Adobe Inc.

Patent Trials and Appeals Board

Sep 23, 2021

2021001163 (P.T.A.B. Sep. 23, 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.
15/783,228 10/13/2017 Jen-Chan Jeff Chien P7176-US 6629
108982 7590 09/23/2021
FIG. 1 Patents
116 W. Pacific Avenue
Suite 200
Spokane, WA 99201
EXAMINER
PATEL, DIPEN M
ART UNIT PAPER NUMBER
3688
NOTIFICATION DATE DELIVERY MODE
09/23/2021 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 JEN-CHAN JEFF CHIEN,
THOMAS WILLIAM RANDALL JACOBS,
KENT ANDREW EDMONDS, KEVIN GARY SMITH,
PETER RAYMOND FRANSEN,
GAVIN STUART PETER MILLER, and
ASHLEY MANNING STILL
__________________

Appeal 2021-001163
Application 15/783,228
Technology Center 3600
____________________

Before ANTON W. FETTING, JAMES P. CALVE, and
ROBERT J. SILVERMAN, 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 1–12, 14, 15, and 21–26, which are all of
the pending claims.2 See Non-Final Act. 2; Appeal Br. 3. 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. 3.
2 Claims 13 and 16–20 are cancelled. Appeal Br. 45 (Claims App.).
Appeal 2021-001163
Application 15/783,228

2
CLAIMED SUBJECT MATTER
Claims 1, 12, and 21 are independent. Claim 1 recites:
1. In a digital medium environment to customize a
time at which digital marketing content is output in relation to a
digital video, a method implemented by at least one computing
device, the method comprising:
receiving, by the at least one computing device, a model
trained using machine learning based on training data, the
training data describing:
user segments of a user population;
user interaction of respective said segments with
training digital marketing content output in conjunction
with a plurality of training digital videos;
an outcome of the user interaction; and
a time at which the training digital marketing
content is output in relation to respective training digital
videos of the plurality of training digital videos;
receiving, by the at least one computing device, the
digital video;
generating, by the at least one computing device, a
suggestion by processing the digital video and data describing a
corresponding said user segment of a subsequent user based on
the model using machine learning, the suggestion specifying a
time and an item of digital marketing content is to be output to
the subsequent user at the time in relation to the output of the
digital video; and
controlling, by the at least one computing device, output
of the item of digital marketing content at the time in relation to
the digital video based on the suggestion.
REJECTIONS
Claims 1–12, 14, 15, and 21–26 are rejected under 35 U.S.C. § 101 as
directed to a judicial exception without significantly more.
Claims 1–11 and 21–25 are rejected under 35 U.S.C. § 102(a)(2) as
anticipated by Bakshi (US 9,736,503 B1, issued August 15, 2017).
Appeal 2021-001163
Application 15/783,228

3
Claims 12, 14, 15, and 26 are rejected under 35 U.S.C. § 103 as
unpatentable over Bakshi and McIntire (US 2007/0250901 A1, published
October 25, 2007).
ANALYSIS
Eligibility of Claims 1–12, 14, 15, and 21–26
The Examiner determines that the claims recite certain methods of
organizing human activity as fundamental economic principles or practices,
commercial or legal interactions, advertising, marketing, or sales activities
or behaviors by collecting and analyzing data, creating an index to search for
and retrieve data, and filtering and tailoring content. Non-Final Act. 2–4.
The Examiner determines the claims recite steps of processing information
through mathematical relations, formulas, and calculations by applying a
model trained by using machine learning to determine an optimal timing for
presenting marketing content. Id. at 5.
The Examiner determines that the additional element of a computing
device that can process data in real time is used as a tool to implement the
abstract idea and apply it without integrating the abstract idea into a practical
application or transforming it into a patent eligible application. Non-Final
Act. 5–8. The Examiner determines that the additional element is described
generically at a high level of generality and appears to automate mental tasks
of selecting and inserting marketing content at the appropriate time through
processes that are well-understood, routine, and conventional activities such
as receiving or transmitting data over a network, performing repetitive
calculations, and organizing and manipulating information through
mathematical correlations without significantly more. Id. at 8–10.
Appeal 2021-001163
Application 15/783,228

4
Principles of Law
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. Laws of nature, natural phenomena, and abstract ideas are
not patentable. 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 claim elements, individually and as an
ordered combination, to determine if any additional elements provide an
inventive concept sufficient to ensure that the claims in practice amount to
significantly more than a patent on the ineligible concept. 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”). To determine if 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.3 Id. at 52–55.

3 “A claim that integrates a judicial exception into a practical application
will apply, rely on, or use the judicial exception in a manner that imposes a
meaningful limit on the judicial exception, such that the claim is more than a
drafting effort designed to monopolize the judicial exception.” Revised
Guidance, 84 Fed. Reg. at 54.
Appeal 2021-001163
Application 15/783,228

5
If a claim (1) recites a judicial exception and (2) does not integrate
that exception into a practical application, we consider whether the claim
(3) provides an inventive concept such as by adding a limitation beyond a
judicial exception that is not “well-understood, routine, conventional” in the
field or (4) appends well-understood, routine, conventional activities
previously known to the industry, specified at a high level of generality, to
the judicial exception. Revised Guidance, 84 Fed. Reg. at 56.
Step 1
We agree with the Examiner that claims 1–11, 12, 14, 15, and 26
recite a method (process) and claims 21–25 recite a system (machine) that
are statutory categories. See Non-Final Act. 3; 35 U.S.C. § 101.
Alice Step One / Revised Guidance Step 2A, Prong One:
Do the Claims Recite a Judicial Exception?
We agree with the Examiner that the claims recite certain methods of
organizing human activity of commercial or legal interactions, advertising,
marketing, or sales activities or behaviors, and the model is trained through
machine learning that involves mathematical concepts. Non-Final Act. 3–5;
Ans. 5–7; Revised Guidance, 84 Fed. Reg. at 52.
The application’s title CUSTOMIZED PLACEMENT OF DIGITAL
MARKETING CONTENT IN A DIGITAL VIDEO indicates the purpose of
the invention is customizing placement of digital marketing in digital videos.
Customized placement increases the likelihood of advertisement conversion,
i.e., a viewer will click on an advertisement and purchase a good or service.
Spec. ¶¶ 16, 20. Training data of user interactions with marketing content is
used to train a model to suggest marketing content to output with a video
and when to do so to lead to a conversion of a good or service. Id. ¶¶ 20, 21.
Appeal 2021-001163
Application 15/783,228

6
The preamble of claim 1 recites this purpose as “a digital medium
environment to customize a time at which digital marketing content is output
in relation to a digital video.” Appeal Br. 41 (Claims App.). The final steps
fulfill this purpose by generating a suggestion “specifying a time and an item
of digital marketing content is to be output to the subsequent user at the time
in relation to the output of the digital video” and “controlling, by the at least
one computing device, output of the item of digital marketing content at the
time in relation to the digital video based on the suggestion.” Id.
Providing customized marketing/advertising targeted to a user based
on user demographics, time of day, user preferences, and other data recites a
fundamental economic practice and a method of organizing human activity.
See Affinity Labs of Tex., LLC v. Amazon.com Inc., 838 F.3d 1266, 1269–70
(Fed. Cir. 2016) (delivering user selected media content to portable devices
is a fundamental economic concept); Intellectual Ventures I LLC v. Capital
One Bank (USA), 792 F.3d 1363, 1369–70 (Fed. Cir. 2015) (providing a user
with tailored information like advertisements based on information known
about the user such as a location, address, or personal characteristics and a
time of day is a fundamental practice long prevalent in our system); In re
Morsa, 809 F. App’x 913, 917 (Fed. Cir. 2020) (customizing information
based on demographic and psychographic information known about a user to
match the user to a specific advertiser is a fundamental economic practice of
organizing human activity); Bridge & Post, Inc. v. Verizon Commc’ns, Inc.,
778 F. App’x 882, 886–88 (Fed. Cir. 2019) (using targeted marketing and
segmentation to tailor information is a fundamental practice to increase the
effectiveness of advertisements in radio, television, print media, or Internet
by using a persistent identifier or a user profile of network usage patterns).
Appeal 2021-001163
Application 15/783,228

7
The other limitations of claim 1 involve receiving a model trained
using machine learning based on training data. See Appeal Br. 41 (Claims
App.). The training data describes user segments of a user population, user
interaction with training digital marketing content output with plural training
digital videos, an outcome of the user interaction, and a time at which the
training digital marketing content is output in relation to the training digital
videos. See id.
The generic model trained by generic machine learning organizes
users’ commercial interactions for advertising, marketing, or sales activities
or behaviors, which is a method of organizing human behavior involving the
fundamental economic practice of tailored marketing. When recited at this
level of generality without any technical details to indicate improvements to
computers, software, or other technologies, the claim recites an abstract idea.
User segments represent the demographics of users. Spec. ¶¶ 40, 51.
User demographic information includes age, geographic location, and other
such factors. See id. ¶ 40. As discussed above, training a model to provide
targeted marketing customized to users based on user demographics is an
abstract idea involving a fundamental economic practice long prevalent in
our system of commerce and a method of organizing human activity. E.g.,
Intellectual Ventures, 792 F.3d at 1369–70. Indeed, Bridge and Post held
that targeted marketing using market segmentation to tailor information was
a fundamental practice when used to target advertisements for television
video media and the Internet. Bridge & Post, 778 F. App’x at 886–88.
Measuring user interactions and outcomes with marketing content to
identify content and output times that increase advertisement conversion is
part of the same abstract idea.
Appeal 2021-001163
Application 15/783,228

8
Advertisements can be output during a commercial break in a video or
as a banner during the video. Spec. ¶¶ 34, 41, 42. Even if we read these
features into “specifying a time,” they do not improve computers, machine
learning, or another technology. Television commercial breaks are used to
show advertisements. Banner headlines can be used to display information
during a television broadcast. See Interval Licensing LLC v. AOL, Inc., 896
F.3d 1335, 1345 (Fed. Cir. 2018) (claimed “attention manager” displaying
two information sets on a display in a non-interfering manner was claimed
without any technical details of how that result was produced to make it non-
abstract); Intellectual Ventures I LLC v. Capital One Fin. Corp., 850 F.3d
1332, 1341–42 (Fed. Cir. 2017) (“dynamic document” that presented content
from multiple electronic records was not patent eligible).
Training a model using machine learning to try to predict an optimal
time to output an advertisement during a video to increase the conversion
rate of the advertisements is part of the customized marketing. No technical
details are claimed for the way machine learning trains the model. Indeed,
the training of the model is claimed as a source limitation (discussed below).
In an analogous situation, training a system to choose optimal product
prices to optimize sales recited a fundamental economic practice. See OIP
Techs., Inc. v. Amazon.com, Inc., 788 F.3d 1359, 1361–63 (Fed. Cir. 2015)
(automatic method of pricing products was similar to fundamental economic
concepts where it tested a plurality of prices by sending electronic messages
with product offers at different prices, gathered statistics during the testing
of sales of the product at each price, automatically determined an outcome of
using each price for the product, selected a price based on the estimated
outcome, and output product offers to customers at the selected price).
Appeal 2021-001163
Application 15/783,228

9
In view of the foregoing analysis, we are unpersuaded by Appellant’s
arguments that the claims do not represent certain methods of organizing
human activity or fundamental economic principles. See Appeal Br. 18–20.
The Specification describes the machine learning techniques generally
as “linear regression, logistic regression, decision trees, structured vector
machines, naïve Bayes, K-means, K-nearest neighbor, random forest, neural
networks, and so forth.” Spec. ¶ 52. These techniques involve mathematical
relationships, formulas/equations, or calculations. October 2019 Update:
Subject Matter Eligibility 3–4. Thus, the Examiner correctly determined
that a model trained by machine learning involves mathematical concepts.
Appellant argues that the claims improve computer-related technology
by controlling the output of digital marketing content at a particular time for
digital videos, and this timing technique can be at a commercial break in the
video or a banner shown during the video. Appeal Br. 15–16. As claimed,
controlling the output of marketing content at “a time” is part of the abstract
idea identified above. Specifying a time that can be a commercial break in
the video or a banner displayed during the video is claimed (and described)
without any technical details to indicate computer or software improvements
are used to specify this timing. See Spec. ¶¶ 3, 21, 34, 36, 42, 54.
Appellant argues that Example 39 of the 2019 Patent Eligibility
Guidelines recites an eligible method of training a neural network. Appeal
Br. 17–18. Example 39 uses training data of digital facial images that are
transformed various ways to create a first training set used to train a neural
network. Then, a second training set is created from the first training set and
non-facial images that were incorrectly detected as facial images during the
first stage. 2019 Subject Matter Eligibility Examples: Abstract Ideas, at 8–9.
Appeal 2021-001163
Application 15/783,228

10
Here, the training data is not transformed or curated in any particular
way in order to train the machine learning model. Nor is the model trained
in stages to improve its accuracy as in Example 39. See Ans. 6. As claimed,
the model uses generic machine learning to analyze training data in some
unspecified way without any technical details of that process or any apparent
improvement to computers or other technology. The Specification’s general
description of machine learning as involving general mathematical concepts
confirms the abstract nature of the claimed model and the machine learning.
See Spec. ¶¶ 52–57, 61–65. Nor is there any indication that the method
improves machine learning or the training of models in any technical way.
Thus, we determine that claim 1 recites the abstract idea identified
above. Claims 2–12, 14, 15, and 21–26 are not argued separately and thus
fall with claim 1. Appeal Br. 14–22; see 37 C.F.R. § 41.37(c)(1)(iv) (2019).
Alice Step One
Revised Guidance Step 2A, Prong Two:
Is There an Integration into a Practical Application?
We next consider whether claim 1 recites additional elements that
integrate the abstract idea into a practical application. Revised Guidance,
84 Fed. Reg. at 54. We agree with the Examiner that the additional element
of a computing device is claimed and described at a high level of generality
as a tool used to implement the abstract idea without improving computers
or another technology. Non-Final Act. 5–8; Ans. 8. Nor does performing a
method on a generic computing device apply the abstract idea with, or by
use of, a particular machine, or transform or reduce a particular article to a
different state or thing, or apply the abstract idea in some other meaningful
way to integrate the judicial exception into a practical application. See
Revised Guidance, 84 Fed. Reg. at 55.
Appeal 2021-001163
Application 15/783,228

11
The Specification describes a computing device 802 that includes
processing system 804, computer-readable media 806, one or more I/O
interfaces 808, and a system bus. Spec. ¶ 66. Processing system 804 may
perform operations using hardware such as hardware element 810 that is
configured as processors, functional blocks, and so forth. Id. ¶ 68. Storage
media 806 may be a variety of generic devices. Id. ¶ 69. This description
confirms the generic nature of the claimed “computing device” that is used
to implement the abstract idea without improving computers or technology.
Appellant argues that the claims are rooted in computer technology to
overcome a problem in computer networks. Appeal Br. 22–24. Appellant
asserts that the method is implemented on a computing device to receive a
model trained using machine learning with training data of user segments,
user interaction, outcomes of user interaction, and a time to output marketing
content and receive a digital video, generate a suggestion, and control output
of digital marketing content. Id. at 23–24.
These limitations are features of the abstract idea not additional
elements that can integrate the abstract idea into a practical application. See
Revised Guidance, 84 Fed. Reg. at 55 n.24 (“additional elements” are claim
features, limitations, and/or steps recited in a claim beyond the identified
judicial exception); see Alice, 573 U.S. at 221 (a claim to an abstract idea
must include additional features to ensure it does not monopolize the
abstract idea; Mayo Collaborative Servs. v. Prometheus Labs., Inc., 566 U.S.
66 (2012) held that a transformation into a patent-eligible application
requires more than simply stating the abstract idea with the words “apply
it”); Synopsys, Inc. v. Mentor Graphics Corp., 839 F.3d 1138, 1151 (Fed.
Cir. 2016) (“But, a claim for a new abstract idea is still an abstract idea.”).
Appeal 2021-001163
Application 15/783,228

12
The final step of controlling output of an item of digital marketing
content at the suggested time recites extra-solution activity. See Elec. Power
Grp., LLC v. Alstom S.A., 830 F.3d 1350, 1354 (Fed. Cir. 2016) (presenting
the results of the abstract processes of collecting and analyzing information
is abstract as an ancillary part of the collection and analysis).
Even if we consider these limitations to be additional elements, they
do not improve computers or other technology or transform an article to a
different state or thing sufficient to integrate the abstract idea into a practical
application. They do not recite an improved way to train a model by using
machine learning. They do not purport to improve machine learning using
basic training data. Nor do they recite an improved technical way to provide
customized advertisements to viewers of digital videos. As claimed, the
limitations recite a generic model trained by generic machine learning using
training data to generate a suggestion of a time and item of digital marketing
content to output with a digital video. See Intellectual Ventures, 792 F.3d at
1371 (Fed. Cir. 2015) (“Requiring the use of a ‘software’ ‘brain’ ‘tasked
with tailoring information and providing it to the user’ provides no
additional limitation beyond applying an abstract idea, restricted to the
Internet, on a generic computer.”).
“Information as such is an intangible.” Elec. Power, 830 F.3d at
1353. Indeed, the model is recited as a source limitation in terms of being
trained by machine learning based on training data without actually reciting
any training steps as positive steps of the method in claim 1. See Biogen MA
Inc. v. EMD Serono, Inc., 976 F.3d 1326, 1334 (Fed. Cir. 2020) (nesting a
product-by-process limitation in a method of treatment claim does not alter
its construction as a product-by-process limitation).
Appeal 2021-001163
Application 15/783,228

13
DDR illustrates why claim 1 here is not integrated into a practical
application. DDR Holdings, LLC v. Hotels.com, L.P., 773 F.3d 1245 (Fed.
Cir. 2014). The claims in DDR recited a new web server configuration. If a
website visitor clicked on an advertisement for a merchant’s product on a
host’s website, the visitor was directed to a hybrid web page that combined
the look and feel elements of the host website with product information of
the merchant’s website on a third party outsource provider’s server. Id. at
1257–58. Here, claim 1 recites a model trained by machine learning without
claiming any technical details of that process or improvements to computer
operations. See Alice, 573 U.S. at 223 (applying an abstract idea with a
generic computer implementation does not make the abstract idea patent-
eligible); Customedia Techs., LLC v. Dish Network Corp., 951 F.3d 1359,
1365 (Fed. Cir. 2020) (generic speed and efficiency improvements that are
inherent in applying the use of a computer to any task, at most, improve the
abstract idea by using a computer as a tool); Credit Acceptance Corp. v.
Westlake Servs., 859 F.3d 1044, 1055 (Fed. Cir. 2017) (automating manual
processes by using generic computers is not a patentable improvement to
computer technology); OIP, 788 F.3d at 1363 (“But relying on a computer to
perform routine tasks more quickly or more accurately is insufficient to
render a claim patent eligible.”).
Appellant also asserts that controlling the time at which the marketing
content is presented is a technological improvement. Appeal Br. 24–25.
However, no technical details are recited. Presenting customized marketing
based on times is an abstract idea without more. See Intellectual Ventures,
792 F.3d at 1370 (presenting tailored information to a user based on the time
of day is a fundamental practice long prevalent in our system).
Appeal 2021-001163
Application 15/783,228

14
Here, the Specification indicates marketing is displayed as a banner or
during a commercial break in a video. Spec. ¶¶ 34, 41, 42. No technical
details are claimed to implement this timing. It’s not clear how computers
or other technology are improved by specifying a time to output marketing
content for a video. See Customedia, 951 F.3d at 1365 (software can make
non-abstract improvements to computer technology, but the software must
improve the functionality of a computer or a network platform).
Appellant asserts that specific techniques are described to control the
output of digital marketing content for a video to address the complexities of
digital video over other types of digital content and to provide a specific
solution to a problem arising in computer networks. Appeal Br. 25.
This argument is not persuasive because no such techniques or details
are claimed. See Ericsson Inc. v. TCL Commc’n Tech. Holdings Ltd., 955
F.3d 1317, 1325 (Fed. Cir. 2020) (“While the specification may be ‘helpful
in illuminating what a claim is directed to . . . the specification must always
yield to the claim language’ when identifying the ‘true focus of a claim.’”)
(citation omitted); ChargePoint, Inc. v. SemaConnect, Inc., 920 F.3d 759,
769–70 (Fed. Cir. 2019) (“Even if ChargePoint’s specification had provided,
for example, a technical explanation of how to enable communication over a
network for device interaction . . ., the claim language here would not
require those details. Instead, the broad claim language would cover any
mechanism for implementing network communication.”); Accenture Global
Servs., GmbH v. Guidewire Software, Inc., 728 F.3d 1336, 1345 (Fed. Cir.
2013) (“[T]he complexity of the implementing software or the level of detail
in the specification does not transform a claim reciting only an abstract
concept into a patent-eligible system or method.”).
Appeal 2021-001163
Application 15/783,228

15
Example 37 of the 2019 PEG Examples recites a user interface that
displays and rearranges icons automatically. Appeal Br. 25–26. Here, the
claims do not recite a user interface or displaying data on a user interface.
Thus, we determine that the claims lack additional elements to
integrate the abstract idea into a practical application.
Alice, Step Two and Revised Guidance Step 2B:
Do the Claims Include an Inventive Concept?
We next consider whether the claims recite 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 the industry. See Berkheimer v. HP Inc., 881 F.3d 1360, 1367
(Fed. Cir. 2018); Revised Guidance, 84 Fed. Reg. at 56 (the second step of
the Alice analysis considers if a claim adds a limitation beyond the judicial
exception that is not “well-understood, routine, conventional” activity).
Individually, the computing device is used as a tool to implement the
abstract idea without improving computers. See Elec. Power, 830 F.3d at
1355 (the claims interpreted in light of the specification required off-the-
shelf conventional computer, network, and display technology to process
information without an inventive concept). As an ordered combination, the
claims recite no more than when the limitations are considered individually.
See BSG Tech LLC v. BuySeasons, Inc., 899 F.3d 1281, 1290–91 (Fed. Cir.
2018) (“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.”).
Thus, we sustain the rejection of claims 1–12, 14, 15, and 21–26.
Appeal 2021-001163
Application 15/783,228

16
Claims 1–11 and 21–25
Anticipated by Bakshi
Appellant argues the claims as a group. Appeal Br. 31–36. We select
claim 1 as representative. 37 C.F.R. § 41.37(c)(1)(iv).
Appellant argues that Bakshi does not disclose training data with user
interactions with digital marketing content output for plural training digital
videos, an outcome of user interaction, and a time when digital marketing
content is output in relation to training videos of the plural training videos.
Appeal Br. 31–32. Appellant asserts that Bakshi instead times the display of
an advertisement during a particular video rather than user interactions with
digital marketing content output with a plurality of training digital videos as
recited in claim 1. Id. at 32–35.
We agree with the Examiner that Bakshi uses optimization component
402 to run a gradient ascent learning mechanism on videos to optimize the
timing to integrate advertisements into respective videos of a plurality of
videos to increase or maintain the viewership retention and advertisement
conversion. Non-Final Act. 12–13; Ans. 11–14. Optimization component
402 experiments with different insertion points for playing an advertisement
during a video to evaluate the effect of playing advertisements at different
insertion points on advertisement conversion. Bakshi, 15:21–43; Ans. 12.
Bakshi uses machine learning to identify user conversion interactions
with advertisements output in conjunction with a plurality of training videos
at different times in relation to each of the training digital videos as claimed.
According to the Specification, training data is “configured to describe user
interaction with the digital video 114 and digital marketing content 124.”
Spec. ¶ 51 (emphasis added). Thus, interactions with a video are analyzed.
Appeal 2021-001163
Application 15/783,228

17
The Specification thus creates the model from user interactions with a
particular digital video 114 and a particular digital marketing content 124.
Training data may include user interactions with a plurality of digital videos
and marketing content, but each user interaction is measured for a particular,
single digital video and a particular digital marketing content, as in Bakshi.
Bakshi uses a learning mechanism of optimization component 402 on
individual videos to identify optimal insertion points of advertisements in
each video of a plurality of “respective videos.” See Bakshi, 15:21–43. The
learning mechanism can learn to select different insertion points in a video
“based on the various parameters described herein employed to identify high
interest segments and points associated with those high interest segments.”
Id. at 15:34–37. The parameters include identifying high interest segments
of videos based on user demographics (e.g., age, gender, location, language,
etc.) or audience characteristics (older, younger). Id. at 10:22–53. Filter 302
identifies high interest segments of videos based on video type/genre such as
sports videos. Id. at 14:15–15:5. Bakshi’s learning mechanism is trained on
user segments, user interactions with segments of “respective videos” of a
plurality of videos, outcomes, and timing of advertisements at high interest
video segments. The trained mechanism is applied to “respective videos” of
plural videos to suggest advertisement times as claimed. See Ans. 12.
Nor does Appellant’s argument apprise us of error in the Examiner’s
findings that Bakshi trains a model by running a gradient ascent learning
mechanism on individual videos of “respective videos” to optimize insertion
points for advertisements that provide the highest advertisement conversion
rates (user interactions with advertisements) in a video and considering other
parameters. Bakshi, 15:21–43; see Non-Final Act. 12, 13, Ans. 14.
Appeal 2021-001163
Application 15/783,228

18
As the Examiner points out, other parameters used to train the model
are video types, user segment demographics, user interactions and outcomes
(i.e., conversions), and a time to output the advertisement in other videos.
Id. at 9:4–11:5, 11:35–12:16, 14:15–15:10, 15:21–16:55; Ans. 13–14; Non-
Final Act. 12–13. Bakshi trains the learning mechanism model, filters and
classifiers on single videos in order to use the trained learning mechanism
model, filters, and classifiers to select insertion points (times) to output
advertisements in other videos as claimed. The learning mechanism thereby
suggests insertion times of advertisements based on learning insertion times
and other parameters (e.g., user segments, video types, conversion rates) in
other videos in the training data as discussed above. Ans. 12–14.
Accordingly, we sustain the rejection of claim 1 and claims 2–11 and
21–25, which fall with claim 1. See Appeal Br. 36.
Claims 12, 14, 15, and 26
Rejected Over Bakshi and McIntire
Independent claim 12 recites a method of training a model using
machine learning based on training data that includes “a tag that describes a
content creation characteristic of how content included within the respective
portions of the at least one training digital video is created.” Appeal Br. 44
(Claims App.). The Examiner cites McIntire to teach this feature. Non-
Final Act. 17–18; Ans. 15–16. The Examiner finds that McIntire provides a
segment identifier of a media stream to indicate a color scheme of a segment
of the media stream so supplemental content is altered to match better with
the media stream such as selecting a background color of dark blue (rather
than pale yellow) to match a night scene of the video stream. Non-Final
Act. 18 (quoting McIntire ¶ 215); Ans. 16 (quoting McIntire ¶ 215).
Appeal 2021-001163
Application 15/783,228

19
Appellant argues that “[n]othing about McIntire teaches ‘how content
. . . is created.’” Reply Br. 9. Appellant asserts that even if portions of
McIntire describe “what” is included in a video, none of these portions teach
or suggest “how content is created” as claimed. Id. at 10. We disagree.
The Specification describes tags 120 as content creation traits within
respective portions of the video to include “colors used, lighting conditions,
digital filters, etc.” Spec. ¶ 30. McIntire uses segment identifiers to identify
lighting conditions of a scene filmed at night where the primary colors in the
night scene are muted so background colors of supplemental content must be
darker. McIntire ¶ 215. This teaching corresponds to the Specification’s
description of content tags describing lighting conditions and colors used.
Accordingly, we sustain the rejection of claim 12 and claims 14, 15,
and 26, which fall with claim 12.
CONCLUSION
In summary:
Claim(s)
Rejected
35 U.S.C.
§
Reference(s)/
Basis
Affirmed Reversed
1–12, 14, 15,
21–26
101 Eligibility 1–12, 14, 15,
21–26

1–11, 21–25 102 Bakshi 1–11, 21–25
12, 14, 15,
26
103 Bakshi,
McIntire
12, 14, 15, 26
Overall
Outcome
1–12, 14, 15,
21–26

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