DETAILED ACTION
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 .
Response to Amendment
This Office Action is in response to applicant’s communication filed 30 December 2025, in response to the Office Action mailed 1 October 2025. The applicant’s remarks and any amendments to the claims or specification have been considered, with the results that follow.
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.
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claim(s) 1-3 and 6-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ojha (US 2022/0254071) in view of Kar (US 11,610,115).
As per claim 1, Ojha teaches a computer-implemented method comprising: receiving, at a computing device, a request to generate a target dataset based on a marginal constraint for a source dataset that is associated with a plurality of objects [the system receives a request from a user, at a client (computing) device, to edit, modify, or generate digital content (target dataset), such as digital images (plurality of objects) (paras. 0042-44, etc.) with a cross-domain consistency constraint (paras. 0020-26; etc.) and an expectation for the target distribution (paras. 0062-67; etc.); where cross-domain consistency constraint and the expectation are the marginal constraint], wherein a first object of a plurality of objects occurs at a source frequency in the source dataset [features (first objects) have a source distribution (frequency) in images of the source model data (source dataset) (paras. 0062-67, 0105; etc.)], the marginal constraint indicates a target frequency for the first object that is separate from the source frequency [the system uses a cross-domain consistency constraint and an expectation for the target distribution to constrain the target feature distribution in samples generated by the target model (paras. 0020-26, 0062-67; etc.)]; accessing, at the computing device, a source generative model that includes a first set of modules including a first module and a second module, wherein each module of the first set of modules is trained on the source dataset [a source generative adversarial network (GAN) that includes a discriminator and generator (first set of first and second modules) trained on source data to generate digital images (paras. 0019-21, 0027-31; figs. 9, 10A; etc.)]; updating, at the computing device, the second module based on the marginal constraint [the system utilizes GAN-to-GAN translation to adapt the generator (second module) for the target GAN (paras. 0019-21; fig. 3; etc.) using a cross-domain consistency constraint and an expectation for the target distribution to constrain the target feature distribution in samples generated by the target model (paras. 0020-26, 0062-67; etc.)]; generating, at the computing device, an adapted generative model that includes a second set of modules including the first module and the updated second module [the system utilizes GAN-to-GAN translation to adapt the generator (second module) for the target GAN (adapted generative model) (paras. 0019-21; fig. 3; etc.) using a cross-domain consistency constraint and an expectation for the target distribution to constrain the target feature distribution in samples generated by the target model (paras. 0020-26, 0062-67; etc.); where the pretrained discriminator and the updated generator of the target GAN are the second set of modules including the first module (discriminator) and updated second module (generator)]; and generating, at the computing device, the target dataset based on the adapted generative model, wherein the first object occurs at the target frequency in the target dataset [the system utilizes GAN-to-GAN translation to adapt the generator (second module) for the target GAN (adapted generative model) (paras. 0019-21; fig. 3; etc.) using a cross-domain consistency constraint and an expectation for the target distribution to constrain the target feature distribution (first object occurs at the target frequency) in samples generated by the target model (paras. 0020-26, 0062-67; etc.)].
While Ojha teaches adapting a GAN using constraints on the target distribution of the target GAN (see above) including constraining pairwise distances and/or relative feature distances between source and target (see, e.g., Ojha: paras. 0033-35; etc.), it has not been relied upon for teaching that the source dataset encodes a set of co-occurrence frequencies for a plurality of object pairs of the plurality of objects and the target dataset encodes the set of co-occurrence frequencies for the plurality of object pairs.
Kar teaches the source dataset encodes a set of co-occurrence frequencies for a plurality of object pairs of the plurality of objects and the target dataset encodes the set of co-occurrence frequencies for the plurality of object pairs [an encoder (col. 7, line 45 to col. 8, line 6; fig. 2A; etc.) encodes sets of probabilities/distributions of co-occurrences of objects in a scene/image (col. 4, lines 4-38; fig. 1; etc.) and trains the model to match the generated distribution to a target distribution (col. 8, line 63 to col. 9, line 14; fig. 2B; etc.); for the source and target datasets and generator models of Ojha, above].
Ojha and Kar are analogous art, as they are within the same field of endeavor, namely adapting generative models, including for generating synthetic image samples.
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to include encoding co-occurrence distributions of objects in a scene for the generative model, as taught by Kar, for the source and target generative model encoders in the GAN-to-GAN translation system taught by Ojha.
Kar provides motivation as [encoding the co-occurrences of objects in the scenes to be generated allows easier and more accurate rendering of synthetic images by the generative model(s) (col. 4, lines 24-58; etc.)].
As per claim 2, Ojha/Kar teaches wherein updating the second module comprises: updating the second module based on a constrained divergence objective function that indicates a variational distance between the first and second modules [utilizing a non-saturating GAN objective, the GAN translation system learns parameters in accordance with an adversarial loss (constrained divergence objective) between a generator neural network G (second module) and a discriminator neural network D (first module – e.g., indicating a likelihood that a generated digital image is real or fake) (Ojha: para. 0062, etc.)].
As per claim 3, Ojha/Kar teaches wherein the source generative model and the adapted generative model are latent variable models [the disclosed systems preserve relative similarities and differences between digital images in the source domain by ensuring cross-domain distance consistency between feature vectors generated by the source and target adversarial neural networks from the same latent vectors (Ojha: para. 0004, etc.); thus, latent variable models].
As per claim 6, Ojha/Kar teaches wherein the first module is associated with the set of co-occurrence frequencies for the plurality of object pairs [features (first objects) have a source distribution (frequency) in images of the source model data (source dataset) (Ojha: paras. 0062-67, 0105; etc.) using an encoder (Kar: col. 7, line 45 to col. 8, line 6; fig. 2A; etc.) that encodes sets of probabilities/distributions of co-occurrences of objects in a scene/image (Kar: col. 4, lines 4-38; fig. 1; etc.) and trains the model to match the generated distribution to a target distribution (Kar: col. 8, line 63 to col. 9, line 14; fig. 2B; etc.); and where the discriminator neural network (first module), in competition with the generator neural network, analyzes a generated digital image from the generator neural network to determine whether the generated digital image is real (e.g., from a set of stored digital images) or fake (e.g., not from the set of stored digital images) (Ojha: para. 0030; etc.); so, the discriminator is associated with the set of co-occurrence frequencies].
As per claim 7, Ojha/Kar teaches wherein the second module is associated with the target frequency of the first object [features (first objects) have a source distribution (frequency) in images of the source model data (source dataset) (Ojha: paras. 0062-67, 0105; etc.) using an encoder (Kar: col. 7, line 45 to col. 8, line 6; fig. 2A; etc.) that encodes sets of probabilities/distributions of co-occurrences of objects in a scene/image (Kar: col. 4, lines 4-38; fig. 1; etc.) and trains the model to match the generated distribution to a target distribution (Kar: col. 8, line 63 to col. 9, line 14; fig. 2B; etc.)].
As per claim 8, Ojha/Kar teaches wherein updating the second module comprises: receiving, at the computing device, the source distribution; and training, at the computing device, the source generative model based on the received source distribution [the disclosed system GAN-to-GAN translation preserves diversity information from a source generative adversarial neural network trained on a large source domain (Ojha: para. 0004, 0019-21, 0027-31; figs. 9, 10A; etc.) where features (first objects) have a source distribution (frequency) in images of the source model data (source dataset) (Ojha: paras. 0062-67, 0105; etc.) using an encoder (Kar: col. 7, line 45 to col. 8, line 6; fig. 2A; etc.) that encodes sets of probabilities/distributions of co-occurrences of objects in a scene/image (Kar: col. 4, lines 4-38; fig. 1; etc.)].
As per claim 9, Ojha/Kar teaches wherein training the source generative model comprises: training, at the computing device, a neural network that implements the source generative model [the disclosed system GAN-to-GAN translation preserves diversity information from a source generative adversarial neural network trained on a large source domain (Ojha: para. 0004, 0019-21, 0027-31; figs. 9, 10A; etc.) where features (first objects) have a source distribution (frequency) in images of the source model data (source dataset) (Ojha: paras. 0062-67, 0105; etc.) using an encoder (Kar: col. 7, line 45 to col. 8, line 6; fig. 2A; etc.) that encodes sets of probabilities/distributions of co-occurrences of objects in a scene/image (Kar: col. 4, lines 4-38; fig. 1; etc.); where the encoders can be neural networks (Ojha: paras. 0030, 0046, 0081; Kar: col. 5, lines 51-65; col. 8, lines 32-62; etc.)].
As per claim 10, Ojha/Kar teaches providing, from the computing device to another computing device that transmitted the request to generate the target distribution, the target distribution [the requests and resulting model outputs can be transmitted between remote computer systems (Ojha: para. 0123; Kar: col. 18, lines 20-39; etc.)].
As per claim 11, see the rejection of claim 1, above, wherein Ojha/Kar also teaches a computing system, comprising: one or more processors; and one or more non-transitory computer-readable media that, when executed by the one or more processors, cause the computing system to perform operations, the operations comprising: [the method] [the system can be implemented via instructions stored on a computer-readable medium and executable by processors of one or more computing devices (Ojha: para. 0099, fig. 1; Kar: col. 18, lines 20-39; etc.)].
As per claim 12, see the rejection of claim 2, above.
As per claim 13, Ojha/Kar teaches wherein the source generative model is at least one of a latent variable model, an autoregressive model, or an energy-based model [the disclosed systems preserve relative similarities and differences between digital images in the source domain by ensuring cross-domain distance consistency between feature vectors generated by the source and target adversarial neural networks from the same latent vectors (Ojha: para. 0004, etc.); thus, latent variable models].
As per claim 14, see the rejections of claims 6-7, above.
As per claim 15, see the rejection of claim 8, above.
As per claim 16, see the rejection of claim 9, above.
As per claim 17, see the rejection of claim 10, above.
As per claim 18, see the rejection of claim 1, above, wherein Ojha/Kar also teaches one or more tangible non-transitory computer-readable media storing computer-readable instructions that when executed by one or more processors cause the one or more processors to perform operations, the operations comprising: [the method] [the system can be implemented via instructions stored on a computer-readable medium and executable by processors of one or more computing devices (Ojha: para. 0099, fig. 1; Kar: col. 18, lines 20-39; etc.)].
As per claim 19, see the rejection of claim 2, above.
As per claim 20, see the rejection of claim 13, above.
Claim(s) 4 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ojha and Kar as applied to claim 1 above, and further in view of Li (US 2022/0208355).
As per claim 4, Ojha/Kar teaches the method of claim 1, as described above.
Ojha/Kar has not been relied upon for teaching wherein the source generative model and the adapted generative model are autoregressive models.
Li teaches wherein the source generative model and the adapted generative model are autoregressive models [an auto regressive model can be integrated into a GAN to generate synthetic images (para. 0090, etc.)].
Ojha/Kar and Li are analogous art, as they are within the same field of endeavor, namely using GANs for image synthesis.
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to implement an autoregressive model for the generative model of the GAN, as taught by Li, for the generative model of the GANs in the system taught by Ojha/Kar.
Li provides motivation as [using the auto regressive model to achieve image synthesis and segmentation allows the Gan to attend to internal model states to efficiently find global, long-range dependencies within the internal representations of the images (para. 0090, etc.)].
Claim(s) 5 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ojha and Kar as applied to claim 1 above, and further in view of Xie (US 2022/0108426).
As per claim 5, Ojha/Kar teaches the method of claim 1, as described above.
Ojha/Kar has not been relied upon for teaching wherein the source generative model and the adapted generative model are energy-based models.
Xie teaches wherein the source generative model and the adapted generative model are energy-based models [an energy-based image-to-image translation model can be used for a GAN (abstract; para. 0034; etc.)].
Ojha/Kar and Xie are analogous art, as they are within the same field of endeavor, namely using GANs for image synthesis.
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to implement an energy-based model for the generative model of the GAN, as taught by Xie, for the generative model of the GANs in the system taught by Ojha/Kar.
Xie provides motivation as [energy-based models (EBM) achieve better performance on image generation, provide better translation quality, and improve memory and time efficiency (abstract; paras. 0034; etc.)].
Response to Arguments
The objections to claims 9 and 11-17 have been withdrawn due to the amendments filed.
Applicant's arguments filed 30 December 2025 have been fully considered but they are not persuasive.
Applicant argues that the cited art does not teach any “marginal constraint [that] indicates a target frequency for the first object.”
However, Ojha teaches features (first objects) have a source distribution (source frequency) in images of the source model data (source dataset) (paras. 0062-67, 0105; etc.), and the system uses a cross-domain consistency constraint and an expectation for the target distribution (target frequency) to constrain the target feature distribution in samples generated by the target model (paras. 0020-26, 0062-67; etc.); while Kar teaches an encoder (col. 7, line 45 to col. 8, line 6; fig. 2A; etc.) encodes sets of probabilities/distributions of co-occurrences of objects in a scene/image (col. 4, lines 4-38; fig. 1; etc.) and trains the model to match the generated distribution to a target distribution (col. 8, line 63 to col. 9, line 14; fig. 2B; etc.); where the target feature distribution in the samples of the target dataset is the target frequency for the first object.
Conclusion
The following is a summary of the treatment and status of all claims in the application as recommended by M.P.E.P. 707.07(i): claims 1-20 are rejected.
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Talbot (Learning Translations from Comparable Corpora, 2003, pgs. 1-86) – discloses a system for training (generative) translation models from comparable datasets.
Al-Turki (US 10,839,269) – discloses a system for visual domain adaptation utilizing Generative Adversarial Distribution Matching (GADM).
Deasy (US 2021/0383538) – discloses a system/method for training a cross-modality model to synthesize images resembling a target distribution, including using a gray level co-occurrence matrix.
The examiner requests, in response to this Office action, that support be shown for language added to any original claims on amendment and any new claims. That is, indicate support for newly added claim language by specifically pointing to page(s) and line number(s) in the specification and/or drawing figure(s). This will assist the examiner in prosecuting the application.
When responding to this office action, Applicant is advised to clearly point out the patentable novelty which he or she thinks the claims present, in view of the state of the art disclosed by the references cited or the objections made. He or she must also show how the amendments avoid such references or objections. See 37 CFR 1.111(c).
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to GEORGE GIROUX whose telephone number is (571)272-9769. The examiner can normally be reached M-F 10am-6pm.
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.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Omar Fernandez Rivas can be reached at 571-272-2589. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
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/apply/patent-center for more information about Patent Center and 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.
/GEORGE GIROUX/Primary Examiner, Art Unit 2128