DETAILED ACTION
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 04/10/2026 has been entered.
Status of Claims
Claims 1-13, 15-17, and 21-24 submitted on 04/10/2026 are pending and have been examined. Claims 1-13 and 15-17 have been amended. Claims 14 and 18-20 have been cancelled. Claims 21-24 are newly added.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Priority
Acknowledgment is made of applicant's provisional Application No. 63/442,287, filed on 01/31/2023.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
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.
Claims 1-13, 15-17, and 21-24 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception without significantly more. The claims recite an abstract idea. This judicial exception is not integrated into a practical application. The claim(s) do not include additional elements that are sufficient to amount to significantly more than the judicial exception.
Step 1
Claims 1-10 are directed to a machine, claims 11-13 and 15-17 are directed to a process, and claims 21-24 are directed to an article of manufacture (see MPEP 2106.03).
Step 2A, Prong 1
Claim 1, taken as representative, recites at least the following limitations that recite an abstract idea:
training by inputting training image feature vectors, and for training item images for portions of data of a combination of data, in a sequence of different types of data to match one or more other data selections in the combination of data;
determining an anchor image embedding for an anchor image for anchor data for which is to be displayed;
determining one or more image embeddings of one or more respective complementary data for the anchor data;
generating, based on a distance between the anchor image embedding and the one or more image embeddings of the one or more respective complementary data for the anchor data, one or more preliminary combinations of data;
determining, respective data for the anchor data for each of the one or more preliminary combinations of data based at least in part on respective visual compatibility of the respective data with respective existing data of each of the one or more preliminary combinations of data based on training by inputting the training image feature vectors; and
transmitting, for display, information based on determining, the respective data for the anchor data.
The above limitation, under its broadest reasonable interpretation, falls within the “Certain Methods of Organizing Human Activity” grouping of abstract ideas, enumerated in MPEP 2106.04(a)(2)(II), in that it recites a commercial interaction. Claims 11 and 21 recites similar limitations as claim 1.
Thus, under Prong 1 of Step 2A, claims 1, 11, and 21 recite an abstract idea.
Step 2A, Prong 2
Claim 1 includes the following additional elements that are bolded:
a system comprising one or more processors and one or more non-transitory computer-readable media storing computing instructions that, when executed on the one or more processors, cause the one or more processors to perform operations comprising:
training a machine learning module by inputting training image feature vectors, generated by an image encoder and for training item images for portions of data of a combination of data from one or more databases, in a sequence of different types of data for the machine learning module to match one or more other data selections in the combination of data;
determining an anchor image embedding for an anchor image for anchor data for which a webpage is to be displayed on an online website;
determining one or more image embeddings of one or more respective complementary data for the anchor data;
generating, based on a distance between the anchor image embedding and the one or more image embeddings of the one or more respective complementary data for the anchor data, one or more preliminary combinations of data for the machine learning module;
determining, via the machine learning module, respective data for the anchor data for each of the one or more preliminary combinations of data based at least in part on respective visual compatibility of the respective data with respective existing data of each of the one or more preliminary combinations of data based on training the machine learning module by inputting the training image feature vectors; and
transmitting, via a computer network and for display on the webpage, information based on determining, via the machine learning module, the respective data for the anchor data.
Claims 11 and 21 includes the same additional elements as claim 1. In addition, claim 21 includes additional elements such as a non-transitory, computer-readable medium comprising instructions that, when executed by a processing resource, cause the processing resource to perform operations comprising.
The additional elements recited in claims 1, 11, and 21 merely invoke such elements as a tool to perform the abstract idea and generally link the use of the abstract idea to a particular technological environment of machine learning modules (see MPEP 2106.05(f) and MPEP 2106.05(h). These additional elements are described at a high level in Applicant’s specification without any meaningful detail about their structure or configuration (see ¶¶0017-0018, ¶0043, and Fig. 2).
As such, under Prong 2 of Step 2A, when considered both individually and as a whole, the additional elements do not integrate the judicial exception into a practical application and, thus, claims 1, 11, and 21 are directed to an abstract idea.
Step 2B
As noted above, while the recitation of the additional elements in independent claims 1, 11, and 21 are acknowledged, claims 1, 11, and 21 merely invoke such additional elements as a tool to perform the abstract idea and generally link the use of the abstract idea to a particular technological environment (see MPEP 2106.05(f) and MPEP 2106.05(h)).
Even when considered as an ordered combination, the additional elements of claim 1, 11, and 21 do not add anything that is not already present when they are considered individually. Therefore, under Step 2B, there are no meaningful limitations in claims 1, 11, and 21 that transform the judicial exception into a patent eligible application such that the claims amount to significantly more than the judicial exception itself (see MPEP 2106.05).
As such, independent claims 1, 11, and 21 are ineligible.
Dependent claims 2-5, 7-10, 12, 13, 15, 17, and 22-23 when analyzed as a whole, are held to be patent ineligible under 35 U.S.C. 101 because they do not add “significantly more” to the abstract idea. More specifically, dependent claims 2-5, 7-10, 12, 13, 15, 17, and 22-23 merely further define the abstract limitations of claims 1, 11, and 21 or provide further embellishments of the limitations recited in independent claims 1, 11, and 21. Claims 2-5, 7-10, 12, 13, 15, 17, and 22-23 do not introduce any further additional elements. Thus, dependent claims 2-5, 7-10, 12, 13, 15, 17, and 22-23 are ineligible.
Furthermore, it is noted that certain dependent claims recite additional elements supplemental to those recited in independent claims 1, 11, and 21: an InceptionV3 or a Contrastive Language-Image Pre-Training (CLIP) (Claims 6, 16, and 24). However, these elements do not integrate the abstract idea into a practical application because they merely amount to using a computer to apply the abstract idea to a particular technological environment or field of use and thus do not act to integrate the abstract idea into a practical application of the abstract idea. Additionally, the additional elements do not amount to significantly more because they merely amount to using a computer to apply the abstract idea and amount to no more than a general link of the use of the abstract idea to a particular technological environment. Thus, dependent claims 6, 16, and 24 are ineligible.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status. This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 1, 2, 6, 7, 10-12, 16, and 17 is/are rejected under 35 U.S.C. 103 as being unpatentable over Collomosse et al. (US 2022/0092108 A1) in view of Wiesel et al. (US 2019/0244407 A1).
Regarding Claim 1, Collomosse et al., hereinafter, Collomosse, discloses a system comprising one or more processors and one or more non-transitory computer-readable media storing computing instructions that, when executed on the one or more processors, cause the one or more processors to perform operations comprising (¶0102[Embodiments of the present disclosure may comprise or utilize a special purpose or general-purpose computer including computer hardware, such as, for example, one or more processors and system memory, as discussed in greater detail below… one or more of the processes described herein may be implemented at least in part as instructions embodied in a non-transitory computer-readable medium and executable by one or more computing devices (e.g., any of the media content access devices described herein). In general, a processor (e.g., a microprocessor) receives instructions, from a non-transitory computer-readable medium, (e.g., a memory, etc.), and executes those instructions, thereby performing one or more processes, including one or more of the processes described herein.]):
training a machine learning module by inputting training image feature vectors, generated by an image encoder and for training item images for portions of data of a combination of data from one or more databases, in a sequence of different types of data for the machine learning module to match one or more other data selections in the combination of data (Figs. 4-6[showing inputting training image feature vectors in a model]; ¶¶0045-0046[To generate style embeddings for digital images, the style search system 102 trains or tunes one or more style extraction neural networks based on training data… As illustrated in FIG. 3, the style search system 102 generates a set of sample digital images 302 for learning parameters of a style extraction neural network through training or tuning. In particular, the style search system 102 generates the set of sample digital images 302 from online collections of digital images] in view of ¶0037[To generate the style embedding, the style search system 102 utilizes one or more style extraction neural networks such as a novel two-branch autoencoder neural network, a weakly supervised discriminative neural network, or a combination of the two.] and ¶0019[In some embodiments, based on extracted style embeddings, the style search system determines a style for a query digital image and searches a repository of digital images to identify other digital images with styles similar to the query digital image.]);
determining an anchor image embedding for an anchor image for anchor data for which a webpage is to be displayed on an online website (Fig. 2[204]; ¶¶0039-0043[the style search system 102 performs an act 204 to generate a style embedding. More specifically, the style search system 102 generates a style embedding that includes one or more features indicating a style of a digital image… a style extraction neural network extracts style codes and generates a style embedding for a digital image such as a query digital image and/or digital images within a digital image repository… The style search system 102 thus provides the identified digital images for display on the client device 108 (e.g., with a style search interface)]);
determining one or more image embeddings of one or more respective complementary data for the anchor data (Fig. 2[206]; ¶0041[As further illustrated in FIG. 2, the style search system 102 performs an act 206 to compare style embeddings. More specifically, the style search system 102 compares a style embedding for a query digital image with style embeddings for stored digital images within a digital image repository (e.g., within the database 114).]);
based on a distance between the anchor image embedding and the one or more image embeddings of the one or more respective complementary data for the anchor data, one or more preliminary data for the machine learning module (Fig. 2[206]; ¶¶0041-0042[Additionally, the style search system 102 performs an act 208 to identify digital images with a similar style to the query digital image. Particularly, the style search system 102 identifies digital images from the digital image repository that have style embeddings within a threshold similarity of the style embedding of the query digital image. For example, based on comparing the style embeddings (e.g., via the act 206), the style search system 102 identifies and selects digital images whose style embeddings are within a threshold distance of the style embedding for the query digital image within the embedding space.]);
determining, via the machine learning module, respective data for the anchor data for each of the one or more data based at least in part on respective visual compatibility of the respective data with respective existing data of each of the one or more data based on training the machine learning module by inputting the training image feature vectors (Fig. 2[208] and Figs. 4-6; ¶0042[Additionally, the style search system 102 performs an act 208 to identify digital images with a similar style to the query digital image. Particularly, the style search system 102 identifies digital images from the digital image repository that have style embeddings within a threshold similarity of the style embedding of the query digital image. For example, based on comparing the style embeddings (e.g., via the act 206), the style search system 102 identifies and selects digital images whose style embeddings are within a threshold distance of the style embedding for the query digital image within the embedding space.]); and
transmitting, via a computer network and for display on the webpage, information based on determining, via the machine learning module, the respective data for the anchor data (Fig. 2[108]; ¶0043[he style search system 102 thus provides the identified digital images for display on the client device 108 (e.g., with a style search interface). Indeed, as illustrated in FIG. 2, the client device 108 presents a style search interface that includes the query digital image (e.g., the hairy feet in fish sandals), along with four selected digital images with similar style.]).
Although Collomosse discloses a distance between image embeddings and data for machine learning modules, Collomosse does not explicitly disclose generating combinations of data, one or more preliminary combinations of data based on compatibility with the one or more preliminary combinations of data.
However, Wiesel et al., hereinafter, Wiesel, teaches generating combinations of data (¶0509[to receive a user-image of said user, to receive a user-query for a clothing article, to generate a plurality of image-based search results in response to said user-query, wherein each image-based search result corresponds to a different instance of said clothing article, to modify each image-based search result into a combined image that depicts a combination of said user wearing said instance of the clothing article; to present to said user a set of image-based search results, wherein each image-based search result comprises a combined image that depicts the combination of said user wearing said instance of the clothing article.]).
The system of Wiesel is applicable to the system of Collomosse as they share characteristics and capabilities, namely, they are both targeted to identifying and manipulating digital images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the training of a machine learning module as disclosed by Collomosse to include generating combinations of data as taught by Wiesel. One of ordinary skill in the art would have been motivated to expand the system of Collomosse in order to generate a realistic image that emulates or simulates how that particular user would appear if he or she wears a particular article of clothing, or other article (¶0004).
Regarding Claim 2, Collomosse in view of Wiesel teaches the system of claim 1, Collomosse further discloses wherein generating the one or more data comprises:
determining the one or more respective complementary data for the anchor data based on multiple algorithms, based on the anchor image embedding, and based on the one or more image embeddings of the one or more respective complementary data for the anchor data (Fig. 2[206]; ¶¶0041-0042[Additionally, the style search system 102 performs an act 208 to identify digital images with a similar style to the query digital image. Particularly, the style search system 102 identifies digital images from the digital image repository that have style embeddings within a threshold similarity of the style embedding of the query digital image. For example, based on comparing the style embeddings (e.g., via the act 206), the style search system 102 identifies and selects digital images whose style embeddings are within a threshold distance of the style embedding for the query digital image within the embedding space.] in view of ¶0044[the style search system 102 utilizes particular acts and algorithms to implement or perform the step for generating the style embedding. Specifically, the description above in relation to the act 204 of FIG. 2, in addition to the corresponding description of FIGS. 4-6, provides acts and algorithms as structure and support for performing a step for generating a style embedding disentangled from image content for the query digital image.]).
Although Collomosse discloses determining complementary data, Collomosse does not explicitly disclose preliminary combinations of data and determining data in each of one or more remaining non-accessory types.
However, Wiesel teaches generating combinations of data and non-accessory types (¶0509[to receive a user-image of said user, to receive a user-query for a clothing article, to generate a plurality of image-based search results in response to said user-query, wherein each image-based search result corresponds to a different instance of said clothing article, to modify each image-based search result into a combined image that depicts a combination of said user wearing said instance of the clothing article; to present to said user a set of image-based search results, wherein each image-based search result comprises a combined image that depicts the combination of said user wearing said instance of the clothing article.] in view of ¶0497[In some embodiments, system 5000 may comprise a realistic wrinkles generator 5009, to receive an image of a clothing article (e.g., an image of a yellow shirt, in response to a user selection of a yellow shirt from an online catalog or from a set of search results); to generate a combined image of said clothing article worn by said user]).
The system of Wiesel is applicable to the system of Collomosse as they share characteristics and capabilities, namely, they are both targeted to identifying and manipulating digital images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the training of a machine learning module as disclosed by Collomosse to include generating combinations of data as taught by Wiesel. One of ordinary skill in the art would have been motivated to expand the system of Collomosse in order to generate a realistic image that emulates or simulates how that particular user would appear if he or she wears a particular article of clothing, or other article (¶0004).
Regarding Claim 6, Collomosse in view of Wiesel teaches the system of claim 1, Collomosse further discloses wherein the operations further comprise: generating the training image feature vectors, for the training item images, using one or more of an InceptionV3 or a Contrastive Language-Image Pre-Training (CLIP) (Fig. 10; ¶0026[In these or other embodiments, the style search system utilizes a contrastive loss function (e.g., a normalized temperature-scaled cross entropy loss function) to learn parameters for fine-grain style coherence with a discriminative neural network]).
Regarding Claim 7, Collomosse in view of Wiesel teaches the system of claim 1, Collomosse further discloses wherein training the machine learning module comprises: training a visual-semantic embedding module based on training item texts to generate visual-semantic embeddings (Figs. 1 and 10; ¶0023[For instance, the two-branch autoencoder correlates effectively for color cues, while the weakly supervised discriminative neural network correlates effectively for semantic cues. Thus, in some embodiments, the style search system combines the first style embedding with the second style embedding for additional accuracy.]); and
training the machine learning module further based on the visual-semantic embeddings (Figs. 1 and 10; ¶¶0023-0025[For instance, the two-branch autoencoder correlates effectively for color cues, while the weakly supervised discriminative neural network correlates effectively for semantic cues. Thus, in some embodiments, the style search system combines the first style embedding with the second style embedding for additional accuracy… the style search system trains the style extraction neural network]).
Regarding Claim 10, Collomosse in view of Wiesel teaches the system of claim 1, Collomosse further discloses wherein the operations further comprise generating one or more combinations of data based on the respective data for the anchor data (Fig. 2; ¶¶0041-0042); and
the one or more combinations of data based on a color matrix (¶¶0023-0024[For instance, the two-branch autoencoder correlates effectively for color cues, while the weakly supervised discriminative neural network correlates effectively for semantic cues… the style search system searches for additional digital images with similar style to the query digital image]).
Although Collomosse discloses determining combinations of data based on color, Collomosse does not explicitly disclose ranking the data.
However, Wiesel teaches ranking the data (¶0485[The system may further order search results in a user-specific manner, taking into account user preferences, user settings, user profile, user questionnaire, user body type or ratio or dimensions, current trends or shopping trends or fashion trends (e.g., ranking higher the products that match a current shopping trend), current weather or season (e g, ranking higher Summer Dresses, if the user searches for “dress” in the United States in July), geographic location, or the like.]).
The system of Wiesel is applicable to the system of Collomosse as they share characteristics and capabilities, namely, they are both targeted to identifying and manipulating digital images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the training of a machine learning module as disclosed by Collomosse to include ranking the data as taught by Wiesel. One of ordinary skill in the art would have been motivated to expand the system of Collomosse in order to generate a realistic image that emulates or simulates how that particular user would appear if he or she wears a particular article of clothing, or other article (¶0004).
Regarding Claim 11, Collomosse discloses a computer-implemented method comprising:
one or more data for a machine learning module based on a distance between an anchor image embedding, for an anchor image for anchor data, and one or more image embeddings of one or more respective complementary data for the anchor data (Fig. 2[206]; ¶¶0041-0042[Additionally, the style search system 102 performs an act 208 to identify digital images with a similar style to the query digital image. Particularly, the style search system 102 identifies digital images from the digital image repository that have style embeddings within a threshold similarity of the style embedding of the query digital image. For example, based on comparing the style embeddings (e.g., via the act 206), the style search system 102 identifies and selects digital images whose style embeddings are within a threshold distance of the style embedding for the query digital image within the embedding space.]),
wherein the machine learning module is trained based on training image feature vectors, for data stored in one or more databases, being inputted into the machine learning module in a sequence for a combination of data (Figs. 4-6[showing inputting training image feature vectors in a model]; ¶¶0045-0046[To generate style embeddings for digital images, the style search system 102 trains or tunes one or more style extraction neural networks based on training data… As illustrated in FIG. 3, the style search system 102 generates a set of sample digital images 302 for learning parameters of a style extraction neural network through training or tuning. In particular, the style search system 102 generates the set of sample digital images 302 from online collections of digital images] in view of ¶0037[To generate the style embedding, the style search system 102 utilizes one or more style extraction neural networks such as a novel two-branch autoencoder neural network, a weakly supervised discriminative neural network, or a combination of the two.] and ¶0019[In some embodiments, based on extracted style embeddings, the style search system determines a style for a query digital image and searches a repository of digital images to identify other digital images with styles similar to the query digital image.]); and
using the machine learning module to determine respective data for the anchor data for each of the one or more data based at least in part on respective visual compatibility of the respective data with respective existing data of each of the one or more of data (Fig. 2[208] and Figs. 4-6; ¶0042[Additionally, the style search system 102 performs an act 208 to identify digital images with a similar style to the query digital image. Particularly, the style search system 102 identifies digital images from the digital image repository that have style embeddings within a threshold similarity of the style embedding of the query digital image. For example, based on comparing the style embeddings (e.g., via the act 206), the style search system 102 identifies and selects digital images whose style embeddings are within a threshold distance of the style embedding for the query digital image within the embedding space.]).
Although Collomosse discloses a distance between image embeddings and data for machine learning modules, Collomosse does not explicitly disclose generating preliminary combinations of data, one or more preliminary combinations of data based on compatibility with the one or more preliminary combinations of data.
However, Wiesel teaches generating combinations of data (¶0509[to receive a user-image of said user, to receive a user-query for a clothing article, to generate a plurality of image-based search results in response to said user-query, wherein each image-based search result corresponds to a different instance of said clothing article, to modify each image-based search result into a combined image that depicts a combination of said user wearing said instance of the clothing article; to present to said user a set of image-based search results, wherein each image-based search result comprises a combined image that depicts the combination of said user wearing said instance of the clothing article.]).
The method of Wiesel is applicable to the method of Collomosse as they share characteristics and capabilities, namely, they are both targeted to identifying and manipulating digital images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the training of a machine learning module as disclosed by Collomosse to include generating combinations of data as taught by Wiesel. One of ordinary skill in the art would have been motivated to expand the method of Collomosse in order to generate a realistic image that emulates or simulates how that particular user would appear if he or she wears a particular article of clothing, or other article (¶0004).
Regarding Claim 12, Collomosse in view of Wiesel teaches the computer-implemented method of claim 11, Collomosse further discloses further comprising: determining the one or more respective complementary data for the anchor data based on multiple algorithms (Fig. 2[206]; ¶¶0041-0042[Additionally, the style search system 102 performs an act 208 to identify digital images with a similar style to the query digital image. Particularly, the style search system 102 identifies digital images from the digital image repository that have style embeddings within a threshold similarity of the style embedding of the query digital image. For example, based on comparing the style embeddings (e.g., via the act 206), the style search system 102 identifies and selects digital images whose style embeddings are within a threshold distance of the style embedding for the query digital image within the embedding space.] in view of ¶0044[the style search system 102 utilizes particular acts and algorithms to implement or perform the step for generating the style embedding. Specifically, the description above in relation to the act 204 of FIG. 2, in addition to the corresponding description of FIGS. 4-6, provides acts and algorithms as structure and support for performing a step for generating a style embedding disentangled from image content for the query digital image.]).
Although Collomosse discloses determining complementary data, Collomosse does not explicitly disclose determining data in each of one or more remaining types.
However, Wiesel teaches generating combinations of data and non-accessory types (¶0509[to receive a user-image of said user, to receive a user-query for a clothing article, to generate a plurality of image-based search results in response to said user-query, wherein each image-based search result corresponds to a different instance of said clothing article, to modify each image-based search result into a combined image that depicts a combination of said user wearing said instance of the clothing article; to present to said user a set of image-based search results, wherein each image-based search result comprises a combined image that depicts the combination of said user wearing said instance of the clothing article.] in view of ¶0497[In some embodiments, system 5000 may comprise a realistic wrinkles generator 5009, to receive an image of a clothing article (e.g., an image of a yellow shirt, in response to a user selection of a yellow shirt from an online catalog or from a set of search results); to generate a combined image of said clothing article worn by said user]).
The method of Wiesel is applicable to the method of Collomosse as they share characteristics and capabilities, namely, they are both targeted to identifying and manipulating digital images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the training of a machine learning module as disclosed by Collomosse to include generating combinations of data as taught by Wiesel. One of ordinary skill in the art would have been motivated to expand the method of Collomosse in order to generate a realistic image that emulates or simulates how that particular user would appear if he or she wears a particular article of clothing, or other article (¶0004).
Regarding Claim 16, Collomosse in view of Wiesel teaches the computer-implemented method of claim 11, Collomosse further discloses further comprising: generating the training image feature vectors, for the training images, using one or more of an InceptionV3 or a Contrastive Language-Image Pre-Training (CLIP) (Fig. 10; ¶0026[In these or other embodiments, the style search system utilizes a contrastive loss function (e.g., a normalized temperature-scaled cross entropy loss function) to learn parameters for fine-grain style coherence with a discriminative neural network]).
Regarding Claim 17, Collomosse in view of Wiesel teaches the computer-implemented method of claim 11, Collomosse further discloses further comprising: training a visual-semantic embedding module based on training item texts to generate visual-semantic embeddings for the training data (Figs. 1 and 10; ¶0023[For instance, the two-branch autoencoder correlates effectively for color cues, while the weakly supervised discriminative neural network correlates effectively for semantic cues. Thus, in some embodiments, the style search system combines the first style embedding with the second style embedding for additional accuracy.]); and
training the machine learning module further based on the visual-semantic embeddings (Figs. 1 and 10; ¶¶0023-0025[For instance, the two-branch autoencoder correlates effectively for color cues, while the weakly supervised discriminative neural network correlates effectively for semantic cues. Thus, in some embodiments, the style search system combines the first style embedding with the second style embedding for additional accuracy… the style search system trains the style extraction neural network]).
Claim(s) 3-5, 13, and 15 is/are rejected under 35 U.S.C. 103 as being unpatentable over Collomosse in view of Wiesel in view of Chen et al. (US 7,058,598 B1).
Regarding Claim 3, Collomosse in view of Wiesel teaches the system of claim 2, Collomosse further discloses wherein: determining the one or more respective complementary data for the anchor data comprises:
determining one or more first respective complementary data for the anchor data based on a respective graph-based similarity between the anchor data and each of the one or more first respective complementary data (Figs. 1-3; ¶¶0049-0050[In addition, the style search system 102 determines a consensus using graph-based vote pooling. For instance, the style search system 102 utilizes an affinity matrix A.sub.i,j to code edges of a style graph, where the affinity matrix A.sub.i,j reflects the number of times digital images i and j were simultaneously selected within an annotation task… Thus, based on receiving responses from labeling devices indicating which digital images in various annotation tasks share common styles, the style search system 102 increases the relationship, or the style similarity, between digital images that are more frequently identified as sharing a common style]); and
determining one or more second respective complementary data for the anchor data based on a respective signal between the anchor data and each of the one or more second respective complementary data (Figs. 1-3; ¶¶0049-0050[In addition, the style search system 102 determines a consensus using graph-based vote pooling. For instance, the style search system 102 utilizes an affinity matrix A.sub.i,j to code edges of a style graph, where the affinity matrix A.sub.i,j reflects the number of times digital images i and j were simultaneously selected within an annotation task… Thus, based on receiving responses from labeling devices indicating which digital images in various annotation tasks share common styles, the style search system 102 increases the relationship, or the style similarity, between digital images that are more frequently identified as sharing a common style]); and
the one or more respective complementary data comprise one or more first respective complementary data and the one or more second respective complementary data (Figs. 1-3; ¶¶0049-0050[In addition, the style search system 102 determines a consensus using graph-based vote pooling. For instance, the style search system 102 utilizes an affinity matrix A.sub.i,j to code edges of a style graph, where the affinity matrix A.sub.i,j reflects the number of times digital images i and j were simultaneously selected within an annotation task… Thus, based on receiving responses from labeling devices indicating which digital images in various annotation tasks share common styles, the style search system 102 increases the relationship, or the style similarity, between digital images that are more frequently identified as sharing a common style]).
Although Collomosse discloses determining complementary data for an anchor data, Collomosse does not explicitly disclose data in each of the one or more remaining non-accessory types.
However, Wiesel teaches generating combinations of data and non-accessory types (¶0509[to receive a user-image of said user, to receive a user-query for a clothing article, to generate a plurality of image-based search results in response to said user-query, wherein each image-based search result corresponds to a different instance of said clothing article, to modify each image-based search result into a combined image that depicts a combination of said user wearing said instance of the clothing article; to present to said user a set of image-based search results, wherein each image-based search result comprises a combined image that depicts the combination of said user wearing said instance of the clothing article.] in view of ¶0497[In some embodiments, system 5000 may comprise a realistic wrinkles generator 5009, to receive an image of a clothing article (e.g., an image of a yellow shirt, in response to a user selection of a yellow shirt from an online catalog or from a set of search results); to generate a combined image of said clothing article worn by said user]).
The system of Wiesel is applicable to the system of Collomosse as they share characteristics and capabilities, namely, they are both targeted to identifying and manipulating digital images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the training of a machine learning module as disclosed by Collomosse to include generating combinations of data as taught by Wiesel. One of ordinary skill in the art would have been motivated to expand the system of Collomosse in order to generate a realistic image that emulates or simulates how that particular user would appear if he or she wears a particular article of clothing, or other article (¶0004).
Although Collomosse discloses data comprising complementary data, Collomosse in view of Wiesel does not explicitly teach a union of the data.
However, Chen et al., hereinafter, Chen, teaches a union of data (Col. 10, lines 10-25[calculating, using algorithms including Cartesian product and set union operations performed on the data stored in the first and second matrices, an approximately optimized rank of all of the items considered as a package; and displaying the approximately optimized ranks to the user in table format.]).
The system of Chen is applicable to the system of Collomosse in view of Wiesel as they share characteristics and capabilities, namely, they are all targeted to browsing for items online. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the manipulation of complementary data as taught by Collomosse in view of Wiesel to include a union of data as taught by Chen. One of ordinary skill in the art would have been motivated to expand the system of Collomosse in view of Wiesel in order to facilitate comparison browsing and shopping (Col. 1, lines 14-17).
Regarding Claim 4, Collomosse in view of Wiesel in view of Chen teaches the system of claim 3, Collomosse further discloses wherein the operations further comprise: determining the respective graph-based similarity based on the distance between the anchor image embedding and a first respective image embedding of the one or more image embeddings of the one or more respective complementary data for the anchor data (Fig. 2[206]; ¶¶0041-0042[Additionally, the style search system 102 performs an act 208 to identify digital images with a similar style to the query digital image. Particularly, the style search system 102 identifies digital images from the digital image repository that have style embeddings within a threshold similarity of the style embedding of the query digital image. For example, based on comparing the style embeddings (e.g., via the act 206), the style search system 102 identifies and selects digital images whose style embeddings are within a threshold distance of the style embedding for the query digital image within the embedding space.] in view of ¶¶0049-0050).
Regarding Claim 5, Collomosse in view of Wiesel in view of Chen teaches the system of claim 3, Collomosse further discloses wherein determining the one or more second respective complementary data for the anchor data further comprises:
determining one or more similar items for the anchor data (¶0024[In at least one embodiment, the style search system searches for additional digital images with similar style to the query digital image. For example, the style search system accesses and searches a repository of digital images for which the style search system has generated corresponding style embeddings]); and
determining the one or more second respective complementary data for the anchor (Figs. 1-3; ¶¶0049-0050[In addition, the style search system 102 determines a consensus using graph-based vote pooling. For instance, the style search system 102 utilizes an affinity matrix A.sub.i,j to code edges of a style graph, where the affinity matrix A.sub.i,j reflects the number of times digital images i and j were simultaneously selected within an annotation task… Thus, based on receiving responses from labeling devices indicating which digital images in various annotation tasks share common styles, the style search system 102 increases the relationship, or the style similarity, between digital images that are more frequently identified as sharing a common style]).
Although Collomosse discloses determining complementary data for an anchor data, Collomosse does not explicitly disclose data in each of the one or more remaining non-accessory types.
However, Wiesel teaches generating combinations of data and non-accessory types (¶0509[to receive a user-image of said user, to receive a user-query for a clothing article, to generate a plurality of image-based search results in response to said user-query, wherein each image-based search result corresponds to a different instance of said clothing article, to modify each image-based search result into a combined image that depicts a combination of said user wearing said instance of the clothing article; to present to said user a set of image-based search results, wherein each image-based search result comprises a combined image that depicts the combination of said user wearing said instance of the clothing article.] in view of ¶0497[In some embodiments, system 5000 may comprise a realistic wrinkles generator 5009, to receive an image of a clothing article (e.g., an image of a yellow shirt, in response to a user selection of a yellow shirt from an online catalog or from a set of search results); to generate a combined image of said clothing article worn by said user]).
The system of Wiesel is applicable to the system of Collomosse as they share characteristics and capabilities, namely, they are both targeted to identifying and manipulating digital images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the training of a machine learning module as disclosed by Collomosse to include generating combinations of data as taught by Wiesel. One of ordinary skill in the art would have been motivated to expand the system of Collomosse in order to generate a realistic image that emulates or simulates how that particular user would appear if he or she wears a particular article of clothing, or other article (¶0004).
Regarding Claim 13, Collomosse in view of Wiesel teaches the computer-implemented method of claim 12, Collomosse further discloses wherein: determining the one or more respective complementary data for the anchor data comprises:
determining one or more first respective complementary data for the anchor data based on a respective graph-based similarity between the anchor data and each of the one or more first respective complementary data (Figs. 1-3; ¶¶0049-0050[In addition, the style search system 102 determines a consensus using graph-based vote pooling. For instance, the style search system 102 utilizes an affinity matrix A.sub.i,j to code edges of a style graph, where the affinity matrix A.sub.i,j reflects the number of times digital images i and j were simultaneously selected within an annotation task… Thus, based on receiving responses from labeling devices indicating which digital images in various annotation tasks share common styles, the style search system 102 increases the relationship, or the style similarity, between digital images that are more frequently identified as sharing a common style]); and
determining one or more second respective complementary data for the anchor data based on a respective signal between the anchor data and each of the one or more second respective complementary data (Figs. 1-3; ¶¶0049-0050[In addition, the style search system 102 determines a consensus using graph-based vote pooling. For instance, the style search system 102 utilizes an affinity matrix A.sub.i,j to code edges of a style graph, where the affinity matrix A.sub.i,j reflects the number of times digital images i and j were simultaneously selected within an annotation task… Thus, based on receiving responses from labeling devices indicating which digital images in various annotation tasks share common styles, the style search system 102 increases the relationship, or the style similarity, between digital images that are more frequently identified as sharing a common style]); and
the one or more respective complementary data comprise one or more first respective complementary data and the one or more second respective complementary data (Figs. 1-3; ¶¶0049-0050[In addition, the style search system 102 determines a consensus using graph-based vote pooling. For instance, the style search system 102 utilizes an affinity matrix A.sub.i,j to code edges of a style graph, where the affinity matrix A.sub.i,j reflects the number of times digital images i and j were simultaneously selected within an annotation task… Thus, based on receiving responses from labeling devices indicating which digital images in various annotation tasks share common styles, the style search system 102 increases the relationship, or the style similarity, between digital images that are more frequently identified as sharing a common style]).
Although Collomosse discloses determining complementary data for an anchor data, Collomosse does not explicitly disclose data in each of one or more remaining types, data in each of one or more remaining non-accessory types, and data in each of the one or more remaining non-accessory types.
However, Wiesel teaches generating combinations of data and non-accessory types (¶0509[to receive a user-image of said user, to receive a user-query for a clothing article, to generate a plurality of image-based search results in response to said user-query, wherein each image-based search result corresponds to a different instance of said clothing article, to modify each image-based search result into a combined image that depicts a combination of said user wearing said instance of the clothing article; to present to said user a set of image-based search results, wherein each image-based search result comprises a combined image that depicts the combination of said user wearing said instance of the clothing article.] in view of ¶0497[In some embodiments, system 5000 may comprise a realistic wrinkles generator 5009, to receive an image of a clothing article (e.g., an image of a yellow shirt, in response to a user selection of a yellow shirt from an online catalog or from a set of search results); to generate a combined image of said clothing article worn by said user]).
The method of Wiesel is applicable to the method of Collomosse as they share characteristics and capabilities, namely, they are both targeted to identifying and manipulating digital images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the training of a machine learning module as disclosed by Collomosse to include generating combinations of data as taught by Wiesel. One of ordinary skill in the art would have been motivated to expand the method of Collomosse in order to generate a realistic image that emulates or simulates how that particular user would appear if he or she wears a particular article of clothing, or other article (¶0004).
Although Collomosse discloses data comprising complementary data, Collomosse in view of Wiesel does not explicitly teach a union of the data.
However, Chen teaches a union of data (Col. 10, lines 10-25[calculating, using algorithms including Cartesian product and set union operations performed on the data stored in the first and second matrices, an approximately optimized rank of all of the items considered as a package; and displaying the approximately optimized ranks to the user in table format.]).
The method of Chen is applicable to the method of Collomosse in view of Wiesel as they share characteristics and capabilities, namely, they are all targeted to browsing for items online. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the manipulation of complementary data as taught by Collomosse in view of Wiesel to include a union of data as taught by Chen. One of ordinary skill in the art would have been motivated to expand the method of Collomosse in view of Wiesel in order to facilitate comparison browsing and shopping (Col. 1, lines 14-17).
Regarding Claim 15, Collomosse in view of Wiesel in view of Chen teaches the computer-implemented method of claim 13, Collomosse further discloses wherein determining the one or more second respective complementary data for the anchor data further comprises:
determining one or more similar data for the anchor data (¶0024[In at least one embodiment, the style search system searches for additional digital images with similar style to the query digital image. For example, the style search system accesses and searches a repository of digital images for which the style search system has generated corresponding style embeddings]); and
determining the one or more second respective complementary data for the anchor data further based on a respective signal between each of the one or more similar data and each of the one or more second respective complementary data (Figs. 1-3; ¶¶0049-0050[In addition, the style search system 102 determines a consensus using graph-based vote pooling. For instance, the style search system 102 utilizes an affinity matrix A.sub.i,j to code edges of a style graph, where the affinity matrix A.sub.i,j reflects the number of times digital images i and j were simultaneously selected within an annotation task… Thus, based on receiving responses from labeling devices indicating which digital images in various annotation tasks share common styles, the style search system 102 increases the relationship, or the style similarity, between digital images that are more frequently identified as sharing a common style]).
Although Collomosse discloses determining complementary data for an anchor data, Collomosse does not explicitly disclose data in each of the one or more remaining non-accessory types.
However, Wiesel teaches generating combinations of data and non-accessory types (¶0509[to receive a user-image of said user, to receive a user-query for a clothing article, to generate a plurality of image-based search results in response to said user-query, wherein each image-based search result corresponds to a different instance of said clothing article, to modify each image-based search result into a combined image that depicts a combination of said user wearing said instance of the clothing article; to present to said user a set of image-based search results, wherein each image-based search result comprises a combined image that depicts the combination of said user wearing said instance of the clothing article.] in view of ¶0497[In some embodiments, system 5000 may comprise a realistic wrinkles generator 5009, to receive an image of a clothing article (e.g., an image of a yellow shirt, in response to a user selection of a yellow shirt from an online catalog or from a set of search results); to generate a combined image of said clothing article worn by said user]).
The method of Wiesel is applicable to the method of Collomosse as they share characteristics and capabilities, namely, they are both targeted to identifying and manipulating digital images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the training of a machine learning module as disclosed by Collomosse to include generating combinations of data as taught by Wiesel. One of ordinary skill in the art would have been motivated to expand the method of Collomosse in order to generate a realistic image that emulates or simulates how that particular user would appear if he or she wears a particular article of clothing, or other article (¶0004).
Claim(s) 8 and 9 is/are rejected under 35 U.S.C. 103 as being unpatentable over Collomosse in view of Wiesel in view of Best et al. (US 2019/0228444 A1).
Regarding Claim 8, Collomosse in view of Wiesel teaches the system of claim 1, Collomosse further discloses wherein the operations further comprise: anchor data from each of one or more combinations of data that are based on the respective data for the anchor data (Fig. 2[204]; ¶¶0039-0043[the style search system 102 performs an act 204 to generate a style embedding. More specifically, the style search system 102 generates a style embedding that includes one or more features indicating a style of a digital image… a style extraction neural network extracts style codes and generates a style embedding for a digital image such as a query digital image and/or digital images within a digital image repository… The style search system 102 thus provides the identified digital images for display on the client device 108 (e.g., with a style search interface)]); and
complementary data determining and respective data determining for the respective simulation anchor data (Fig. 2[206]; ¶0041[As further illustrated in FIG. 2, the style search system 102 performs an act 206 to compare style embeddings. More specifically, the style search system 102 compares a style embedding for a query digital image with style embeddings for stored digital images within a digital image repository (e.g., within the database 114).]).
Although Collomosse discloses anchor and complementary data, Collomosse in view of Wiesel does not explicitly teach choosing a respective simulation data and simulating.
However, Best et al., hereinafter, Best teaches choosing a simulation and simulating data (¶¶0022-0024[The system may communicate, in real time, input data to one or more servers configured to generate at least one simulated product based on the input data to display the simulated product in the state selected by the user]).
The system of Best is applicable to the system of Collomosse in view of Wiesel as they share characteristics and capabilities, namely, they are all targeted to browsing for items online. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the manipulation of complementary data as taught by Collomosse in view of Wiesel to include simulation of data as taught by Best. One of ordinary skill in the art would have been motivated to expand the system of Collomosse in view of Wiesel in order to provide the user with further choices other than those directly offered by the retailer (¶0003).
Regarding Claim 9, Collomosse in view of Wiesel in view of Best teaches the system of claim 8, Collomosse further discloses wherein the operations further comprise one or more of:
the one or more combinations of data based on a color matrix (¶¶0023-0024[For instance, the two-branch autoencoder correlates effectively for color cues, while the weakly supervised discriminative neural network correlates effectively for semantic cues… the style search system searches for additional digital images with similar style to the query digital image]); or
updating a plurality of templates based on impression signals associated with historical combinations of data created based on the plurality of templates.
Although Collomosse discloses determining combinations of data based on color, Collomosse does not explicitly disclose ranking the data.
However, Wiesel teaches ranking the data (¶0485[The system may further order search results in a user-specific manner, taking into account user preferences, user settings, user profile, user questionnaire, user body type or ratio or dimensions, current trends or shopping trends or fashion trends (e.g., ranking higher the products that match a current shopping trend), current weather or season (e g, ranking higher Summer Dresses, if the user searches for “dress” in the United States in July), geographic location, or the like.]).
The system of Wiesel is applicable to the system of Collomosse as they share characteristics and capabilities, namely, they are both targeted to identifying and manipulating digital images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the training of a machine learning module as disclosed by Collomosse to include ranking the data as taught by Wiesel. One of ordinary skill in the art would have been motivated to expand the system of Collomosse in order to generate a realistic image that emulates or simulates how that particular user would appear if he or she wears a particular article of clothing, or other article (¶0004).
Claim(s) 21-24 is/are rejected under 35 U.S.C. 103 as being unpatentable over Collomosse in view of Wiesel in view of Wade et al. (US 2022/0383400 A1).
Regarding Claim 21, Collomosse discloses a non-transitory, computer-readable medium comprising instructions that, when executed by a processing resource, cause the processing resource to perform operations comprising (¶0102[Embodiments of the present disclosure may comprise or utilize a special purpose or general-purpose computer including computer hardware, such as, for example, one or more processors and system memory, as discussed in greater detail below… one or more of the processes described herein may be implemented at least in part as instructions embodied in a non-transitory computer-readable medium and executable by one or more computing devices (e.g., any of the media content access devices described herein). In general, a processor (e.g., a microprocessor) receives instructions, from a non-transitory computer-readable medium, (e.g., a memory, etc.), and executes those instructions, thereby performing one or more processes, including one or more of the processes described herein.]):
one or more data for a machine learning module based on a distance between an anchor image embedding, for an anchor image for anchor data stored in one or more databases, and one or more image embeddings of one or more respective complementary data stored in the one or more databases for the anchor data (Fig. 2[206]; ¶¶0041-0042[Additionally, the style search system 102 performs an act 208 to identify digital images with a similar style to the query digital image. Particularly, the style search system 102 identifies digital images from the digital image repository that have style embeddings within a threshold similarity of the style embedding of the query digital image. For example, based on comparing the style embeddings (e.g., via the act 206), the style search system 102 identifies and selects digital images whose style embeddings are within a threshold distance of the style embedding for the query digital image within the embedding space.]);
using the machine learning module to determine respective data for the anchor data for each of the one or more data based at least in part on respective visual compatibility of the data with respective existing data of each of the one or more data (Fig. 2[208] and Figs. 4-6; ¶0042[Additionally, the style search system 102 performs an act 208 to identify digital images with a similar style to the query digital image. Particularly, the style search system 102 identifies digital images from the digital image repository that have style embeddings within a threshold similarity of the style embedding of the query digital image. For example, based on comparing the style embeddings (e.g., via the act 206), the style search system 102 identifies and selects digital images whose style embeddings are within a threshold distance of the style embedding for the query digital image within the embedding space.]); and
data that includes the respective data based on signals associated with the combination of data (Figs. 1-3; ¶¶0049-0050[In addition, the style search system 102 determines a consensus using graph-based vote pooling. For instance, the style search system 102 utilizes an affinity matrix A.sub.i,j to code edges of a style graph, where the affinity matrix A.sub.i,j reflects the number of times digital images i and j were simultaneously selected within an annotation task… Thus, based on receiving responses from labeling devices indicating which digital images in various annotation tasks share common styles, the style search system 102 increases the relationship, or the style similarity, between digital images that are more frequently identified as sharing a common style]).
Although Collomosse discloses a distance between image embeddings and data for machine learning modules, Collomosse does not explicitly disclose generating preliminary combinations of data, one or more preliminary combinations of data based on compatibility with the one or more preliminary combinations of data.
However, Wiesel teaches generating combinations of data (¶0509[to receive a user-image of said user, to receive a user-query for a clothing article, to generate a plurality of image-based search results in response to said user-query, wherein each image-based search result corresponds to a different instance of said clothing article, to modify each image-based search result into a combined image that depicts a combination of said user wearing said instance of the clothing article; to present to said user a set of image-based search results, wherein each image-based search result comprises a combined image that depicts the combination of said user wearing said instance of the clothing article.]).
The system of Wiesel is applicable to the system of Collomosse as they share characteristics and capabilities, namely, they are both targeted to identifying and manipulating digital images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the training of a machine learning module as disclosed by Collomosse to include generating combinations of data as taught by Wiesel. One of ordinary skill in the art would have been motivated to expand the system of Collomosse in order to generate a realistic image that emulates or simulates how that particular user would appear if he or she wears a particular article of clothing, or other article (¶0004).
Although Collomosse discloses data association based on signals, Collomosse in view of Wiesel does not explicitly teach removing, from a plurality of templates stored in a database, a template used to create a combination of data that includes data based on impression signals, thereby conserving memory resources used to store the template in the database and processing resources used to create combinations of data based on the template.
However, Wade et al., hereinafter, Wade, teaches removing templates stored in database based on impression signals and to conserve memory resources (¶0117[Consider a case in which a merchant has an existing 3D composite model of a product bundle. Individual 3D models representing the products in the product bundle may be removed to obtain a blank template for the 3D composite model. The blank template could then be stored in the template data 417] in view of ¶0065[The product bundling engine 300 may also help conserve computational resources at the e-commerce platform 100. Automatically producing digital media using the product bundling engine 300 may mean that less product media needs to be stored at the e-commerce platform 100 at any given time, which may help conserve computer storage resources]).
The system of Wade is applicable to the system of Collomosse in view of Wiesel as they share characteristics and capabilities, namely, they are all targeted to browsing for items online. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the manipulation of complementary data as taught by Collomosse in view of Wiesel to include removing templates based on impression signals as taught by Wade. One of ordinary skill in the art would have been motivated to expand the system of Collomosse in view of Wiesel in order to more efficiently produce product media for the product bundle (¶0004).
Regarding Claim 22, Collomosse in view of Wiesel in view of Wade teaches the non-transitory, computer-readable medium of claim 21, Collomosse further discloses wherein the operations further comprise: training the machine learning module by inputting, into the machine learning module, training image feature vectors in a sequence of different types of data (Figs. 4-6[showing inputting training image feature vectors in a model]; ¶¶0045-0046[To generate style embeddings for digital images, the style search system 102 trains or tunes one or more style extraction neural networks based on training data… As illustrated in FIG. 3, the style search system 102 generates a set of sample digital images 302 for learning parameters of a style extraction neural network through training or tuning. In particular, the style search system 102 generates the set of sample digital images 302 from online collections of digital images] in view of ¶0037[To generate the style embedding, the style search system 102 utilizes one or more style extraction neural networks such as a novel two-branch autoencoder neural network, a weakly supervised discriminative neural network, or a combination of the two.] and ¶0019[In some embodiments, based on extracted style embeddings, the style search system determines a style for a query digital image and searches a repository of digital images to identify other digital images with styles similar to the query digital image.]).
Regarding Claim 23, Collomosse in view of Wiesel in view of Wade teaches the non-transitory, computer-readable medium of claim 22, Collomosse further discloses wherein the operations further comprise: generating, using an image encoder, the training image feature vectors based on training images for a portions of data of a combination of data from one or more databases (Figs. 4-6[showing inputting training image feature vectors in a model]; ¶¶0045-0046[To generate style embeddings for digital images, the style search system 102 trains or tunes one or more style extraction neural networks based on training data… As illustrated in FIG. 3, the style search system 102 generates a set of sample digital images 302 for learning parameters of a style extraction neural network through training or tuning. In particular, the style search system 102 generates the set of sample digital images 302 from online collections of digital images] in view of ¶0037[To generate the style embedding, the style search system 102 utilizes one or more style extraction neural networks such as a novel two-branch autoencoder neural network, a weakly supervised discriminative neural network, or a combination of the two.] and ¶0019[In some embodiments, based on extracted style embeddings, the style search system determines a style for a query digital image and searches a repository of digital images to identify other digital images with styles similar to the query digital image.])
Regarding Claim 24, Collomosse in view of Wiesel in view of Wade teaches the non-transitory, computer-readable medium of claim 22, Collomosse further discloses wherein the operations further comprise: generating the training image feature vectors using one or more of an InceptionV3 or a Contrastive Language-Image Pre-Training (CLIP) (Fig. 10; ¶0026[In these or other embodiments, the style search system utilizes a contrastive loss function (e.g., a normalized temperature-scaled cross entropy loss function) to learn parameters for fine-grain style coherence with a discriminative neural network]).
Response to Arguments
Applicant’s arguments on pages 13-14 of the remarks filed 04/10/2026, with respect to the previous 35
USC § 101 rejections have been fully considered but are not persuasive.
Applicant argues on page 13 of the remarks that the amended claims do not recite a commercial interaction. Examiner respectfully disagrees. The amended claims are directed to an abstract idea and are specifically categorized under the Certain Methods of Organizing Human Activity Enumerated Grouping (see MPEP 2106). Training by inputting training image feature vectors, and for training item images for portions of data of a combination of data, in a sequence of different types of data to match one or more other data selections in the combination of data; determining an anchor image embedding for an anchor image for anchor data for which is to be displayed; determining one or more image embeddings of one or more respective complementary data for the anchor data; generating, based on a distance between the anchor image embedding and the one or more image embeddings of the one or more respective complementary data for the anchor data, one or more preliminary combinations of data; determining, respective data for the anchor data for each of the one or more preliminary combinations of data based at least in part on respective visual compatibility of the respective data with respective existing data of each of the one or more preliminary combinations of data based on training by inputting the training image feature vectors; and transmitting, for display, information based on determining, the respective data for the anchor data in the amended claim 1 are commercial or legal interactions because they are directed to advertising, marketing or sales activities or behaviors, and business relations, see MPEP 2106.04(a)(2)(II). Furthermore, ¶0003 of the instant specification states that the invention improves recommendation of commercial items to users.
The applicant argues on page 14 of the remarks that the amended claims integrate the abstract idea into a practical application by improving a computer or another technical field. Examiner respectfully disagrees. The mere application of the abstract idea on generic and high-level components such as machine learning module, image encoder, online website, webpage, processors, or memory does not integrate the abstract idea into a practical application or provide a technical improvement. These components are described at a high level and as generic in the instant specification ¶¶0017-0018, ¶0043, and Fig. 2.
Applicant further cites to ¶0077 of the instant specification to provide support for improving machine learning. Examiner respectfully disagrees. ¶0077 of the instant specification merely states that the abstract idea is performed on a high level and generic “machine learning model” and does not describe the process or components used in order to implement a technical improvement.
Accordingly, Examiner maintains that the invention is directed to a judicial exception without
significantly more. The claims recite an abstract idea. This judicial exception is not integrated into a practical application. The claim(s) do not include additional elements that are sufficient to amount to significantly more than the judicial exception. Thus the 35 USC §101 rejections are maintained.
Applicant’s arguments on pages 14-15 of the remarks filed 04/10/2026, with respect to the previous 35 USC § 103 rejections have been fully considered but are moot in view of the new 103 rejection of the amended claims.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to AHOORA LADONI whose email is Ahoora.Ladoni@uspto.gov and telephone number is (703) 756-5617. The examiner can normally be reached M-F 0900–1700 ET.
Examiner interviews are available via telephone, in-person and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/AHOORA LADONI/Examiner, Art Unit 3689
/ANNA MAE MITROS/Examiner, Art Unit 3689
Prosecution Insights