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
The amendment filed 11/21//2025 has been entered. No Claims have been amended. No Claims have been/remained canceled. Claims 1-20 remain pending in the application.
Response to Arguments
Regarding Applicant’s arguments, on page 9-15 of the remark filed on 11/21/2025, on the limitations of independent Claims 1: “receiving first image data representing a first plurality of images captured using a first camera; training a first machine learning model using the first plurality of images to identify first features resulting from physical defects corresponding to one or more optical elements of the first camera or pixels of an image sensor of the first camera;.”, arguments are not persuasive.
Applicant argues on Pages 11 that the cited references fail to teach machine learning model trained using a plurality of images captured using a first camera to identify features resulting from physical defects corresponding to optical elements or pixels of an image sensor of the first camera. Applicant’s interpretation of the reference has been noted; however, examiner respectfully disagrees. Stoppe teaches on Par. (0009) one or more defects of test object and trained machine learning model of optical effects of test object. Stoppe describes on Par. (0012) that the test object would be lens of camera.
Applicant further argues on pages 12-13 that the cited references do not teach “processing the second image using the first machine learning model to determine a first probability that a first image was captured using the first camera”. applicant’s interpretation of the reference has been noted; however, examiner respectfully disagrees. Hojjati teaches on Par. (0111) a threshold value with a percentage to determine a likelihood an image was altered using a machine learning model. Hojjati discloses a plurality of images being processed and validated and on Par. (0059-0060) teaching verifying an image was directly from a camera.
Applicant further argues on page 13 first paragraph that the cited references do not teach using images captured using a camera to train machine learning model trained to identify features resulting from physical defects corresponding to optical elements or pixels of an image sensor of the first camera. Applicant’s interpretation of the reference has been noted; however, examiner respectfully disagrees. Stoppe teaches on Par. (0009) one or more defects of test object and trained machine learning model of optical effects of test object. Stoppe describes on Par. (0012) that the test object would be lens of camera. Stoppe teaches these images to be captured by a first camera on Par. (0011-0014). Therefore, the rejection is maintained.
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 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.
Claims 1-3, is/are rejected under 35 U.S.C. 103 as being unpatentable over Hojjati et al. (U.S Pub. No. 20230344650, hereinafter referred to as “Hojjati”), Baldwin et al. (U.S Pub. No. 20230409756, hereinafter referred to as “Baldwin”) further in view of Stoppe et al. (U.S Pub. No. 20210279858, hereinafter referred to as “Stoppe”)
In regards to Claim 1, Hojjati teaches a computer-implemented method comprising:
receiving first image data representing a first plurality of images captured using a first camera; (Par. (0060-0061); receive image and corresponding image data directly from camera)
verifying, using the public key, that the first image data was digitally signed using a private key corresponding to the public key; (Par. (0027); signing token In image), (Par. (0044); verify image for signed token of the image), (Par. (0066); image validation corresponding to public key), (Par. (0073); signed token in image data is verified), (Par. (0086); determines if validated image has signed token image data), (Par. (0024); signing the image with private key)
receiving a first request to verify second image data corresponding to the public key; (Par. (0039); receiving request to validate image (plurality of images that are validated for consumers)), (Par. (0029); receiving request to validate image)
verifying, using the public key, that the second image data was digitally signed using the private key; (Par. (0041); verification of digitally signed token in image), (Par. (0024); signing token and image executed using private key)
processing the second image using the first machine learning model to determine a first probability that a first image was captured using the first camera. ((Par. (0111); threshold value of 75% is verified to determine likelihood of image being altered and value is checked to see if value satisfies probability threshold), (Par. (0111); image is verified and satisfies threshold value of particular likelihood that image is not altered), (Par. (0013 and 0037); determine trustworthiness of image), (Par. (0059-0060); verification of image corresponding to image being directly from camera), (Par. (0111-0113); machine learning model) (Par. (0019 further explains); embed image with digitally signed token), (Par. (0021 further explains); second image (caption) can be included with digitally signed token metadata and image considered from original source)), (Par. (0024) further explains); generated signed token is associated with publisher that is verified, determining if image is taken from publisher/camera))
Hojjati does not explicitly teach receiving a public key corresponding to a first user device; training a first machine learning model using the first plurality of images to identify first features, the first features resulting from physical defects corresponding to one or more optical elements of the first camera or pixels of an image sensor of the first camera; associating the first machine learning model with the public key; retrieving the first machine learning model using the public key; and
Wherein Baldwin teaches receiving a public key corresponding to a first user device; (Par. (0067); receiving request with public key associated with message from computing device), (Par. (0250); providing of the public key for attestation), (Par. (0062); additional information includes public key) (Par. (0133); additional information that includes public key is received)
associating the first machine learning model with the public key; (Par.(0111); validating images), (Par. (0118); machine learning models with public keys)
retrieving the first machine learning model using the public key; and (Par.(0118); machine learning model is verified against public key), (Par.(0147-0149); received encrypted information of machine learning model using public key)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati to incorporate the teaching of Baldwin to utilize the above feature because of the analogous concept of validation of images using machine learning, with the motivation of linking public key encryption with trained data models of artificial intelligence to further detect possible deep fake, fraud or modified images and to help decipher authentic depictions against possible A.I generated or false data. This helps maintain the integrity of the image detection system as a whole and adds a secure layer of protection with public keys used to retrieve models and exchange data (Baldwin Par. (0026)).
Hojjati and Baldwin do not explicitly teach training a first machine learning model using the first plurality of images to identify first features, the first features resulting from physical defects corresponding to one or more optical elements of the first camera or pixels of an image sensor of the first camera; using the first machine learning model to determine a first probability that a first image was captured using the first camera;
Wherein Stoppe teaches training a first machine learning model using the first plurality of images to identify first features, (Par. (0008-0010); training artificial network and machine learning to identify optical defects on at least two images that are tested), (Par. (0027-0030 and (0124-0129); machine learning corresponding to identifying defects in images ))
the first features resulting from physical defects corresponding to one or more optical elements of the first camera or pixels of an image sensor of the first camera; (Par. (0008-00010); testing and checking defects of images using machine learning and receive input based on optical standards), (Par. (0027-0030 and (0124-0129); machine learning corresponding to identifying defects in images ))
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati and Brown to incorporate the teaching of Stoppe to utilize the above feature because of the analogous concept of validation of images using machine learning, with the motivation of being able to detect visible defects to identify authenticity and errors in images. (Stoppe Par. (0002-0005))
In regards to Claim 2, the combination of Hojjati, Baldwin and Stoppe teach the method of claim one, Hojjati further teaches the computer-implemented method of claim 1, further comprising: determining that the first probability satisfies a first condition; and (Par. (0111); determine a particular likelihood that image has been alerted or modified [..] threshold value of 75% [..] value is verified and satisfies the probability threshold)
based at least on determining that the first probability satisfies the first condition, (Par. (0111); threshold value of 75% is verified to determine likelihood of image being altered and value is checked to see if value satisfies probability threshold)
determining first certification data indicating that the first plurality of images originated from the first camera. (Par. (0111); image is verified and satisfies threshold value of particular likelihood that image is not altered), (Par. (0013 and 0037); determine trustworthiness of image), (Par. (0059-0060); verification of image corresponding to image being directly from camera)
In regards to Claim 3, the combination of Hojjati, Baldwin, and Stoppe teach the method of claim one, Hojjati further teaches the computer-implemented method of claim 2, further comprising: storing the first certification data in a distributed ledger. (Par. (0071); image validation corresponding to distributed ledger technology with examples of blockchain)
Claim 4, is/are rejected under 35 U.S.C. 103 as being unpatentable over Hojjati et al. (U.S Pub. No. 20230344650, hereinafter referred to as “Hojjati”), Baldwin et al. (U.S Pub. No. 20230409756, hereinafter referred to as “Baldwin”) and Stoppe et al. (U.S Pub. No. 20210279858, hereinafter referred to as “Stoppe”)
further in view of Coenders et al. (U.S Pub. No. 20210150411, hereinafter referred to as “Coenders”)
In regards to Claim 4, the combination of Hojjati, Baldwin and Stoppe do not explicitly teach prior to receiving the first request: storing, in a decentralized storage system, first model data corresponding to the first machine learning model, determining first model hash data corresponding to the first model hash data, and storing the first model hash data in a distributed ledger.
Wherein Coenders teaches prior to receiving the first request: (Par. (0104-0105); storing of A.I model before request is made)
storing, in a decentralized storage system, first model data corresponding to the first machine learning model, (Par. (0089); blockchain storage with records of A.I models and related hash values)
determining first model hash data corresponding to the first model hash data, and storing the first model hash data in a distributed ledger. (Par. (0106); hash of A.I data model is stored), (Par. (0091); determining the first model hash data (results of hash that is determined), (Par. (0107); determining the first model hash data (validating the hash values of the local models)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati, Baldwin and Stoppe to incorporate the teaching of Coenders to utilize the above feature because of the analogous concept of blockchain technologies using machine learning, with the motivation of storing data models in a blockchain ledger to prevent tampering, fraudulent behavior or possible malware due to the immutable storage created. This allows hash values of the data models to be compared and retrieved when verifying the integrity of each model and provides a further enhance level of secure protection (Coenders Par. (0006)).
Claims 5, is/are rejected under 35 U.S.C. 103 as being unpatentable over Hojjati et al. (U.S Pub. No. 20230344650, hereinafter referred to as “Hojjati”), Baldwin et al. (U.S Pub. No. 20230409756, hereinafter referred to as “Baldwin”) Stoppe et al. (U.S Pub. No. 20210279858, hereinafter referred to as “Stoppe”)and Coenders et al. (U.S Pub. No. 20210150411, hereinafter referred to as “Coenders”) further in view of Kong et al. (U.S Pub. No. 20220067570, hereinafter referred to as “Kong”)
In regards to Claim 5, the combination of Hojjati, Baldwin and Stoppe do not explicitly teach retrieving, using the public key, the first model hash data from the distributed ledger; and retrieving, using the first model hash data, the first model data from the decentralized storage system.
Wherein Coenders teaches retrieving, using the first model hash data, the first model data from the decentralized storage system. (Par. (0008); computing hash of AI model), (Par. (0089); A.I model stored in blockchain registry), (Par. (0107); retrieving the first model hash data (when user wishes to perform [..] access model from blockchain registry and request made to go through local model hash to identify)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati, Baldwin and Stoppe to incorporate the teaching of Coenders to utilize the above feature because of the analogous concept of blockchain technologies using machine learning, with the motivation of storing data models in a blockchain ledger to prevent tampering, fraudulent behavior or possible malware due to the immutable storage created. This allows hash values of the data models to be compared and retrieved when verifying the integrity of each model and provides a further enhance level of secure protection (Coenders Par. (0006)).
Hojjati, Baldwin, Stoppe and Coenders do not explicitly teach retrieving, using the first model hash data, the first model data from the decentralized storage system.
Wherein Kong teaches retrieving, using the public key, the first model hash data from the distributed ledger; and (Par. (0034-0035); hashes of training data and machine learning model corresponding to retrieving a public key associated with blockchain), (Par. (0045); public key of machine learning model on blockchain), (Par. (0048); first model hash data (hashes of trained data of machine learning model)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati, Baldwin, Stoppe and Coenders to incorporate the teaching of Kong to utilize the above feature because of the analogous concept of hash based verification of data models using blockchain technologies and machine learning, with the motivation of utilizing a public key to redrive hash data models in the distributed ledger to effective locate and verify valid data models from invalid ones based on the corresponding key. This helps safeguard the data and promote high efficiency in the logs (Kong Par. (0002) and (0058)).
Claims 6-7, is/are rejected under 35 U.S.C. 103 as being unpatentable over Hojjati et al. (U.S Pub. No. 20230344650, hereinafter referred to as “Hojjati”), Baldwin et al. (U.S Pub. No. 20230409756, hereinafter referred to as “Baldwin”) Stoppe et al. (U.S Pub. No. 20210279858, hereinafter referred to as “Stoppe”) Patel et al. (U.S Pub. No. 20220277065, hereinafter referred to as “Patel”) Jakobsson et al. (U.S Pub. No. 20230011621, hereinafter referred to as “Jakobsson”) further in view of Griffin et al. (U.S Pub. No. 20200351075, hereinafter referred to as “Griffin”)
In regards to Claim 6, the combination of Hojjati, Baldwin and Stoppe do not explicitly teach receiving a second request for verification that a third image originated from the first camera, the third image corresponding to the public key; determining first feature data representing first image features extracted from a second plurality of images corresponding to the public key, the second plurality of images received prior to receiving the second request; determining second feature data representing second image features extracted from the third image data; and determining, using the first feature data and the second feature data, a second probability that the third image corresponds to an adversarial attack.
Wherein Patel teaches determining first feature data representing first image features extracted from a second plurality of images ….. (Par. (0092); comparing fourth image with image data)
the second plurality of images received prior to receiving the second request; (Par. (0101); requests for two images)
determining second feature data representing second image features extracted from the third image; and (Par. (0062); portion of second image data is matched with thirds image data)
determining, using the first feature data and the second feature data, a second probability that the third image corresponds to an adversarial attack. (Par. (0082-0083); first and second feature data of third image (eyes, mouth of third image) are matched for criteria) , (Par.(0086); . (Par. (0082-0083); first and second feature data of third image (features of third image), (Par. (0090); corresponds to an adversarial attack (fails to meet criteria later it is flagged and does not meet authorization)), (Par. (0063); a second probability that the third image data ( image data associated with third image fails to meet matching criteria)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati, Baldwin and Stoppe to incorporate the teaching of Patel to utilize the above feature because of the analogous concept of detecting and authenticating images, with the motivation of extracting features from multiple sets of images to determine the provability of an attack. This helps the user distinguish authentic images from possible spoof or fraudulent ones. This proves vital when browsing the web, conducting purchases and viewing various data to help mitigate possible harm based on tampered images and in return ensure high quality detection to users, (Patel (Par. (0005)).
Hojjati, Baldwin, Holland and Patel do not explicitly teach receiving a second request for verification that a third image data originated from the first camera, the third image corresponding to the public key; …..a second plurality of images corresponding to the public key,
Wherein Jakobsson teaches receiving a second request for verification that a third image originated from the first camera, (Figure 6 labels 670, 630; multiple request by miners 1, 2 and 3 with proof), (Par. (0145); request made by users to authenticate NFT), (Par. (0163); devices 505 making request to NFT), (Par. (0366); second request (multiple users request NFT content), (Par. (0394); content creators of NFT receiving requests)(Par. (0242); that first image data originated from a first camera (origination of NFT images), (Par. (0301); artifact origination of NFT based on digital camera to determine likelihood of origin)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati, Baldwin, Stoppe and Patel to incorporate the teaching of Jakobsson to utilize the above feature because of the analogous concept of detecting fraudulent images using blockchain technologies and machine learning, with the motivation of implementing a public key linked with image data and determining a probability based on machine learning the origins of an image. This proves important for users conducting purchases or browsing the web to prevent harm from images that are falsely created and in return allows the user to be aware of fraud, (Jakobsson Par. (0121-0125)).
Hojjati, Baldwin, Holland, Patel and Jakobsson do not explicitly teach, the third image corresponding to the public key; …. a second plurality of images corresponding to the public key,
Wherein Griffin teaches the third image corresponding to the public key; (Par. (0073); each block encoded as images (third image data or blocks 401-405 that are images)), (Par. (0099); each block with image and public key)
…a second plurality of images corresponding to the public key, (Par. (0078-0079); fourth image layer of the block), (Par. (0099); each block with image and public key)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati, Baldwin, Stoppe, Patel and Jakobsson to incorporate the teaching of Griffin to utilize the above feature because of the analogous concept of detecting and authenticating images, with the motivation of using encryption keys such as public keys as a form of comparison and to add another layer of effective determination. By using the public key linked with multiple images the correct users and authentic images can be matched an in return create valid results to distinguish authentic images from possible harm, (Griffin Par. (0006-0007)).
In regards to Claim 7, the combination of Hojjati, Baldwin and Stoppe do not explicitly teach determining that the second probability fails to satisfy a second condition; and in response to determining that the second probability fails to satisfy the second condition, outputting an indication of a possible adversarial attack.
Wherein Patel teaches determining that the second probability fails to satisfy a second condition; and (Par. (0082-0083); first and second feature data of third image (eyes, mouth of third image) are matched for criteria) , (Par.(0086); . (Par. (0082-0083); first and second feature data of third image (features of third image), (Par. (0090); corresponds to an adversarial attack (fails to meet criteria later it is flagged and does not meet authorization)), (Par. (0063); a second probability that the third image data ( image data associated with third image fails to meet matching criteria)
in response to determining that the second probability fails to satisfy the second condition, outputting an indication of a possible adversarial attack. (Par. (0090); fails to meet criteria image is flagged to user and does not meet authorization))
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati, Baldwin, Stoppe, Jakobsson and Griffin to incorporate the teaching of Patel for the reasons discussed in dependent claim 6 stated above.
Claims 8, 12, 14 and 17, is/are rejected under 35 U.S.C. 103 as being unpatentable over Hojjati et al. (U.S Pub. No. 20230344650, hereinafter referred to as “Hojjati”), Jakobsson et al. (U.S Pub. No. 20230011621, hereinafter referred to as “Jakobsson”) further in view of Stoppe et al. (U.S Pub. No. 20210279858, hereinafter referred to as “Stoppe”)
In regards to Claim 8, Hojjati teaches a computer-implemented method comprising: the first image corresponding to a public key; (Par. (0064); obtained public key with identity of user)
using the public key, (Par. (0064); obtained public key with identity of user)
determining that the first probability satisfies a first condition; (Par. (0111); threshold value of 75% is verified to determine likelihood of image being altered and value is checked to see if value satisfies probability threshold)
in response to determining that the first probability satisfies the first condition, determining first certification data indicating that the first image originated from the first camera; and (Par. (0111); image is verified and satisfies threshold value of particular likelihood that image is not altered), (Par. (0013 and 0037); determine trustworthiness of image), (Par. (0059-0060); verification of image corresponding to image being directly from camera)
sending, to the first user device, a first indication that the first image has been certified. (Figure 3B label 320, 322; user notification with message “validated” on image 320 displayed to user)
Hojjati does not explicitly teach receiving, from a first user device, a first request for certification that a first image originated from a first camera, retrieving, …., a first machine learning model trained using a plurality of images to identify first features resulting from physical defects corresponding to one or more optical elements of the first camera or pixels of an image sensor of the first camera; processing the first image using the first machine learning model to determine a first probability that the first image originated from the first camera;
Wherein Jakobsson teaches receiving, from a first user device, a first request for certification that first image data originated from a first camera, (Par. (0135); that first image data originated from a first camera (verify authenticity of NFT of content creator), (Par. (0129); that first image data originated from a first camera (NFT can include images), (Par. (0242); that first image data originated from a first camera (origination of NFT images), (Par. (0301); artifact origination of NFT based on digital camera to determine likelihood of origin), (Par. (0145); receiving, from a first user device, a first request for certification (user can request authentication of NFT image), (Par. (0265); requesting origin proof of NFT), (Par. (0394); receiving request associated with content creators of NFT),
processing the first image using the first machine learning model to determine a first probability that the first image originated from the first camera; (Par. (0291-0294); artifact corresponding to images files with precision score to indicate a likelihood of content is forgery or not), (Par. (0300-0302); likelihood of origination of artifact and image files, score of 1000 of out 1000 or 10 out of 1000 indicates likelihood of origination of artifact)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati to incorporate the teaching of Jakobsson to utilize the above feature because of the analogous concept of detecting fraudulent images using blockchain technologies and machine learning, with the motivation of implementing a public key linked with image data and determining a probability based on machine learning the origins of an image. This proves important for users conducting purchases or browsing the web to prevent harm from images that are falsely created and in return allows the user to be aware of fraud, (Jakobsson Par. (0121-0125)).
Hojjati and Jakobsson do not explicitly teach retrieving, …., a first machine learning model trained using a plurality of images to identify first features resulting from physical defects corresponding to one or more optical elements of the first camera or pixels of an image sensor of the first camera;
Wherein Stoppe teaches retrieving, …., a first machine learning model trained using a plurality of images to identify first features resulting from physical defects corresponding to one or more optical elements of the first camera or pixels of an image sensor of the first camera (Par. (0008-0010); training artificial network and machine learning to identify optical defects on at least two images that are tested), (Par. (0027-0030 and (0124-0129); machine learning corresponding to identifying defects in images )), (Par. (0008-00010); testing and checking defects of images using machine learning and receive input associated with ML model based on optical standards), (Par. (0027-0030 and (0124-0129); machine learning corresponding to identifying defects in images ))
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati and Brown to incorporate the teaching of Stoppe to utilize the above feature because of the analogous concept of validation of images using machine learning, with the motivation of being able to detect visible defects to identify authenticity and errors in images. (Stoppe Par. (0002-0005))
In regards to Claim 12, the combination of Hojjati, Jakobsson and Stoppe teach the method of claim 8, Hojjati further teaches the computer-implemented method of claim 8, further comprising: determining first image hash data corresponding to the first image, wherein first certification data includes the first image hash data; (Par. (0017); verification of image with hash), (Par. (0021); certificate with image and hash)
determining second image hash data corresponding to the second image; (Par. (0079); second image data (multiple images being validated), (Par. (0069-0070); hash of portion of image data and validating the image data)
retrieving, using the second image hash data, the first certification data from a distributed ledger, (Par. (0090); image and image hash stored on blockchain and obtained from blockchain)
the second image hash data corresponding to the first image hash data; and (Par. (0090); matching hash from image to hash and image stored in data store, image is validated)
sending the first certification data to the second user device in response to the second request. (Figure 3B label 320, 322; sending notification of “validated” as response)
Hojjati does not explicitly teach receiving, from a second user device, a second request to verify second image data corresponding to the public key;
Wherein Jakobssen further teaches receiving, from a second user device, a second request to verify second image corresponding to the public key; (Figure 6 labels 670, 630; multiple request by miners 1, 2 and 3 with proof), (Par. (0145); request made by users to authenticate NFT), (Par. (0163); devices 505 making request to NFT), (Par. (0366); second request (multiple users request NFT content), (Par. (0394); content creators of NFT receiving requests)(Par. (0242); that first image data originated from a first camera (origination of NFT images), (Par. (0301); artifact origination of NFT based on digital camera to determine likelihood of origin), (Par. (0145); request for NFT based on public key), (Par. (0229); NFT verified using public key)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati and Stoppe to incorporate the teaching of Jakobsson for the reasons discussed in independent claim 8 stated above.
In regards to Claim 14, Hojjati teaches a computer-implemented method comprising: the first image corresponding to a public key; (Par. (0064); obtained public key with identity of user)
using the public key ((Par. (0064); obtained public key with identity of user)
determining that the first probability satisfies a first condition; (Par. (0111); threshold value of 75% is verified to determine likelihood of image being altered and value is checked to see if value satisfies probability threshold)
in response to determining that the first probability satisfies the first condition, determining first certification data indicating that the first image originated from the first camera; and (Par. (0111); image is verified and satisfies threshold value of particular likelihood that image is not altered), (Par. (0013 and 0037); determine trustworthiness of image), (Par. (0059-0060); verification of image corresponding to image being directly from camera)
sending the first certification data to the first user device. (Figure 3B label 320, 322; user notification with message “validated” on image 320 displayed to user)
Hojjati does not explicitly teach receiving, from a first user device, a first request for verification that first image originated from a first camera, retrieving, …., a first machine learning model trained using a plurality of images to identify first features, the first features resulting from physical defects corresponding to one or more optical elements of the first camera or pixels of an image sensor of the first camera; processing the first image using the first machine learning model to determine a first probability that the first image originated from the first camera;
Wherein Jakobsson teaches receiving, from a first user device, a first request for verification that first image originated from a first camera, (Par. (0135); that first image data originated from a first camera (verify authenticity of NFT of content creator), (Par. (0129); that first image data originated from a first camera (NFT can include images), (Par. (0242); that first image data originated from a first camera (origination of NFT images), (Par. (0301); artifact origination of NFT based on digital camera to determine likelihood of origin), (Par. (0145); receiving, from a first user device, a first request for certification (user can request authentication of NFT image), (Par. (0265); requesting origin proof of NFT), (Par. (0394); receiving request associated with content creators of NFT),
processing the first image using the first machine learning model to determine a first probability that the first image data originated from the first camera; (Par. (0291-0294); artifact corresponding to images files with precision score to indicate a likelihood of content is forgery or not), (Par. (0300-0302); likelihood of origination of artifact and image files, score of 1000 of out 1000 or 10 out of 1000 indicates likelihood of origination of artifact)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati to incorporate the teaching of Jakobsson to utilize the above feature because of the analogous concept of detecting fraudulent images using blockchain technologies and machine learning, with the motivation of implementing a public key linked with image data and determining a probability based on machine learning the origins of an image. This proves important for users conducting purchases or browsing the web to prevent harm from images that are falsely created and in return allows the user to be aware of fraud, (Jakobsson Par. (0121-0125)).
Hojjati and Jakobsson do not explicitly retrieving, …, a first machine learning model trained using a plurality of images to identify first features, the first features resulting from physical defects corresponding to one or more optical elements of the first camera or pixels of an image sensor of the first camera;
Wherein Stoppe teaches retrieving, …, a first machine learning model trained using a plurality of images to identify first features, the first features resulting from physical defects corresponding to one or more optical elements of the first camera or pixels of an image sensor of the first camera; (Par. (0008-0010); training artificial network and machine learning to identify optical defects on at least two images that are tested), (Par. (0027-0030 and (0124-0129); machine learning corresponding to identifying defects in images )), (Par. (0008-00010); testing and checking defects of images using machine learning and receive input associated with ML model based on optical standards), (Par. (0027-0030 and (0124-0129); machine learning corresponding to identifying defects in images ))
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati and Brown to incorporate the teaching of Stoppe to utilize the above feature because of the analogous concept of validation of images using machine learning, with the motivation of being able to detect visible defects to identify authenticity and errors in images. (Stoppe Par. (0002-0005))
In regards to Claim 17, the combination of Hojjati, Jakobsson and Stoppe teach the method of claim 14, Hojjati further teaches the computer-implemented method of claim 14, further comprising: verifying, using the public key, that the second image was digitally signed using a private key corresponding to the public key; (Par. (0027); signing token In image), (Par. (0044); verify image for signed token of the image), (Par. (0066); image validation corresponding to public key), (Par. (0073); signed token in image data is verified), (Par. (0086); determines if validated image has signed token image data), (Par. (0024); signing the image with private key), (Par. (0079); second image data (multiple images being validated), (Par. (0069-0070); hash of portion of image data and validating the image data)
determining first image hash data corresponding to the second image; (Par. (0090); matching hash from image to hash and image stored in data store, image is validated)
determining that the first image hash data corresponds to second certification data stored in a distributed ledger, (Par. (0090); image and image hash stored on blockchain and obtained from blockchain),
in response to determining that the first image hash data corresponds to the second certification data, (Par. (0090); matching hash from image to hash and image stored in data store, image is validated)
sending the second certification data in response to the second request. (Figure 3B label 320, 322; sending notification of “validated” as response)
Hojjati does not explicitly teach receiving a second request for verification that a second image originated from the first camera, the second image corresponding to the public key; the second certification data corresponding to second image previously certified as originating from the first camera; and
Wherein Jakobsson teaches receiving a second request for verification that a second image data originated from the first camera, (Figure 6 labels 670, 630; multiple request by miners 1, 2 and 3 with proof), (Par. (0145); request made by users to authenticate NFT), (Par. (0163); devices 505 making request to NFT), (Par. (0366); second request (multiple users request NFT content), (Par. (0394); content creators of NFT receiving requests)(Par. (0242); that first image data originated from a first camera (origination of NFT images), (Par. (0301); artifact origination of NFT based on digital camera to determine likelihood of origin)
the second image corresponding to the public key; (Par. (0145); NFT based on public key), (Par. (0129); that first image data originated from a first camera (NFT can include images),
the second certification data corresponding to second image previously certified as originating from the first camera; and ((Par. (0135); that first image data originated from a first camera (verify authenticity of NFT’s of a plurality of NFTs of content creator), (Par. (0129); that first image data originated from a first camera (NFT can include images), (Par. (0242); that first image data originated from a first camera (origination of NFT’s images), (Par. (0301); artifact origination of NFT based on digital camera to determine likelihood of origin), (Par. (0129-0135); second certification data indicating that the second image data (each NFT with image is verified for authenticity)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati and Stoppe to incorporate the teaching of Jakobsson for the reasons discussed in independent claim 14 stated above.
Claims 9 and 15, is/are rejected under 35 U.S.C. 103 as being unpatentable over Hojjati et al. (U.S Pub. No. 20230344650, hereinafter referred to as “Hojjati”), Jakobsson et al. (U.S Pub. No. 20230011621, hereinafter referred to as “Jakobsson”) and Stoppe et al. (U.S Pub. No. 20210279858, hereinafter referred to as “Stoppe”) further in view of Kong et al. (U.S Pub. No. 20220067570, hereinafter referred to as “Kong”)
In regards to Claim 9, the combination of Hojjati, Jakobsson and Stoppe do not explicitly teach prior to processing the first image using the first machine learning model: retrieving, using the public key, first model hash data from a distributed ledger, and retrieving, using the first model hash data, first model data corresponding to the first machine learning model from a decentralized storage system.
Wherein Kong teaches prior to processing the first image using the first machine learning model: (Par. (0052); machine learning model determines feature or object in an image after steps of Par. (0035-0049) of retrieving hashes)
retrieving, using the public key, first model hash data from a distributed ledger, and (Par. (0034-0035); receive request for hashes, hashes provided by storage system via blockchain; hashes provided from storage with public key of the blockchain; encrypted with public key))(Par. (0048); first model hash data ( set of hashes corresponding to machine learning model)
retrieving, using the first model hash data, first model data corresponding to the first machine learning model from a decentralized storage system. (Par. (0048); comparing based on first and second set of hashes corresponding to machine learning model and training data; obtaining training data of machine learning model after verifying correct set of training data based on hashes), (Par. (0049) and (0056); hashes used to verify training data stored in blockchain), (Par. (0064); distributed ledger storing training data that was obtained)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati, Jakobsson and Stoppe to incorporate the teaching of Kong to utilize the above feature because of the analogous concept of detecting images using blockchain technologies and machine learning, with the motivation of safeguarding digital media such as images, videos and text by using training data and models to protect confidentiality and prevent copying of the data. By using hashing techniques coupled with a blockchain ledger tampering, fraudulent behavior or possible malware is prevented due to the immutable storage created. This allows hash values of the data models to be compared and retrieved when verifying the integrity of each model and provides a further enhance level of secure protection (Kong Par. (0012-0014 and 0033-0034)).
In regards to Claim 15, the combination of Hojjati, Jakobsson and Stoppe do not explicitly teach prior to processing the first image using the first machine learning model: retrieving, using the public key, first model hash data from a distributed ledger, and retrieving, using the first model hash data, first model data corresponding to the first machine learning model from a decentralized storage system.
Wherein Kong teaches prior to processing the first image using the first machine learning model: (Par. (0052); machine learning model determines feature or object in an image after steps of Par. (0035-0049) of retrieving hashes)
retrieving, using the public key, first model hash data from a distributed ledger, and (Par. (0034-0035); receive request for hashes, hashes provided by storage system via blockchain; hashes provided from storage with public key of the blockchain; encrypted with public key))(Par. (0048); first model hash data ( set of hashes corresponding to machine learning model)
retrieving, using the first model hash data, first model data corresponding to the first machine learning model from a decentralized storage system. (Par. (0048); comparing based on first and second set of hashes corresponding to machine learning model and training data; obtaining training data of machine learning model after verifying correct set of training data based on hashes), (Par. (0049) and (0056); hashes used to verify training data stored in blockchain), (Par. (0064); distributed ledger storing training data that was obtained)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati, Jakobsson and Stoppe to incorporate the teaching of Kong to utilize the above feature because of the analogous concept of detecting images using blockchain technologies and machine learning, with the motivation of safeguarding digital media such as images, videos and text by using training data and models to protect confidentiality and prevent copying of the data. By using hashing techniques coupled with a blockchain ledger tampering, fraudulent behavior or possible malware is prevented due to the immutable storage created. This allows hash values of the data models to be compared and retrieved when verifying the integrity of each model and provides a further enhance level of secure protection (Kong Par. (0012-0014 and 0033-0034)).
Claims 10-11 and 16, is/are rejected under 35 U.S.C. 103 as being unpatentable over Hojjati et al. (U.S Pub. No. 20230344650, hereinafter referred to as “Hojjati”), Jakobsson et al. (U.S Pub. No. 20230011621, hereinafter referred to as “Jakobsson”) and Stoppe et al. (U.S Pub. No. 20210279858, hereinafter referred to as “Stoppe”) further in view of Stonehouse et al. (U.S Pub. No. 20210192340, hereinafter referred to as “Stonehouse”)
In regards to Claim 10, the combination of Hojjati, Jakobsson and Stoppe teach the method of claim 8, Hojjati further teaches using the retrieving, using the public key, the first machine learning model; (Par. (0064 and 0111-0113); public key corresponding to machine learning model input data obtained)
Hojjati does not explicitly teach receiving, from a second user device, a second request for verification that a second image originated from the first camera, the second image corresponding to the public key; processing the second image using the first machine learning model to determine a second probability that the second image data originated from the first camera; determining that the second probability satisfies a second condition;
Jakobssen further teaches receiving, from a second user device, a second request for verification that a second image originated from the first camera, (Figure 6 labels 670, 630; multiple request by miners 1, 2 and 3 with proof), (Par. (0145); request made by users to authenticate NFT), (Par. (0163); devices 505 making request to NFT), (Par. (0366); second request (multiple users request NFT content), (Par. (0394); content creators of NFT receiving requests)(Par. (0242); that first image data originated from a first camera (origination of NFT images), (Par. (0301); artifact origination of NFT based on digital camera to determine likelihood of origin)
the second image corresponding to the public key; (Par. (0229); NFT with public key), (Par. (0129); second image data (NFTs with digital asset images)
processing the second image using the first machine learning model to determine a second probability that the second image data originated from the first camera; (Par. (0301); determine likelihood using machine learning methods of camera image and likelihood of origin), (Par. (0129); second image data (NFTs with digital asset images), (Par. (0302); origination assessment of artifact with an assessment score corresponding to NFT and insurance associated with originator of NFT and current owner), (Par. (0291-0294); artifact corresponding to images files with precision score to indicate a likelihood of content is forgery or not), (Par. (0300-0302); likelihood of origination of artifact and image files, score of 1000 of out 1000 or 10 out of 1000 indicates likelihood of origination of artifact)
determining that the second probability satisfies a second condition; (Par. (0129); second image data (NFTs with digital asset images), (Par. (0302); second probability (origination assessment of artifact with an assessment score corresponding to NFT and insurance associated with originator of NFT and current owner), (Par. (0291); original assessment score corresponding to satisfying or exceeding a threshold)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati to incorporate the teaching of Jakobsson to utilize the above feature because of the analogous concept of detecting fraudulent images using blockchain technologies and machine learning, with the motivation of implementing a public key linked with image data and determining a probability based on machine learning the origins of an image. This proves important for users conducting purchases or browsing the web to prevent harm from images that are falsely created and in return allows the user to be aware of fraud, (Jakobsson Par. (0121-0125)).
Hojjati, Jakobsson and Stoppe do not explicitly teach in response to determining that the second probability satisfies the second condition, determining second certification data indicating that the second image originated from the first camera; and sending, to the second user device, a second indication that the second image has been verified as originating from the first camera.
Wherein Stonehouse teaches in response to determining that the second probability satisfies the second condition, (Par. (0016-0017); determine likelihood that an image is authentic corresponding to different camera devices and machine learning models), (Par. (0048-0049); second probability (likelihood image is authentic based on 82%), (Par. (0118); first probability (length greater than 0.01mm to determine likelihood image is authentic)
determining second certification data indicating that the second image originated from the first camera; and (Par. (0048-0049); determining image of subject consumer good is authentic by displaying on window of interface “Authentic”), (Par. (0091-0092); determining authentic images on cameras of iPhone, Samsung, Huawei etc.)
sending, to the second user device, a second indication that the second image has been verified as originating from the first camera. (Figure 1C; label 76; indication of “Authentic displayed on user device interface), (Par. (0081); sending user device (one or more mobile devices), (Par. (0086); outputting on device classifying image as authentic)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati, Jakobsson and Stoppe to incorporate the teaching of Stonehouse to utilize the above feature because of the analogous concept of detecting and authenticating images, with the motivation of extracting features from multiple sets of images to determine the provability of an attack. This helps the user distinguish authentic images from possible counterfeit, forged or inauthentic images. This proves vital when browsing the web, conducting purchases and viewing various data to help mitigate possible harm based on tampered images and in return ensure high quality detection to users, (Stonehouse (Par. (0002-0003)).
In regards to Claim 11, the combination of Hojjati, Jakobsson and Stoppe teach the method of claim 8, Hojjati further teaches using the retrieving, using the public key, the first machine learning model; (Par. (0064 and 0111-0113); public key corresponding to machine learning model input data obtained)
Hojjati does not explicitly teach receiving, from a second user device, a second request for verification that a second image originated from the first camera, the second image [[data]] corresponding to the public key; processing the second image using the first machine learning model to determine a second probability that the second image originated from the first camera; determining that the second probability fails to satisfy a second condition; and in response to determining that the second probability fails to satisfy the second condition, sending, to the second user device, a second indication that the second image could not be verified as originating from the first camera.
Jakobssen further teaches receiving, from a second user device, a second request for verification that a second image originated from the first camera, (Figure 6 labels 670, 630; multiple request by miners 1, 2 and 3 with proof), (Par. (0145); request made by users to authenticate NFT), (Par. (0163); devices 505 making request to NFT), (Par. (0366); second request (multiple users request NFT content), (Par. (0394); content creators of NFT receiving requests)(Par. (0242); that first image data originated from a first camera (origination of NFT images), (Par. (0301); artifact origination of NFT based on digital camera to determine likelihood of origin)
processing the second image using the first machine learning model to determine a second probability that the second image originated from the first camera; (Par. (0301); determine likelihood using machine learning methos of camera image and likelihood of origin), (Par. (0129); second image data (NFTs with digital asset images), (Par. (0302); origination assessment of artifact with an assessment score corresponding to NFT and insurance associated with originator of NFT and current owner), (Par. (0291-0294); artifact corresponding to images files with precision score to indicate a likelihood of content is forgery or not), (Par. (0300-0302); likelihood of origination of artifact and image files, score of 1000 of out 1000 or 10 out of 1000 indicates likelihood of origination of artifact)
determining that the second probability fails to satisfy a second condition; and ((Par. (0129); second image data (NFTs with digital asset images), (Par. (0302); second probability (origination assessment of artifact with an assessment score corresponding to NFT and insurance associated with originator of NFT and current owner), (Par. (0291); original assessment score corresponding to satisfying or exceeding a threshold)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati to incorporate the teaching of Jakobsson to utilize the above feature because of the analogous concept of detecting fraudulent images using blockchain technologies and machine learning, with the motivation of implementing a public key linked with image data and determining a probability based on machine learning the origins of an image. This proves important for users conducting purchases or browsing the web to prevent harm from images that are falsely created and in return allows the user to be aware of fraud, (Jakobsson Par. (0121-0125)).
Hojjati and Jakobsson do not explicitly teach in response to determining that the second probability fails to satisfy the second condition, sending, to the second user device, a second indication that the second image data could not be verified as originating from the first camera.
Hojjati, Jakobsson and Stoppe do not explicitly teach in response to determining that the second probability fails to satisfy the second condition, sending, to the second user device, a second indication that the second image data could not be verified as originating from the first camera.
Wherein Stonehouse teaches in response to determining that the second probability fails to satisfy the second condition, (Par. (0016-0017); determine likelihood that an image is authentic corresponding to different camera devices and machine learning models), (Par. (0048-0049); second probability (likelihood image is authentic based on 82%), (Par. (0118); first probability (length greater than 0.01mm to determine likelihood image is authentic)
sending, to the second user device, a second indication that the second image could not be verified as originating from the first camera. (Par. (0044-0048); outputting classification that image is non-authentic), Par. (0081); sending user device (one or more mobile devices), (Par. (0086); outputting on device classifying image as authentic)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati, Jakobsson and Stoppe to incorporate the teaching of Stonehouse to utilize the above feature because of the analogous concept of detecting and authenticating images, with the motivation of extracting features from multiple sets of images to determine the provability of an attack. This helps the user distinguish authentic images from possible counterfeit, forged or inauthentic images. This proves vital when browsing the web, conducting purchases and viewing various data to help mitigate possible harm based on tampered images and in return ensure high quality detection to users, (Stonehouse (Par. (0002-0003)).
In regards to Claim 16, claim 16 recites similar limitations to dependent claim 11 and the teachings of Hojjati, Jakobsson, Horowitz and Stonehouse address all the limitations discussed in claim 11 and are thereby rejected under the same grounds.
Claim 13, is/are rejected under 35 U.S.C. 103 as being unpatentable over Hojjati et al. (U.S Pub. No. 20230344650, hereinafter referred to as “Hojjati”), Jakobsson et al. (U.S Pub. No. 20230011621, hereinafter referred to as “Jakobsson”) and Stoppe et al. (U.S Pub. No. 20210279858, hereinafter referred to as “Stoppe”) further in view of Franco et al. (U.S No. 11908167, hereinafter referred to as “Franco”)
In regards to Claim 13, the combination of Hojjati, Jakobsson and Horowitz do not explicitly teach computing a zero-knowledge proof that the first image was processed using the first machine learning model and that the first probability satisfies the first condition, wherein first certification includes the zero-knowledge proof.
Wherein Franco teaches computing a zero-knowledge proof that the first image was processed using the first machine learning model and that the first probability satisfies the first condition, wherein first certification includes the zero-knowledge proof. (Col. 4 lines 6-20; machine learning ), (Col. 21 lines 1-45; model of machine learning), (Col. 24 lines 23-40; zero-knowledge proof associated with digital image), (Col. 23 lines 60-67 and Col. 24 lines 1-15; images satisfy predetermined threshold)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati, Jakobsson and Stoppe to incorporate the teaching of Franco to utilize the above feature because of the analogous concept of authenticating image data using machine learning, with the motivation of determining a probability and condition that is satisfied using zero-knowledge proof to alert users possible compromise or threats. By using machine learning data models can be trained and effectively used to have a condition or threshold that must be met and create detection mechanism to notify before harm of image data and exchanges to users (Franco Col. 2 lines 10-25).
Claims 18-19, is/are rejected under 35 U.S.C. 103 as being unpatentable over Hojjati et al. (U.S Pub. No. 20230344650, hereinafter referred to as “Hojjati”), Jakobsson et al. (U.S Pub. No. 20230011621, hereinafter referred to as “Jakobsson”) and Stoppe et al. (U.S Pub. No. 20210279858, hereinafter referred to as “Stoppe”) Patel et al. (U.S Pub. No. 20220277065, hereinafter referred to as “Patel”) further in view of Griffin et al. (U.S Pub. No. 20200351075, hereinafter referred to as “Griffin”)
In regards to Claim 18, the combination of Hojjati, Jakobsson and Stoppe teach the method of claim 14, Jakobsson further teaches receiving a second request for verification that a second image originated from the first camera, (Figure 6 labels 670, 630; multiple request by miners 1, 2 and 3 with proof), (Par. (0145); request made by users to authenticate NFT), (Par. (0163); devices 505 making request to NFT), (Par. (0366); second request (multiple users request NFT content), (Par. (0394); content creators of NFT receiving requests)(Par. (0242); that first image data originated from a first camera (origination of NFT images), (Par. (0301); artifact origination of NFT based on digital camera to determine likelihood of origin)
the second image corresponding to the public key; ((Par. (0145); NFT based on public key), (Par. (0129); that first image data originated from a first camera (NFT can include images),
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati to incorporate the teaching of Jakobsson to utilize the above feature because of the analogous concept of detecting fraudulent images using blockchain technologies and machine learning, with the motivation of implementing a public key linked with image data and determining a probability based on machine learning the origins of an image. This proves important for users conducting purchases or browsing the web to prevent harm from images that are falsely created and in return allows the user to be aware of fraud, (Jakobsson Par. (0121-0125)).
Hojjati, Jakobsson and Stoppe do not explicitly teach determining first feature data representing first image features extracted from third image data, the third image data representing a plurality of images corresponding to the public key, the plurality of images received prior to receiving the second request; determining second feature data representing second image features extracted from the second image data; and determining, using the first feature data and the second feature data, a second probability that the second image data corresponds to an adversarial attack.
Wherein Patel teaches determining first feature data representing first image features extracted from a plurality of images …., (Par. (0092); comparing fourth image with image data), (Par. (0062); portion of second image data is matched with image data)
determining second feature data representing second image features extracted from the second image; and (Par. (0062); portion of second image data is matched with image data)
determining, using the first feature data and the second feature data, a second probability that the second image corresponds to an adversarial attack. (Par. (0082-0083); first and second feature data of third image (eyes, mouth of third image) are matched for criteria) , (Par.(0086); . (Par. (0082-0083); first and second feature data of third image (features of third image), (Par. (0090); corresponds to an adversarial attack (fails to meet criteria later it is flagged and does not meet authorization)), (Par. (0063); a second probability that the third image data ( image data associated with third image fails to meet matching criteria)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati, Jakobsson and Horowitz to incorporate the teaching of Patel to utilize the above feature because of the analogous concept of detecting and authenticating images, with the motivation of extracting features from multiple sets of images to determine the provability of an attack. This helps the user distinguish authentic images from possible spoof or fraudulent ones. This proves vital when browsing the web, conducting purchases and viewing various data to help mitigate possible harm based on tampered images and in return ensure high quality detection to users, (Patel (Par. (0005)).
Hojjati, Jakobsson, Horowitz and Patel do not explicitly teach a plurality of images corresponding to the public key, the third image data representing a plurality of images corresponding to the public key, the plurality of images received prior to receiving the second request;
Wherein Griffin teaches a plurality of images corresponding to the public key, (Par. (0099); each block with image and public key), (Par. (0043-0048); encoded data in blocks with multi-layered images before client request the original data)
the plurality of images received prior to receiving the second request; (Par. (0090); new block with encoded image being added followed by client requesting to propose new transaction)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati, Jakobsson, Stoppe and Patel to incorporate the teaching of Griffin to utilize the above feature because of the analogous concept of detecting and authenticating images, with the motivation of using encryption keys such as public keys as a form of comparison and to add another layer of effective determination. By using the public key linked with multiple images the correct users and authentic images can be matched an in return create valid results to distinguish authentic images from possible harm, (Griffin Par. (0006-0007)).
In regards to Claim 19, the combination of Hojjati, Jakobsson and Stoppe do not explicitly teach determining that the second probability fails to satisfy a second condition; and in response to determining that the second probability fails to satisfy the second condition, outputting an indication of a possible adversarial attack.
Wherein Patel teaches determining that the second probability fails to satisfy a second condition; and (Par. (0082-0083); first and second feature data of third image (eyes, mouth of third image) are matched for criteria) , (Par.(0086); . (Par. (0082-0083); first and second feature data of third image (features of third image), (Par. (0090); corresponds to an adversarial attack (fails to meet criteria later it is flagged and does not meet authorization)), (Par. (0063); a second probability that the third image data ( image data associated with third image fails to meet matching criteria)
in response to determining that the second probability fails to satisfy the second condition, outputting an indication of a possible adversarial attack. (Par. (0090); fails to meet criteria image is flagged to user and does not meet authorization))
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati, Jakobsson, Horowitz and Griffin to incorporate the teaching of Patel for the reasons discussed in dependent claim 18 stated above.
Claim 20, is/are rejected under 35 U.S.C. 103 as being unpatentable over Hojjati et al. (U.S Pub. No. 20230344650, hereinafter referred to as “Hojjati”), Stoppe et al. (U.S Pub. No. 20210279858, hereinafter referred to as “Stoppe”) and Coenders et al. (U.S Pub. No. 20210150411, hereinafter referred to as “Coenders”) further in view of Kong et al. (U.S Pub. No. 20220067570, hereinafter referred to as “Kong”)
In regards to Claim 20, Hojjati teaches a computer-implemented method comprising: receiving first image data representing a first plurality of images captured using a first camera; (Par. (0060-0061); receive image and corresponding image data directly from camera)
verifying, using the public key, that the first image data was digitally signed using a private key corresponding to the public key; (Par. (0027); signing token In image), (Par. (0044); verify image for signed token of the image), (Par. (0066); image validation corresponding to public key), (Par. (0073); signed token in image data is verified), (Par. (0086); determines if validated image has signed token image data), (Par. (0024); signing the image with private key)
the first features resulting from at least one physical characteristic of the first camera; (Fig 3A label 321, 320, physical characteristics of houses and land taken features by camera)
Hojjati does not explicitly teach receiving a public key corresponding to a first user device; training a first machine learning model using the first plurality of images to identify first features, resulting from physical defects corresponding to one or more optical elements of the first camera or pixels of an image sensor of the first camera; determining first model hash data corresponding to the first machine learning model; associating the first model hash data with the public key in a distributed ledger; associating the first model hash data with first model data representing the first machine learning model; and storing the first model data in a storage system.
Wherein Stoppe teaches training a first machine learning model using the first plurality of images to identify first features, resulting from physical defects corresponding to one or more optical elements of the first camera or pixels of an image sensor of the first camera; ((Par. (0008-0010); training artificial network and machine learning to identify optical defects on at least two images that are tested), (Par. (0027-0030 and (0124-0129); machine learning corresponding to identifying defects in images )), (Par. (0008-00010); testing and checking defects of images using machine learning and receive input based on optical standards), (Par. (0027-0030 and (0124-0129); machine learning corresponding to identifying defects in images ))
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati and Brown to incorporate the teaching of Stoppe to utilize the above feature because of the analogous concept of validation of images using machine learning, with the motivation of being able to detect visible defects to identify authenticity and errors in images. (Stoppe Par. (0002-0005))
Hojjati and Stoppe do not explicitly teach receiving a public key corresponding to a first user device; determining first model hash data corresponding to the first machine learning model; associating the first model hash data with the public key in a distributed ledger; associating the first model hash data with first model data representing the first machine learning model; and storing the first model data in a storage system.
Wherein Coenders teaches determining first model hash data corresponding to the first machine learning model; (Par. (0106); hash of A.I data model is stored), (Par. (0091); determining the first model hash data (results of hash that is determined), (Par. (0107); determining the first model hash data (validating the hash values of the local models)
associating the first model hash data with first model data representing the first machine learning model; and (Par. (0006); hash with A.I model that is stored)
storing the first model data in a storage system. (Par. (0089); blockchain storage with records of A.I models and related hash values)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati and Stoppe to incorporate the teaching of Coenders to utilize the above feature because of the analogous concept of blockchain technologies using machine learning, with the motivation of storing data models in a blockchain ledger to prevent tampering, fraudulent behavior or possible malware due to the immutable storage created. This allows hash values of the data models to be compared and retrieved when verifying the integrity of each model and provides a further enhance level of secure protection (Coenders Par. (0006)).
Hojjati, Stoppe and Coenders do not explicitly teach receiving a public key corresponding to a first user device; associating the first model hash data with the public key in a distributed ledger;
Wherein Kong teaches receiving a public key corresponding to a first user device; (Par. (0045); training package encrypted using public key of computing device), (Figure 4 label 415, 140 and 120; receiving public key (computing device transmits training package with public key)
associating the first model hash data with the public key in a distributed ledger; (Par. (0034-0035); hashes of training data and machine learning model corresponding to retrieving a public key associated with blockchain), (Par. (0045); public key of machine learning model on blockchain), (Par. (0048); first model hash data (hashes of trained data of machine learning model)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Hojjati, Stoppe and Coenders to incorporate the teaching of Kong to utilize the above feature because of the analogous concept of hash based verification of data models using blockchain technologies and machine learning, with the motivation of utilizing a public key to redrive hash data models in the distributed ledger to effective locate and verify valid data models from invalid ones based on the corresponding key. This helps safeguard the data and promote high efficiency in the logs (Kong Par. (0002) and (0058)).
Relevant Prior Art
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
UNNIKRISHNAN; Jayakrishnan (U.S Pub. No. 20220237402) “STATIC OCCUPANCY TRACKING”. Considered this reference because it addressed probability of images using machine learning.
WAGNER; Matthias. (U.S Pub. No. 20220284574) “PLATFORMS AND SYSTEMS FOR AUTOMATED CELL CULTURE”. Considered this application because it relates to determine the origin of a camera image and a probability of a condition being met.
QUELLEC; Gwenolé (U.S Pub. No. 20220237900) “AUTOMATIC IMAGE ANALYSIS METHOD FOR AUTOMATICALLY RECOGNISING AT LEAST ONE RARE CHARACTERISTIC”. Considered this application because it addressed detection and analysis of images based a probability to identify discrepancies.
Conclusion
THIS ACTION IS MADE FINAL. 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 HASSAN A HUSSEIN whose telephone number is (571)272-3554. The examiner can normally be reached on 7:30am-5pm.
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, Eleni Shiferaw can be reached on (571)272-3867. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see https://ppair-y.uspto.gov/pair/PrivatePair. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/H.A.H./Examiner, Art Unit 2497
/BRYAN F WRIGHT/Examiner, Art Unit 2497