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 . Claims 1-11 are presented in the case.
Priority
Acknowledgment is made of applicant's claim for foreign priority based on application JP2023-120001 filed in Japan on 07/24/2023. Receipt is acknowledged of certified copies of papers required by 37 CFR 1.55.
Claim Objections
Claims 1 and 4 are objected to because of the following informalities:
Claim 1, line 9 recites the phrase “function that inputs therein the object sample” which should be “function that receives the object sample as an input”
Claim 4, line 5 recites the phrase “each of classes” which should be “each class”
For the informalities above and wherever else they may occur appropriate correction is required.
Specification
The title of the invention is not descriptive. A new title is required that is clearly indicative of the invention to which the claims are directed.
The following title is suggested: “Sample-Specific Data-Augmentation Parameter Learning”.
Information Disclosure Statement
The information disclosure statements submitted on 02/29/2024 and 04/07/2026 are in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statements are being considered by the examiner.
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 2, 7, and 9 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 2 recites the phrase “decrease a prediction error from each of the object sample”. This appears before any prediction steps are performed and it is also unclear whether this can be known during training. In general, learning models attempt to reduce loss or error during training iterations which may not be achieved. One of ordinary skills in art would not be reasonably apprised of the scope of the invention.
The terms “augmentation parameter that is large”, “augmentation parameter that is small” and “which is not completely trained” in claims 7 and 9 are a relative term which renders the claim indefinite. The terms are not defined by the claim, the specification does not provide a standard for ascertaining the requisite degree, and one of ordinary skill in the art would not be reasonably apprised of the scope of the invention. These can be replaced with thresholds or value ranges or removed for a more concrete interpretation.
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 of this title, 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, 5-7 and 9-11 are rejected under 35 U.S.C. 103 as being unpatentable over MOUNSAVENG et al. (US 20210241041 A1) hereinafter MOUNSAVENG in view of Jaderberg et al. (US 20180330185 A1) hereinafter Jaderberg.
As to independent claim 1, MOUNSAVENG teaches a machine learning apparatus comprising processing circuitry, the processing circuitry being configured to: [training server with processor ¶89]
acquire a training sample including an object sample and a target value correlated with the object sample; [training images (object sample) received with labels (target value) ¶164-166 "At each training loop iteration 305, a given training image 312 is received from the subset of training images 285. The given training image 312 is annotated or labelled with a class label 314 indicative of a class of one or more objects depicted in the image."]
generate a first augmented sample by applying data augmentation to the object sample in accordance with a first data augmentation parameter; and [generates transformed training images (first augmented sample) by applying zoom etc. (data augmentation) in accordance to parameters ¶169-172 "receives as an input the given training image 312 and applies a transformation on the given training image 312 to obtain a respective augmented or transformed training image 322. The augmentation network 270 applies a transformation on the given training image 312 vector using the noise vector 318"…"The transformations are performed according to the current values of the set of transformation parameters 275."]
generate a parameter output function that inputs therein the object sample and outputs [[a second data augmentation parameter]] corresponding to the object sample], by machine learning based on the object sample, the target value and the first augmented sample. [generated network classifies (generates output) the transformed image and calculates loss based on label and transformed image to update the network for further transformations ¶182, ¶174-176 "classification network 250 extracts a set of image features from the transformed training image 322 to perform classification and output a class prediction 332"]
MOUNSAVENG does not specifically teach output function that inputs therein the object sample and outputs a second data augmentation parameter corresponding to the object sample.
However, Jaderberg teaches output function that inputs therein the object sample and outputs a second data augmentation parameter corresponding to the object sample. [outputs a second data augmentation (parameter of transformation) based on input image and ML function (localization neural network) ¶25, ¶32 "a neural network that is configured to process the input feature map 106 to generate the parameters of the transformation"]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the object augmentation by MOUNSAVENG by incorporating the output function that inputs therein the object sample and outputs a second data augmentation parameter corresponding to the object sample disclosed by Jaderberg because both techniques address the same field of machine learning and by incorporating Jaderberg into MOUNSAVENG improves neural network models with improved benchmark beating performance in its transformation [Jaderberg ¶6].
As to dependent claim 3, the rejection of claim 1 is incorporated, MOUNSAVENG and Jaderberg further teach wherein the processing circuitry is configured to determine, based on the second data augmentation parameter, a third data augmentation parameter for generating a second augmented sample that is used for training a target value prediction model that inputs therein the object sample and outputs a predicted value of a target value corresponding to the object sample. [MOUNSAVENG updates network according to transformation parameters iteratively and train classification (each iteration has a new data augmentation parameter) ¶169-172, ¶139-141 "applying one step of gradient descent on the classification network 250, instead of a plurality of iterations (e.g. hundreds) as with gradient descent"]
As to dependent claim 5, the rejection of claim 3 is incorporated, MOUNSAVENG and Jaderberg further teach wherein the processing circuitry is configured to:
generate the second augmented sample by applying data augmentation in accordance with the third data augmentation parameter; and [MOUNSAVENG generates transformed training images (first augmented sample) by applying zoom etc. (data augmentation) in accordance to parameters ¶169-172]
generate the target value prediction model that inputs therein the object sample and outputs the predicted value, by machine learning based on the target value and the second augmented sample. [MOUNSAVENG trains (generates model) accordingly ¶42]
As to dependent claim 6, the rejection of claim 5 is incorporated, MOUNSAVENG and Jaderberg further teach wherein the processing circuitry is configured to train the target value prediction model in such a manner as to decrease a loss based on the predicted value and the target value. [MOUNSAVENG loss is used to improved classification and reduce error via training ¶134, ¶28 ]
As to dependent claim 7, the rejection of claim 6 is incorporated, MOUNSAVENG and Jaderberg further teach wherein the processing circuitry is configured to train the target value prediction model in such a manner as to decrease the loss in a case where the object sample with the second data augmentation parameter that is large is weighted by a greater value than the object sample with the second data augmentation parameter that is small. [MOUNSAVENG modifies weights of nodes to optimize model ¶51, ¶134, ¶28]
As to dependent claim 9, the rejection of claim 1 is incorporated, MOUNSAVENG and Jaderberg further teach wherein the first data augmentation parameter is a data augmentation parameter acquired by applying a parameter output function, which is not completely trained, to the object sample. [Jaderberg applies localization subnetwork (output function) to image (sample) ¶32, ¶16 " image processing neural network for localizing objects in images that can be modified to include one or more spatial transformer modules"], [during training (not completely trained) ¶48 "during the training of the image processing neural network system, the system can backpropagate gradients computed from errors between the neural network outputs generated by the image processing neural network and known outputs for the training images to train the spatial transformer module, i.e., to adjust the values of the parameters of the localisation subnetwork and, optionally, of the sampling kernel"]
As to independent claim 10, MOUNSAVENG teaches a machine learning method comprising: [¶176]
acquire a training sample including an object sample and a target value correlated with the object sample; [training images (object sample) received with labels (target value) ¶164-166 "At each training loop iteration 305, a given training image 312 is received from the subset of training images 285. The given training image 312 is annotated or labelled with a class label 314 indicative of a class of one or more objects depicted in the image."]
generate a first augmented sample by applying data augmentation to the object sample in accordance with a first data augmentation parameter; and [generates transformed training images (first augmented sample) by applying zoom etc. (data augmentation) in accordance to parameters ¶169-172 "receives as an input the given training image 312 and applies a transformation on the given training image 312 to obtain a respective augmented or transformed training image 322. The augmentation network 270 applies a transformation on the given training image 312 vector using the noise vector 318"…"The transformations are performed according to the current values of the set of transformation parameters 275."]
generate a parameter output function that inputs therein the object sample and outputs [[a second data augmentation parameter]] corresponding to the object sample], by machine learning based on the object sample, the target value and the first augmented sample. [generated network classifies (generates output) the transformed image and calculates loss based on label and transformed image to update the network for further transformations ¶182, ¶174-176 "classification network 250 extracts a set of image features from the transformed training image 322 to perform classification and output a class prediction 332"]
MOUNSAVENG does not specifically teach output function that inputs therein the object sample and outputs a second data augmentation parameter corresponding to the object sample.
However, Jaderberg teaches output function that inputs therein the object sample and outputs a second data augmentation parameter corresponding to the object sample. [outputs a second data augmentation (parameter of transformation) based on input image and ML function (localization neural network) ¶25, ¶32 "a neural network that is configured to process the input feature map 106 to generate the parameters of the transformation"]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the object augmentation by MOUNSAVENG by incorporating the output function that inputs therein the object sample and outputs a second data augmentation parameter corresponding to the object sample disclosed by Jaderberg because both techniques address the same field of machine learning and by incorporating Jaderberg into MOUNSAVENG improves neural network models with improved benchmark beating performance in its transformation [Jaderberg ¶6].
As to independent claim 11, MOUNSAVENG teaches a non-transitory computer readable medium including computer executable instructions, wherein the instructions, when executed by a processor, cause the processor to perform operations comprising: [training server with processor ¶89, ¶212]
acquire a training sample including an object sample and a target value correlated with the object sample; [training images (object sample) received with labels (target value) ¶164-166 "At each training loop iteration 305, a given training image 312 is received from the subset of training images 285. The given training image 312 is annotated or labelled with a class label 314 indicative of a class of one or more objects depicted in the image."]
generate a first augmented sample by applying data augmentation to the object sample in accordance with a first data augmentation parameter; and [generates transformed training images (first augmented sample) by applying zoom etc. (data augmentation) in accordance to parameters ¶169-172 "receives as an input the given training image 312 and applies a transformation on the given training image 312 to obtain a respective augmented or transformed training image 322. The augmentation network 270 applies a transformation on the given training image 312 vector using the noise vector 318"…"The transformations are performed according to the current values of the set of transformation parameters 275."]
generate a parameter output function that inputs therein the object sample and outputs [[a second data augmentation parameter]] corresponding to the object sample], by machine learning based on the object sample, the target value and the first augmented sample. [generated network classifies (generates output) the transformed image and calculates loss based on label and transformed image to update the network for further transformations ¶182, ¶174-176 "classification network 250 extracts a set of image features from the transformed training image 322 to perform classification and output a class prediction 332"]
MOUNSAVENG does not specifically teach output function that inputs therein the object sample and outputs a second data augmentation parameter corresponding to the object sample.
However, Jaderberg teaches output function that inputs therein the object sample and outputs a second data augmentation parameter corresponding to the object sample. [outputs a second data augmentation (parameter of transformation) based on input image and ML function (localization neural network) ¶25, ¶32 "a neural network that is configured to process the input feature map 106 to generate the parameters of the transformation"]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the object augmentation by MOUNSAVENG by incorporating the output function that inputs therein the object sample and outputs a second data augmentation parameter corresponding to the object sample disclosed by Jaderberg because both techniques address the same field of machine learning and by incorporating Jaderberg into MOUNSAVENG improves neural network models with improved benchmark beating performance in its transformation [Jaderberg ¶6].
Claim 2 is rejected under 35 U.S.C. 103 as being unpatentable over MOUNSAVENG in view of Jaderberg as applied in the rejection of claim 1 above, and further in view of Wang et al. (US 20230095041 A1) hereinafter Wang.
As to dependent claim 2, MOUNSAVENG and Jaderberg teach the method of claim 1 above that is incorporated,
MOUNSAVENG and Jaderberg further teach wherein the processing circuitry is configured to train the parameter output function in such a manner as to decrease a prediction error from each of the object sample and the first augmented sample to the target value. [MOUNSAVENG loss is used to improve classification and reduce error via training ¶134, ¶28 "a first classification loss indicative of a classification error of the classification network. The processor is configured for updating, based on the first classification loss, the set of classification parameters to obtain updated values for the set of classification parameters, classifying, by the classification network, another given digital document of the set of labelled digital documents to obtain a second predicted class"]
MOUNSAVENG and Jaderberg do not specifically teach while decreasing a similarity between the object sample and the first augmented sample.
However, Wang teaches while decreasing a similarity between the object sample and the first augmented sample. [penalizes similarity (decreases) ¶27 "diversity loss penalizes a similarity between two outputted images of the generator for different target style code "]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the neural network disclosed by MOUNSAVENG and Jaderberg by incorporating the while decreasing a similarity between the object sample and the first augmented sample disclosed by Wang because all techniques address the same field of machine learning for image transformation and by incorporating Wang into MOUNSAVENG and Jaderberg enhances images while preserving styles and improve model outputs during training [Wang ¶3-4]
Claim 4 is rejected under 35 U.S.C. 103 as being unpatentable over MOUNSAVENG in view of Jaderberg as applied in the rejection of claim 3 above, and further in view of Yin et al. (US 20230121812 A1) hereinafter Yin.
As to dependent claim 4, MOUNSAVENG and Jaderberg teach the method of claim 3 above that is incorporated,
MOUNSAVENG and Jaderberg do not specifically teach wherein the processing circuitry is configured to determine the third data augmentation parameter, based on a frequency distribution of the second data augmentation parameter, in regard to each of classes of the target value.
However, Yin teaches wherein the processing circuitry is configured to determine the third data augmentation parameter, based on a frequency distribution of the second data augmentation parameter, in regard to each of classes of the target value. [histograms (distributions) metrics for selecting parameters claim 5, ¶29 "features and distribution of data to determine whether the original data set was suitable for training an artificial intelligence model, or if data augmentation is needed at block 2 "]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the neural network disclosed by MOUNSAVENG and Jaderberg by incorporating the wherein the processing circuitry is configured to determine the third data augmentation parameter, based on a frequency distribution of the second data augmentation parameter, in regard to each of classes of the target value disclosed by Yin because all techniques address the same field of machine learning for image transformation and by incorporating Yin into MOUNSAVENG and Jaderberg provides more accurate models with data enhancements [Yin ¶2]
Claim 8 is rejected under 35 U.S.C. 103 as being unpatentable over MOUNSAVENG in view of Jaderberg as applied in the rejection of claim 5 above, and further in view of Gopalan et al. (US 20210064700 A1) hereinafter Gopalan
As to dependent claim 8, MOUNSAVENG and Jaderberg teach the method of claim 5 above that is incorporated,
MOUNSAVENG and Jaderberg do not specifically teach wherein the processing circuitry is configured to train the target value prediction model by excluding the object sample having the second data augmentation parameter corresponding to an outlier.
However, Gopalan teaches wherein the processing circuitry is configured to train the target value prediction model by excluding the object sample having the second data augmentation parameter corresponding to an outlier. [excludes (removes) outliers ¶9 "determine whether a sentence within a set of training data for a machine learning model is an outlier. This process allows the document analysis system to remove sentences that would degrade the accuracy and performance of training data for a machine learning model."]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the neural network disclosed by MOUNSAVENG and Jaderberg by incorporating the wherein the processing circuitry is configured to train the target value prediction model by excluding the object sample having the second data augmentation parameter corresponding to an outlier disclosed by Gopalan because all techniques address the same field of machine learning for image transformation and by incorporating Gopalan into MOUNSAVENG and Jaderberg reduce processing and memory requirement while providing improved understanding and identification of text [Gopalan ¶2-3] .
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. Applicant is required under 37 C.F.R. § 1.111(c) to consider these references fully when responding to this action.
Karulin et al. (US 12046057 B1) teaches training models using augmented image data (see Fig. 3 and Col. 7 ln. 4-15)
It is noted that any citation to specific pages, columns, lines, or figures in the prior art references and any interpretation of the references should not be considered to be limiting in any way. A reference is relevant for all it contains and may be relied upon for all that it would have reasonably suggested to one having ordinary skill in the art. In re Heck, 699 F.2d 1331, 1332-33, 216 U.S.P.Q. 1038, 1039 (Fed. Cir. 1983) (quoting In re Lemelson, 397 F.2d 1006, 1009, 158 U.S.P.Q. 275, 277 (C.C.P.A. 1968)).
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Beau Spratt whose telephone number is 571 272 9919. The examiner can normally be reached 8:30am to 5:00pm (PST).
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, Jennifer Welch can be reached at 571 272 7212. The fax phone number for the organization where this application or proceeding is assigned is 571 483 7388.
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 http://pair-direct.uspto.gov. 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.
/BEAU D SPRATT/Primary Examiner, Art Unit 2143
Prosecution Insights