DETAILED ACTION
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
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and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114.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 submissions filed on 2/26/2026, 4/1/2026, 7/10/2026 have been entered.
Claims 1-20 have been presented for examination based on the application filed on 2/26/2026.
Claim 1 is newly objected to minor informality.
Specification submitted on 2/26/2026 is entered and the objection to specification is withdrawn upon further consideration.
Claims 1-20 remain rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph.
Claims 1-20 remain rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement.
Claims 1-20 remain rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the enablement requirement.
This action is made Non-Final.
---- This page is left blank after this line ----
Response to applicant’s Arguments
Applicant’s remarks and mapping for new limitations are appreciated and the rejection in view of citations is updated.
(Argument 1) Applicant has argued in Remarks Pg.9-10:
PNG
media_image1.png
294
642
media_image1.png
Greyscale
(Response 1) Applicant has summarily restated the claim. No mapping in the specification other than general citation for support in [0032]-[0037] is shown. Please see updated mapping in view of the amendments. Applicants are encouraged to request an interview before responding to this action to clarify their position.
---- This page is left blank after this line ----
Claim Objections
Claim 1 is objected to because of the following informalities:
MPEP 714(II)(C)(B) states:
(B) Markings to Show the Changes: All claims being currently amended must be presented with markings to indicate the changes that have been made relative to the immediate prior version.
Claim 1 last determining step does not show all the markings and adds new limitations without proper markings.
PNG
media_image2.png
303
666
media_image2.png
Greyscale
The dashed red line above is not appropriately underlined in the new amendment as new limitation. Appropriate correction is required.
---- This page is left blank after this line ----
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1-20 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim 1 recites:
(Currently amended) A computer-implemented method comprising:
receiving a first plurality of time sequenced spatial data images at a first resolution;
determining from one or more of the plurality of spatial data images one or more physics laws applicable to the one or more spatial data images;
subdividing each of the one or more plurality of spatial data images into a plurality of small spatial region images;
applying each of the physics laws to each small spatial region image by applying a regional physics law loss function [A] and a neural network [B];
generating subsections for each small spatial region image, each subsection having a value predicted [C] by the neural network;
determining a most applicable regional physics law [D] for each small spatial region image based on the regional physics law loss function and corresponding network model [E] which produces a smallest loss function[F] for the predicted value;
generating a second higher-resolution image than the first resolution by applying the neural network for the most applicable regional physics law to the first plurality of time sequenced images.
As per [A], The specification does not disclose what is the physics law loss function. While the specification discusses the loss function in context of neural network with generator and discriminator, it does not discuss the physics law loss function. Loss function are evaluations of difference between neural network output (like generator neural network – GAN output) against some other output (like output of discriminator neural network)1. Limited support in specification recited below fails to show the what is physics law loss function:
PNG
media_image3.png
642
628
media_image3.png
Greyscale
As per [B] in
applying each of the physics laws to each small spatial region image by applying a regional physics law loss function [A] and a neural network [B];
, it is unclear what is the use of neural network in applying the physics law. Applying the physics law is performing exact physics-based simulation which is different than using a neural network. If the intent of this step is to train the neural network based on physics-based simulation, then it should be stated clearly.
As per [C] in
generating subsections for each small spatial region image, each subsection having a value predicted [C] by the neural network;
, its unclear what value is predicted by the neural network?
Specification [0036] states:
[0036] In process 208, IE program 112 subdivides each section into two or more subsections. For example, for each pixel or data element in an image, IE program 112 may double the pixel count creating a 2×2 pixel block for each 1×1 pixel. IE program 112 may select any scaling factor since, as previously discussed, IE program 112 does not utilize any pixel interpolation as prior image enhancement solutions have utilized. In process 210, IE program 112 trains neural network 114 based on the identified applicable physics laws from process 206. In some scenarios, IE program 112 is supervised by a user while training neural network 114 with a training data set. In other scenarios, IE program 112 performs unsupervised training on neural network 114. In such scenarios where an adversarial neural network is deployed, IE program 112 may automatically retrain neural network 114 if the loss function exceeds a threshold value for a predetermined number of pixels (e.g., over half of the subpixels have minimized loss functions above a certain error rate).
Is the predicted value a variable in the physics law (see specification [0023] table and [0024], variables like density/pressure)? Or are these pixel value (See specification [0036] above). This makes it further unclear what is the output of physics model, what are the inputs to train the neural network (if any)? what does neural network output?
As per [D] in
determining a most applicable regional physics law [D] for each small spatial region image based on the regional physics law loss function and corresponding network model [E] which produces a smallest loss function [F] for the predicted value;
, its unclear for plurality of reasons.
(1) The phrase “a most applicable regional physics law” is a relative term, even when considered in view of annotated (F). Specifically the even if read like “… a most applicable regional physics law… which produces a smallest loss function…”, there are no metes and bounds on how the smallest loss (new instance not related physics law loss function) is computed.
(2) Specification [0023] shows plurality of physics type models. What is does not show is variants of any one physics type model from which “most applicable regional physics law” can be chosen based on loss function. As an example If we pick Incompressible Ideal Fluid Flow as an example for tidal wave flooding coastal region as model, other models like Electromagnetism model cannot be chosen as alternate (as it does not compute tidal wave based flooding). So there are not plurality of versions of Incompressible Ideal Fluid Flow to choose to begin with, let alone determining a most applicable regional physics law. In alternate also see specification [0020]-[0021] also.
As per [E], association of physics law loss function and network model is unclear.
As per [F], it is unclear (1) what is the smallest loss function, as no specific loss function is disclosed in the specification. Further it is unclear (2) if the loss function referred here is the physics law loss function or something different.
Claims 1, 8 and 15 suffer from above deficiency and are rejected as above.
Dependent claims 2-7, 9-14 and 16-20 do not cure this deficiency and are rejected likewise also.
---- This page is left blank after this line ----
Claim Rejections - 35 USC § 112(a) Written Description Requirement
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
Claims 1-20 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for pre-AIA the inventor(s), at the time the application was filed, had possession of the claimed invention.
MPEP 2161.01 states:
For instance, generic claim language in the original disclosure does not satisfy the written description requirement if it fails to support the scope of the genus claimed. Ariad, 598 F.3d at 1349-50, 94 USPQ2d at 1171 ("[A]n adequate written description of a claimed genus requires more than a generic statement of an invention’s boundaries.") (citing Eli Lilly, 119 F.3d at 1568, 43 USPQ2d at 1405-06); Enzo Biochem, Inc. v. Gen-Probe, Inc., 323 F.3d 956, 968, 63 USPQ2d 1609, 1616 (Fed. Cir. 2002) (holding that generic claim language appearing in ipsis verbis in the original specification did not satisfy the written description requirement because it failed to support the scope of the genus claimed); Fiers v. Revel, 984 F.2d 1164, 1170, 25 USPQ2d 1601, 1606 (Fed. Cir. 1993) (rejecting the argument that "only similar language in the specification or original claims is necessary to satisfy the written description requirement").
Exemplary claim 1 recites:
Claim 1 recites:
(Currently amended) A computer-implemented method comprising:
receiving a first plurality of time sequenced spatial data images at a first resolution;
determining from one or more of the plurality of spatial data images one or more physics laws applicable to the one or more spatial data images;
subdividing each of the one or more plurality of spatial data images into a plurality of small spatial region images;
applying each of the physics laws to each small spatial region image by applying a regional physics law loss function [A] and a neural network [B];
generating subsections for each small spatial region image, each subsection having a value predicted by the neural network;
determining a most applicable regional physics law [C] for each small spatial region image based on the regional physics law loss function and corresponding network model [E] which produces a smallest loss function[F] for the predicted value;
generating a second higher-resolution image than the first resolution by applying the neural network for the most applicable regional physics law to the first plurality of time sequenced images.
As per [A]: Specification does not show how the physics law and neural network are applied to same small spatial region. Further the specification has no disclosure (1) what is the physics law loss function (2) how it is associated with neural network. See citation from specification below.
PNG
media_image3.png
642
628
media_image3.png
Greyscale
As per [A] & [B] for the limitation applying each of the physics laws to each small spatial region image by applying a regional physics law loss function and a neural network, the specification, other than stating a general adversarial neural network or GANs is used for training, provides no details to support or describe training the process of training, such that it may be provided to perform the generating (as annotated in [C]). See Specification:
PNG
media_image4.png
424
755
media_image4.png
Greyscale
The written description requirement is what the applicant claims. In this case the by applying a regional physics law loss function.
As can be seen from [0027] there is no mention of loss function in context of training with the physics law or providing the neural network based on loss function. E. g. specification does not describe the loss function for any of the cited physics models (as in specification ¶[0023]).
Loss function is only discussed in general in view of GAN (in specification ¶[0029], [0037]-in context of using the trained model), but there is no mention of reference to physics law during the training process [Emphasis added]. Therefore other than the claim, the specification does not describe applying each of the physics laws to each small spatial region image by applying a regional physics law loss function and a neural network.
As per [C] for the limitation determining a most applicable regional physics law for each small spatial region image based on the regional physics law loss function and corresponding network model which produces a smallest loss function for the predicted value, the specification fails to describe the what is the most applicable regional physics law. Specification [0026] states:
[0026] In some scenarios, the original image data sets from satellite may not be enough in size and quality to train the neural network and simulations based on the dominant physics laws can be generated using the physics laws, where the training images can be generated at arbitrary spatial and temporal resolution. In such scenarios, IE program 112 generates simulated images based on a physics simulation using physics module 113, using the simulated images as training data for neural network 114.
How is the dominant (or most applicable) physics law determined? The specification does not provide disclosure for determining how dominant/most applicable is determined. Specification ¶[0037] states:
PNG
media_image5.png
104
552
media_image5.png
Greyscale
Since it is unclear where the specification discloses computing the loss function during the providing a neural network determining is also not supported in view of specification ¶[0037]. Claims 1, 8 and 15 suffer from above deficiency and are rejected as above. Dependent claims 2-7, 9-14 and 16-20 do not cure this deficiency and are rejected likewise also.
Claim Rejections - 35 USC § 112 Enablement
Claims 1-20 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the enablement requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to enable one skilled in the art to which it pertains, or with which it is most nearly connected, to make and/or use the invention..
It is noted from MPEP that while applying In re Wands test that While the analysis and conclusion of a lack of enablement are based on the factors discussed in MPEP § 2164.01(a) and the evidence as a whole, it is not necessary to discuss each factor in the written enablement rejection. The language should focus on those factors, reasons, and evidence that lead the examiner to conclude that the specification fails to teach how to make and use the claimed invention without undue experimentation, or that the scope of any enablement provided to one skilled in the art is not commensurate with the scope of protection sought by the claims.
Applying In re Wands, 858 F.2d 731, 737, 8 USPQ2d 1400, 1404 (Fed. Cir. 1988) factors:
(A) The breadth of the claims - The focus of the examination inquiry is whether everything within the scope of the claim is enabled (MPEP 2164.08). First, in this case applicant has removed the limitation stating “applying each of the physics laws to each small spatial region image by applying a regional physics law loss function and a neural network”. This would have been important as it appears that based on the solving the physics law would create the simulated data that would appear to pertinent to providing the trained neural network. Specification [0026] states:
[0026] In some scenarios, the original image data sets from satellite may not be enough in size and quality to train the neural network and simulations based on the dominant physics laws can be generated using the physics laws, where the training images can be generated at arbitrary spatial and temporal resolution. In such scenarios, IE program 112 generates simulated images based on a physics simulation using physics module 113, using the simulated images as training data for neural network 114.
Therefore now the claim and the disclosure fails to provide training (missing step) and subsequently the now claimed step of providing a neural network to apply each of the physics laws to each small spatial region image by applying a regional physics law loss function. A generic statement the neural network is trained (Specificaiton [0036]) is not sufficient to show how it is trained for each of the claimed physics models (Specification [0023]).
(B) & (C) The nature of the invention & The state of the prior art - The nature of the invention becomes the backdrop to determine the state of the art and the level of skill possessed by one skilled in the art. The state of the prior art is what one skilled in the art would have known, at the time the application was filed, about the subject matter to which the claimed invention pertains. The relative skill of those in the art refers to the skill of those in the art in relation to the subject matter to which the claimed invention pertains at the time the application was filed (MPEP 2164.05(a)). In this case the Wang (NPL “Physics-Informed Neural Network Super Resolution
for Advection-Diffusion Models”, 2020 cited by applicant) is the closest prior art of record it does not appear to teach the claimed invention as it does not explicitly teach “providing a neural network to apply each of the physics laws to each small spatial region image by applying a regional physics law loss function; generating subsections for each small spatial region image, each subsection having a value predicted by the neural network; determining the most applicable regional physics law for each small spatial region image based on the regional physics law loss function for the predicted value;”. Wang only solves one physics model for the tile (Wang §3.1). The training of the neural network the loss computation is between in the high resolution (HR) and low resolution (LR) image. The instant claim only has one resolution image therefore Wang would not teach training in the same context as claimed invention. Similar teaching is presented Onishi et al (NPL “Super-Resolution Simulation for Real-Time Prediction of Urban Micrometeorology”, 2019) which also uses both the HR and LR images for training (Onishi §2.2).
Further, “The state of the prior art is also related to the need for working examples in the specification.” (MPEP 2164.05(a)) which is not present in this case.
(D) The level of one of ordinary skill - MPEP 2164.05(b) states “The relative skill of those in the art refers to the skill of those in the art in relation to the subject matter to which the claimed invention pertains at the time the application was filed. Here Wang and Onishi show the ordinary skill in the art. Further patent documents show the current state of the art using physics based models and related to super resolution image creation:
US 20230214661 A1 2023-07-06 Perdikaris; Paris Georgios et al.
US 20190114744 A1 2019-04-18 Albrecht; Conrad M. et al.
US 20220215601 A1 2022-07-07 Massanes Basi; Francesc dAssis
US 20180075581 A1 2018-03-15 Shi; Wenzhe et al. (teach use of GAN, superiority over MSE – [0037]).
Many other relevant prior arts are cited on PTO892 to show relevant prior arts as state of the art.
(E) The level of predictability in the art - The “predictability or lack thereof” in the art refers to the ability of one skilled in the art to extrapolate the disclosed or known results to the claimed invention. If one skilled in the art can readily anticipate the effect of a change within the subject matter to which the claimed invention pertains, then there is predictability in the art. On the other hand, if one skilled in the art cannot readily anticipate the effect of a change within the subject matter to which that claimed invention pertains, then there is lack of predictability in the art. Accordingly, what is known in the art provides evidence as to the question of predictability. In particular, the court in In re Marzocchi, 439 F.2d 220, 223-24, 169 USPQ 367, 369-70 (CCPA 1971) (MPEP 2164.03). As seen from the citations/prior art computation of loss function is hallmark of physics-informed neural networks. None of the prior arts providing a neural network to apply each of the physics laws to each small spatial region image by applying a regional physics law loss function; generating subsections for each small spatial region image, each subsection having a value predicted by the neural network; determining the most applicable regional physics law for each small spatial region image based on the regional physics law loss function for the predicted value. Without knowing how the loss function of each pixel for each of the models is computed for the same small spatial region images (based on how the neural network is trained for each of the physics model and subsequent determination of loss function) and how it is compared to select the most applicable physics model, the teachings of such would be apparent from the art below.
Most of the prior art show increasing the resolution based on neural network:
US 20240185383 A1 METHOD OF PROCESSING IMAGE BASED ON SUPER-RESOLUTION WITH DEEP LEARNING AND METHOD OF PREDICTING CHARACTERISTIC OF SEMICONDUCTOR DEVICE USING THE SAME
US 20070116381 A1 Method for deformable registration of images
US 20230359888 A1 OPTIMIZATION AND DIGITAL TWIN OF CHROMATOGRAPHY PURIFICATION PROCESS USING PHYSICS-INFORMED NEURAL NETWORKS
US 20230196745 A1 ADVERSARIAL ATTACK METHOD FOR MALFUNCTIONING OBJECT DETECTION MODEL WITH SUPER RESOLUTION
US 20200293594 A1 PHYSICS INFORMED LEARNING MACHINE
US 20180121389 A1 COGNITIVE INITIALIZATION OF LARGE-SCALE ADVECTION-DIFFUSION MODELS
US 20230195949 A1 NEURAL OPERATORS FOR FAST WEATHER AND CLIMATE PREDICTIONS
US 20240404001 A1 RAPID RECONSTRUCTION OF HIGH RESOLUTION IMAGES FROM LOWER RESOLUTION IMAGES
US 20250299442 A1 DYNAMIC IMAGE RECONSTRUCTION USING PHYSICS-INFORMED NEURAL NETWORK
US 20230062600 A1 ADAPTIVE DESIGN AND OPTIMIZATION USING PHYSICS-INFORMED NEURAL NETWORKS
US 20230214661 A1 COMPUTER SYSTEMS AND METHODS FOR LEARNING OPERATORS
US 20190114744 A1 ENHANCING OBSERVATION RESOLUTION USING CONTINUOUS LEARNING
US 20200126205 A1 IMAGE PROCESSING METHOD, IMAGE PROCESSING APPARATUS, COMPUTING DEVICE AND COMPUTER-READABLE STORAGE MEDIUM
US 20210110531 A1 PHYSICS-CONSTRAINED NETWORK AND TRAINING THEREOF
US 20250036935 A1 COUPLED PHYSICS-INFORMED NEURAL NETWORK FOR SOLVING DISPLACEMENT DISTRIBUTION OF BOUNDED VIBRATION STRING UNDER UNKNOWN EXTERNAL DRIVING FORCE
US 20200082041 A1 ESTIMATING PHYSICAL PARAMETERS OF A PHYSICAL SYSTEM BASED ON A SPATIAL-TEMPORAL EMULATOR
US 20220215601 A1 Image Reconstruction by Modeling Image Formation as One or More Neural Networks
US 20220414832 A1 X-RAY IMAGING RESTORATION USING DEEP LEARNING ALGORITHMS
US 20210089275 A1 Physics Informed Neural Network for Learning Non-Euclidean Dynamics in Electro-Mechanical Systems for Synthesizing Energy-Based Controllers
US 20180075581 A1 SUPER RESOLUTION USING A GENERATIVE ADVERSARIAL NETWORK
US 20240005065 A1 SUPER RESOLVED SATELLITE IMAGES VIA PHYSICS CONSTRAINED NEURAL NETWORK
US 20240193414 A1 PHYSICS AWARE TRAINING FOR DEEP PHYSICAL NEURAL NETWORKS
US 10290083 B2 Multiple phase method for image deconvolution
US 20210271934 A1 Method and System for Predicting Wildfire Hazard and Spread at Multiple Time Scales
US 10970820 B2 System and method for deep learning image super resolution
US 12124961 B2 System for continuous update of advection-diffusion models with adversarial networks
(G) The existence of working examples - MPEP 2164.02 states “When considering the factors relating to a determination of non-enablement, if all the other factors point toward enablement, then the absence of working examples will not by itself render the invention non-enabled.” There is no working example shown in the specification that takes even one model from beginning to the end to show how the high resolution images are generated starting from model(s) (singular and plural) being evaluated and then used to train the neural networks (singular/plural one for each network) and then neural network choosing one model based on comparison of predicted image.
(H) The quantity of experimentation needed to make or use the invention based on the content of the disclosure - MPEP 2164.06(a) related to ELECTRICAL AND MECHANICAL DEVICES OR PROCESSES - gives guidance that drawings by block diagrams with functional labels, was held to be nonenabling in In re Gunn, 537 F.2d 1123, 1129, 190 USPQ 402, 406 (CCPA 1976).- The block diagrams do not resolve all the plurality of physics models compute the same loss function when each of them may evaluate different types of variables. Further none of the diagrams show the generation of predicted image, let alone determination of the difference so that most applicable model is chosen for generating the higher resolution image. See Fig. 2 which does not resolve the issue above. The specification [0023] discloses plurality of physics based models however fails to show even one complete embodiment. The experimentation needed to provide (trained) neural network model for any one of the model may have been possible (see factor A above for one such possible model), but would definitely require heavy undue experimentation to generate/provide trained network models for each of the models (as in specification [0023] & associated table), when flow for even one model is not disclosed in the specification. This is important as the selection of the most applicable regional physics law is based on the regional physics loss function for each one of the model (See claim 1 last determining limitation). The loss function evaluation, is based on trained/provided neural network specific to a physics law (model). Without output from each of neural network (at least one for each of the physics law/model) the output cannot be compared to determined to determine most applicable. In re Wands test performed above also apply to claims 8 and 15. Dependent claims are rejected with similar rationale for inheriting the deficiencies. Dependent claims 2-7, 9-14 and 16-20 do not cure this deficiency and are rejected likewise also.
---- This page is left blank after this line ----
Communication
Any inquiry concerning this communication or earlier communications from the examiner should be directed to AKASH SAXENA whose telephone number is (571)272-8351. The examiner can normally be reached Mon-Fri, 7AM-3:30PM.
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, RYAN PITARO can be reached on (571) 272-4071. 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.
AKASH SAXENA
Primary Examiner
Art Unit 2188
/AKASH SAXENA/Primary Examiner, Art Unit 2188 Friday, July 24, 2026
1 See Specification [0029]-[0033]; [0032] "...Neural network 114, as discussed herein, generates a prediction for each subpixel based on a trained GAN network based on minimizing a loss function between a generator neural network and a discriminator neural network...."
Prosecution Insights