of the
vast amounts of data needed when training a policy” and training a policy to
determine advertisements to present to a new customer requires a computer.
Appeal Br. 12. Appellant asserts that conventional methods use humans to
identify advertisements to present to customers but “considering the sheer
amount of data used to represent customers, these conventional techniques
may be error-prone and time-consuming.” Id. at 12–13 (citing Spec. ¶ 11).
Claim 1 does not train a policy with vast amounts of data. It trains a
policy by compressing a sub-set of features by removing at least one feature.
Automating an abstract idea does not transform it. See Cellspin Soft, Inc. v.
Fitbit, Inc., 927 F.3d 1306, 1316 (Fed. Cir. 2019) (“[T]he need to perform
tasks automatically is not a unique technical problem.”); OIP, 788 F.3d at
1363 (“At best, the claims describe the automation of the fundamental
economic practice of offer-based price optimization through the use of
generic-computer functions[]” to make such methods more efficient).
Appeal 2020-005411
Application 14/542,112

14
Improvements in efficiency or speed resulting from implementing an
abstract idea on generic computers does not make the abstract idea patent
eligible. Credit Acceptance Corp. v. Westlake Servs., 859 F.3d 1044, 1056
(Fed. Cir. 2017) (“But merely ‘configur[ing]’ generic computers in order to
‘supplant and enhance’ an otherwise abstract manual process is precisely the
sort of invention that the Alice Court deemed ineligible for patenting.”)
(citation omitted); Intellectual Ventures, 792 F.3d at 1370 (“[O]ur precedent
is clear that merely adding computer functionality to increase the speed or
efficiency of the process does not confer patent eligibility on an otherwise
abstract idea.”); OIP, 788 F.3d at 1363 (“[R]elying on a computer to perform
routine tasks more quickly or more accurately is insufficient to render a
claim patent eligible.”); Bancorp Servs., L.L.C. v. Sun Life Assur. Co. of
Can. (U.S.), 687 F.3d 1266, 1279 (Fed. Cir. 2012) (“[T]he computer merely
permits one to manage a stable value protected life insurance policy more
efficiently than one could to mentally. Using a computer to accelerate an
ineligible mental process does not make that process patent-eligible.”); see
Credit Acceptance, 859 F.3d at 1055 (automation of manual processes using
generic computers is not a patentable improvement in computer technology).
Appellant also asserts it is not possible for conventional computers to
process large amounts of training data so compressing the second subset to a
smaller data set to train a policy decreases processing time and increases the
effectiveness of the learning algorithms to process information faster and
train a policy more quickly than conventional methods. Appeal Br. 13–14.
This argument is not commensurate with claim 1, which does not
recite training data sizes or methods of training a policy beyond compressing
it by removing at least one feature from the second subset. Appeal Br. 59.
Appeal 2020-005411
Application 14/542,112

15
The Specification’s description of the process of training a policy is
equally broad thus confirming the broad scope of claim 1 as encompassing
mental steps that experts perform when they identify small features subsets
to optimize sales. See Spec. ¶¶ 2, 11, 16, 31, 36, 44.
Claim 1 only trains “a policy.” However, the Specification indicates
that a variety of policies should be generated and evaluated to identify top-
performing policies, and only the top-performing policies are provided to the
learning algorithm to optimize its effectiveness in identifying advertisements
presented to users. Spec. ¶ 23. Claim 1 does not train a variety of policies.
Nor does it identify a top-performing policy.
Because a policy is trained using a random subset of the dataset and a
random projection (random selection of customer features), “it is unknown
whether the policy is an effective policy.” Spec. ¶ 44 (emphasis added). As
a result, “the policy should be tested to determine its effectiveness and/or
efficiency.” Id. However, claim 1 does not test a policy after training it.
Policies can be scored based on the number of simulated users that
purchase a product based on an advertisement selected by the policy. The
more simulated customers that purchase a product based on advertisements
defined by the policy, the more effective the policy. Id. ¶ 51. Claim 1 does
not score a policy to determine its effectiveness.
Removing features does not necessarily improve efficiency as argued.
The Specification indicates that removing irrelevant features “can increase
processing times of the learning algorithm by providing fewer inputs to the
learning algorithm.” Spec. ¶ 27 (emphasis added). Even if a reduced set of
features can train a policy faster, the Specification indicates that a smaller set
is possibly more efficiently than traditional techniques. See id. ¶ 49.
Appeal 2020-005411
Application 14/542,112

16
As claimed, the method lacks features described in the Specification
as improving efficiency and accuracy. See Appeal Br. 12–16; Spec. ¶¶ 23,
27, 44, 49, 51. Even so, any improvements are to the abstract idea rather
than to computers or other technology.
The random projection matrix used to compress (reduce) datasets to
improve efficiency and accuracy of the method, “can represent a baseline set
of values that were determined by an expert.” Id. ¶ 16 (emphasis added).
This description of the random projection matrix used to compress datasets
to achieve the alleged efficiencies confirms that claim 1 embodies mental
processes rather than improvements to computers or other technologies.
The concept of using a small subset of features for behavior analysis
is known in the art. “Conventional policy learning algorithms typically
utilize a relatively small set of carefully constructed features so that optimal
actions can be easily and quickly identified.” Spec. ¶ 2. An expert manually
designs a set of features to train a learning algorithm. Id. ¶¶ 2, 11.
Other aspects of the claimed method asserted to be improvements to
the functioning of computer are features of the abstract idea identified under
Prong One. Appeal Br. 14–16. Thus, they cannot serve as “additional
elements” that integrate the abstract idea into a practical application. See
Revised Guidance, 84 Fed. Reg. at 55 n.24 (“USPTO guidance uses the term
‘additional elements’ to refer to claim features, limitations, and/or steps that
are recited in the claim beyond the identified judicial exception.”); Alice,
573 U.S. at 221 (holding that a claim to an abstract idea must include
“additional features” to ensure it does not monopolize the abstract idea)
(citing Mayo Collaborative Servs. v. Prometheus Labs., Inc., 566 U.S. 66, 77
(2012)).
Appeal 2020-005411
Application 14/542,112

17
Appellant also contends that using the random projection matrix to
compress the second subset of features transforms a particular article to a
different state or thing. Appeal Br. 15. This argument fails because it also
relies on a feature of the abstract idea to serve as an additional element for
integrating that abstract idea into a practical application. Even so, using a
mathematical algorithm to manipulate existing information to generate
additional information is not sufficient for a transformation. See Digitech
Image, 758 F.3d at 1351. Here, the random projection matrix reduces a
dataset of features in some undefined way without any transformation.
Combining one abstract idea (certain methods of organizing human
activity) with another abstract idea (mental processes) does not integrate an
abstract idea. See RecogniCorp, LLC v. Nintendo Co., 855 F.3d 1322, 1327
(Fed. Cir. 2017) (“Adding one abstract idea (math) to another abstract idea
(encoding and decoding) does not render the claim non-abstract.”); Apple,
Inc. v. Ameranth, Inc., 842 F.3d 1229, 1240 (Fed. Cir. 2016) (“An abstract
idea can generally be described at different levels of abstraction.”);
Synopsys, Inc. v. Mentor Graphics Corp., 839 F.3d 1138, 1151 (Fed. Cir.
2016) (“[A] claim for a new abstract idea is still an abstract idea.”).
The additional element of hardware of one or more computing devices
recites a generic computer that performs generic functions as a tool. It may
be a server, a client device, an on-chip system, or other suitable computing
device or system. Spec. ¶ 57. It may include such hardware as processors,
computer-readable memory or a storage medium, I/O interfaces, and input
devices. Id. ¶¶ 57–61, Fig. 5. As such, it is insufficient for an integration.
See Alice, 573 U.S. at 223 (the mere recitation of a generic computer cannot
transform a patent-ineligible abstract idea into a patent-eligible invention).
Appeal 2020-005411
Application 14/542,112

18
Using a random projection matrix of random values (Spec. ¶¶ 16, 25,
42) to compress (reduce) a feature dataset without reciting any technical
details of the process does not improve computers or tie the method to a
particular machine integral to the claim. See Adaptive Streaming Inc. v.
Netflix, Inc., No. 2020-1310, 2020 WL 7334126, *3 (Fed. Cir. Dec. 14,
2020) (claims “do not incorporate anything more that would suffice to
transform their subject matter into an eligible application of the abstract
idea” when “there is no identification in the claims or written description of
specific, unconventional encoding, decoding, compression, or broadcasting
techniques.”); Voit Techs, LLC v. Del-Ton, Inc., 757 F. App’x 1000, 1002,
1003 (Fed. Cir. 2019) (“The Asserted Claims are directed to the abstract idea
of entering, transmitting, locating, compressing, storing, and displaying data
. . . to facilitate the buying and selling of items.”; “Voit fails to explain how
employing different formats, as claimed, improves compression techniques
or the functioning of the computer. Instead, the specification demonstrates
that the Asserted Claims are directed to use of generic computer components
performing conventional compression techniques to carry out the claimed
invention.”); Bascom Global Internet Servs., Inc. v. AT&T Mobility LLC,
827 F.3d 1341, 1348 (Fed. Cir. 2016) (“We agree with the district court that
filtering content is an abstract idea because it is a longstanding, well-known
method of organizing human behavior, similar to concepts previously found
to be abstract.”). Here, claim 1 recites no technical details of compressing a
dataset of plural features with the random projection matrix or compressing
the second subset of the plurality of features by removing at least one feature
therefrom. The Specification’s equally broad description of compressing
data confirms the abstract nature of the claim. See Spec. ¶¶ 25–28, 42–44.
Appeal 2020-005411
Application 14/542,112

19
As the court held in Voit, “[g]eneral statements of ‘advanced image
data compression’ or faster communications will not suffice where it is
unclear how the different compression format claim limitations actually
achieve the alleged improvements.” Voit, 757 F. App’x at 1004.
Accordingly, we determine that claim 1 lacks additional elements that
are sufficient to integrate the abstract idea into a practical application.
Step 2B: Do the Claims Include an Inventive Concept?
We next consider whether claim 1 recites any additional elements,
individually or as an ordered combination, to provide an inventive concept.
Alice, 573 U.S. at 217–18. This step is satisfied when the claim limitations
involve more than well-understood, routine, and conventional activities that
are known in industry. Berkheimer, 881 F.3d at 1367; Revised Guidance,
84 Fed. Reg. at 56 (the second step of the Alice analysis considers if a claim
adds a specific limitation beyond the recited judicial exception that also is
not “well-understood, routine, conventional” activity in the field).
An invention’s use of the ineligible concept to which it is directed
cannot supply the inventive concept that renders it “significantly more” than
the ineligible concept. BSG, 899 F.3d at 1290; SAP Am., Inc. v. InvestPic,
LLC, 898 F.3d 1161, 1163 (Fed. Cir. 2019) (“The claims here are ineligible
because their innovation is an innovation in ineligible subject matter. Their
subject is nothing but a series of mathematical calculations based on selected
information and the presentation of the results of those calculations (in the
plot of a probability distribution function). No matter how much of an
advance in the finance field the claims recite, the advance lies entirely in the
realm of abstract ideas, with no plausibly alleged innovation in the non-
abstract realm. An advance of that nature is ineligible for patenting.”).
Appeal 2020-005411
Application 14/542,112

20
Individually, the additional element of generic computing devices
does not provide an inventive concept. Alice, 573 U.S. at 222 (“[S]imply
implementing a mathematical principle on a physical machine, namely a
computer, [i]s not a patentable application of that principle.”) (citation
omitted); TLI, 823 F.3d at 615 (a “control unit” that predictably controls
image resolution using known image compression techniques and controls
transmission rates in vague terms does not transform the abstract idea).
As an ordered combination, the limitation recites no more than it does
individually. See BSG, 899 F.3d at 1290–91 (“If a claim’s only ‘inventive
concept’ is the application of an abstract idea using conventional and well-
understood techniques, the claim has not been transformed into a patent-
eligible application of an abstract idea.”); SAP, 898 F.3d at 1169 (“[T]his
court has ruled many times that ‘such invocations of computers . . . that are
not even arguably inventive are insufficient to pass the test of an inventive
concept in the application of an abstract idea.’”) (citation omitted); see also
Bancorp, 687 F.3d at 1280 (“The district court correctly held that without
the computer limitations nothing remains in the claims but the abstract idea
of managing a stable value protected life insurance policy by performing
calculations and manipulating the results.”). As discussed under Prong Two,
the claimed computing devices and hardware are described as conventional.
See Spec. ¶¶ 57–62, Fig. 5.
Appellant’s reliance on Amdocs (Israel) Ltd. v. Openet Telecom, Inc.,
841 F.3d 1288 (Fed. Cir. 2016) is misplaced. Appeal Br. 16–18; Reply Br.
6–7. An inventive concept was found in the unique distributed architecture
that allowed the system to collect network usage information in a way that
enabled load distribution and minimized impact on the network. Id. at 1303.
Appeal 2020-005411
Application 14/542,112

21
In Amdocs, the claims recited distributed filtering and aggregation that
eliminated system capacity bottlenecks by allowing granular data to reside in
the system peripheries close to the information sources. Id. This distributed
architecture avoided or reduced network congestion and bottlenecks while
allowing data to be accessed from a central location. Id. Gatherers provided
distributed filtering and aggregation that improved scalability and system
efficiency by reducing a volume of data sent to a central event manager. Id.
Here, in contrast, claim 1 recites generic computing devices rather
than an innovative network architecture or distributed processing function.
Any reduction in data transmission results from abstract compression that
replicates conventional mental processes of experts selecting a smaller set of
customer features so optimal actions can be easily and quickly identified.
Spec. ¶¶ 2, 11, 16. No distributed network processing is claimed. Nor is an
improvement made to network load distribution. See Tenstreet, LLC v.
DriverReach, LLC, 826 F. App’x 925, (Fed. Cir. 2020) (“The test for patent-
eligible subject matter is not whether the claims are advantageous over the
previous method. Even if the ’575 patent provides advantages over manual
collection of data, the patent claims no technological improvement beyond
the use of a generic computer network.”); Bancorp, 687 F.3d at 1278, 1279
(“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.”;
“Using a computer to accelerate an ineligible mental process does not make
that process patent-eligible.”). Because claim 1 does not involve adaptive
monitoring of network traffic, as in Example 40 of the 2019 PEG Examples,
Example 40 illustrates why claim 1 here is not patent eligible. Reply Br. 4.
Appeal 2020-005411
Application 14/542,112

22
Accordingly, we determine claim 1 lacks an inventive concept to
transform the abstract idea into patent eligible subject matter. We sustain
the rejection of claim 1 and claims 2–9, which fall therewith.
Appellant argues that independent claims 10 and 16 are patent-eligible
because they do not recite an abstract idea, and they provide an integration
and inventive concept for the same reasons as claim 1. Appeal Br. 18–20.
These arguments are not persuasive for the reasons discussed above for
claim 1. Accordingly, we sustain the rejection of claim 10 and 16 and their
respective dependent claims 11–15, 17, 18, 20, and 21, which are not argued
separately by Appellant. See id. at 20; see also 37 C.F.R. § 41.37(c)(1)(iv).
CONCLUSION
In summary:
Claims
Rejected
35 U.S.C.
§
Reference(s)/
Basis
Affirmed Reversed
1–18, 20, 21 101 Eligibility 1–18, 20, 21
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)(1)(iv).
AFFIRMED