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 .
Specification
The disclosure is objected to because of the following informality: In paragraph 0019, line 2, “compressors 108, 110, ad 112” should read “compressors 108, 110, and 112”.
Appropriate correction is required.
Claim Objections
Claim 6 objected to because of the following informality: “KL divergences” should define the acronym “KL”. Appropriate correction is required.
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 4, 6-8, 14, and 16-18 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.
The term “substantially” in claim 4, line 5, is a relative term which renders the claim indefinite. The term “substantially” is 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.
Claim 14 recites substantially similar limitations to claim 4, and is therefore rejected under the same analysis.
Claim 6 recites the limitation "the decompressed perturbed sample data" in lines 2-3. There is insufficient antecedent basis for this limitation in the claim. It is unclear if “the decompressed perturbed sample data” is referring to specifically the “perturbed sample data” from claim 1. For the purposes of examination, the Examiner has interpreted these instances and all subsequent instances in the dependent claims as the perturbed sample data.
Claim 16 recites substantially similar limitations to claim 6, and is therefore rejected under the same analysis.
Claim(s) 7-8, 17-18 are rejected for at least the same reasons as claims 6 and 16 since they depend on claims 6 and 16.
Claims 7 and 17 recite the limitation “the relationship is defined as
PNG
media_image1.png
37
351
media_image1.png
Greyscale
”
It is unclear what R(q), the kl function, f(xi), f(x̃i), i, n, or the Average function represent. For purposes of Examination, Examiner has interpreted:
R(q) as the relationship between the compression quality q of a compressor to the average of the KL divergences obtained from the classification of the original data and the classification of the decompressed data (spec paragraph [0037])
The kl function as the KL divergence of two data classifications (spec paragraph [0037])
f(xi) as the classification of the original sample data (spec paragraph [0037])
f(x̃i) as the classification perturbed sample data (spec paragraph [0034])
i as the i-th sample data from either x (original data) or x̃ (perturbed data) (spec paragraph [0037])
n as the total number of samples from the input data (spec paragraph [0028])
and the Average function as a function that calculates the average of the given KL divergences. (spec paragraph [0037])
Claims 8 and 18 recite the limitation “determining the compression quality includes selecting a compression quality that ensures that the global divergence threshold is satisfied or such that
PNG
media_image2.png
21
179
media_image2.png
Greyscale
”
It is unclear what R(q), Lkl , q, or x represent. For purposes of Examination, Examiner has interpreted:
R(q) as the relationship between the compression quality q of a compressor to the average of the KL divergences obtained from the classification of the original data and the classification of the decompressed data (spec paragraph [0037])
Lkl as a global divergence threshold (spec paragraph [0042])
q as the compression quality of a compressor (spec paragraph [0037])
and x as the original data (spec paragraph [0037]).
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claim(s) 1-20 are rejected under 35 U.S.C. 101 because the claims are directed towards an abstract idea without significantly more.
Regarding Claim 1:
Subject Matter Eligibility Analysis Step 1:
Claim 1 recites a method and is thus a process, one of the four statutory categories of patentable subject matter.
Subject Matter Eligibility Analysis Step 2A Prong 1:
Claim 1 recites:
determining a global divergence threshold for input data based on sample data, a compressor, and a tolerance parameter (This limitation is a mental process as it encompasses a human mentally determining a global divergence threshold and is thus an evaluation.)
determining a relationship between a compression quality and divergences generated using the sample data that is processed by a classifier and using perturbed sample data that is processed by the classifier (This limitation is a mental process as it encompasses a human mentally determining a relationship and is thus an evaluation.)
determining a compression quality for the input data based on the global divergence threshold and the relationship (This limitation is a mental process as it encompasses a human mentally determining a compression quality and is thus an evaluation.)
Therefore, claim 1 recites an abstract idea.
Subject Matter Eligibility Analysis Step 2A Prong 2:
Claim 1 further recites additional elements of:
wherein the compressor is configured to compress the input data based on the compression quality (This element does not integrate the abstract idea into a practical application because it amounts to mere “apply it on a computer” (see MPEP 2106.05(f)).)
Therefore, claim 1 is not integrated into a practical application.
Subject Matter Eligibility Analysis Step 2B:
The additional elements of claim 1 do not provide significantly more than the abstract
idea itself, taken alone and in combination because
wherein the compressor is configured to compress the input data based on the compression quality uses a computer as a tool to perform the abstract idea and cannot provide significantly more (see MPEP 2106.05(f)).
Therefore, claim 1 is subject-matter ineligible.
Regarding Claim 2:
Subject Matter Eligibility Analysis Step 2A Prong 1:
Claim 2 recites the same abstract idea as claim 1. Therefore, claim 2 recites an abstract idea.
Subject Matter Eligibility Analysis Step 2A Prong 2:
Claim 2 further recites additional elements of:
wherein the global divergence threshold comprises a global KL (Kullback-Leibler) threshold (This element does not integrate the abstract idea into a practical application because it recites a technological environment in which to apply a judicial exception (see MPEP 2106.05(h)).)
Therefore, claim 2 is not integrated into a practical application.
Subject Matter Eligibility Analysis Step 2B:
The additional elements of claim 2 do not provide significantly more than the abstract
idea itself, taken alone and in combination because
wherein the global divergence threshold comprises a global KL (Kullback-Leibler) threshold specifies a technological environment to perform the abstract idea and cannot provide significantly more (see MPEP 2106.05(h)).
Therefore, claim 2 is subject-matter ineligible.
Regarding Claim 3:
Subject Matter Eligibility Analysis Step 2A Prong 1:
Claim 3 recites:
determining an individual KL threshold for each sample in the sample data and each sample in the perturbed sample data. (This limitation is a mental process as it encompasses a human mentally determining a KL threshold and is thus an evaluation.)
Therefore, claim 3 recites an abstract idea.
Subject Matter Eligibility Analysis Step 2A Prong 2:
Claim 3 does not further recite any additional elements. Therefore, claim 3 is not
integrated into a practical application.
Subject Matter Eligibility Analysis Step 2B:
Since there are no additional elements, claim 3 does not provide significantly more than
the abstract idea itself, taken alone and in combination. Therefore, claim 3 is subject-matter
ineligible.
Regarding Claim 4:
Subject Matter Eligibility Analysis Step 2A Prong 1:
Claim 4 recites:
wherein an accuracy of the classifier is determined according to the tolerance parameter if a perturbation applied to the perturbed sample data satisfies a relationship such that a divergence between results of the classifier applied to the sample data and results of the classifier applied to the perturbed data is smaller than the global divergence threshold for substantially all samples in the sample data (This limitation is a mental process as it encompasses a human mentally determining an accuracy and is thus an evaluation.)
Therefore, claim 4 recites an abstract idea.
Subject Matter Eligibility Analysis Step 2A Prong 2:
Claim 4 does not further recite any additional elements. Therefore, claim 4 is not
integrated into a practical application.
Subject Matter Eligibility Analysis Step 2B:
Since there are no additional elements, claim 4 does not provide significantly more than
the abstract idea itself, taken alone and in combination. Therefore, claim 4 is subject-matter
ineligible.
Regarding Claim 5:
Subject Matter Eligibility Analysis Step 2A Prong 1:
Claim 5 recites:
wherein determining the global divergence threshold includes evaluating the sample data with the classifier and evaluating the perturbed sample data with the classifier (This limitation is a mental process as it encompasses a human mentally determining a threshold and is thus an evaluation.)
Therefore, claim 5 recites an abstract idea.
Subject Matter Eligibility Analysis Step 2A Prong 2:
Claim 5 further recites additional elements of:
wherein the perturbed sample data has been compressed and decompressed prior to processing by the classifier (This element does not integrate the abstract idea into a practical application because it amounts to mere “apply it on a computer” (see MPEP 2106.05(f)).)
Therefore, claim 5 is not integrated into a practical application.
Subject Matter Eligibility Analysis Step 2B:
The additional elements of claim 5 do not provide significantly more than the abstract
idea itself, taken alone and in combination because
wherein the perturbed sample data has been compressed and decompressed prior to processing by the classifier uses a computer as a tool to perform the abstract idea and cannot provide significantly more (see MPEP 2106.05(f)).
Therefore, claim 5 is subject-matter ineligible.
Regarding Claim 6:
Subject Matter Eligibility Analysis Step 2A Prong 1:
Claim 6 recites:
determining average KL divergences between classifying the original sample data with the classifier and classifying the decompressed perturbed sample data with the classifier for each of multiple compression quality values applied to the compressor. (This limitation is a mental process as it encompasses a human mentally determining average KL divergences between two classifications and is thus an evaluation.)
Therefore, claim 6 recites an abstract idea.
Subject Matter Eligibility Analysis Step 2A Prong 2:
Claim 6 does not further recite any additional elements. Therefore, claim 6 is not
integrated into a practical application.
Subject Matter Eligibility Analysis Step 2B:
Since there are no additional elements, claim 6 does not provide significantly more than
the abstract idea itself, taken alone and in combination. Therefore, claim 6 is subject-matter
ineligible.
Regarding Claim 7:
Subject Matter Eligibility Analysis Step 2A Prong 1:
Claim 7 recites:
the relationship is defined as
PNG
media_image3.png
28
345
media_image3.png
Greyscale
(This limitation is a mathematical concept as it encompasses a mathematical formula.)
Therefore, claim 7 recites an abstract idea.
Subject Matter Eligibility Analysis Step 2A Prong 2:
Claim 7 does not further recite any additional elements. Therefore, claim 7 is not
integrated into a practical application.
Subject Matter Eligibility Analysis Step 2B:
Since there are no additional elements, claim 7 does not provide significantly more than
the abstract idea itself, taken alone and in combination. Therefore, claim 7 is subject-matter
ineligible.
Regarding Claim 8:
Subject Matter Eligibility Analysis Step 2A Prong 1:
Claim 8 recites:
wherein determining the compression quality includes selecting a compression quality that ensures that the global divergence threshold is satisfied (This limitation is a mental process as it encompasses a human mentally determining and selecting a compression quality and is thus an evaluation.)
or such that
PNG
media_image4.png
21
179
media_image4.png
Greyscale
(This limitation is a mathematical concept as it encompasses a mathematical formula.)
Therefore, claim 8 recites an abstract idea.
Subject Matter Eligibility Analysis Step 2A Prong 2:
Claim 8 does not further recite any additional elements. Therefore, claim 8 is not
integrated into a practical application.
Subject Matter Eligibility Analysis Step 2B:
Since there are no additional elements, claim 8 does not provide significantly more than
the abstract idea itself, taken alone and in combination. Therefore, claim 8 is subject-matter
ineligible.
Regarding Claim 9:
Subject Matter Eligibility Analysis Step 2A Prong 1:
Claim 9 recites the same abstract idea as claim 1. Therefore, claim 9 recites an abstract idea.
Subject Matter Eligibility Analysis Step 2A Prong 2:
Claim 9 further recites additional elements of:
further comprising compressing the input data based on the compression quality (This element does not integrate the abstract idea into a practical application because it amounts to mere “apply it on a computer” (see MPEP 2106.05(f)).)
Therefore, claim 9 is not integrated into a practical application.
Subject Matter Eligibility Analysis Step 2B:
The additional elements of claim 9 do not provide significantly more than the abstract
idea itself, taken alone and in combination because
further comprising compressing the input data based on the compression quality uses a computer as a tool to perform the abstract idea and cannot provide significantly more (see MPEP 2106.05(f)).
Therefore, claim 9 is subject-matter ineligible.
Regarding Claim 10:
Subject Matter Eligibility Analysis Step 2A Prong 1:
Claim 10 recites the same abstract idea as claim 1. Therefore, claim 10 recites an abstract idea.
Subject Matter Eligibility Analysis Step 2A Prong 2:
Claim 10 further recites additional elements of:
further comprising dynamically adjusting the compression quality based on network conditions (This element does not integrate the abstract idea into a practical application because it amounts to mere “apply it on a computer” (see MPEP 2106.05(f)).)
Therefore, claim 10 is not integrated into a practical application.
Subject Matter Eligibility Analysis Step 2B:
The additional elements of claim 10 do not provide significantly more than the abstract
idea itself, taken alone and in combination because
further comprising dynamically adjusting the compression quality based on network conditions uses a computer as a tool to perform the abstract idea and cannot provide significantly more (see MPEP 2106.05(f)).
Therefore, claim 10 is subject-matter ineligible.
Regarding Claim 11:
Subject Matter Eligibility Analysis Step 1:
Claim 11 recites a non-transitory storage medium and is thus a machine, one of the four statutory categories of patentable subject matter.
Subject Matter Eligibility Analysis Step 2A Prong 1:
Claim 11 recites:
determining a global divergence threshold for input data based on sample data, a compressor, and a tolerance parameter (This limitation is a mental process as it encompasses a human mentally determining a global divergence threshold and is thus an evaluation.)
determining a relationship between a compression quality and divergences generated using the sample data that is processed by a classifier and using perturbed sample data that is processed by the classifier (This limitation is a mental process as it encompasses a human mentally determining a relationship and is thus an evaluation.)
determining a compression quality for the input data based on the global divergence threshold and the relationship (This limitation is a mental process as it encompasses a human mentally determining a compression quality and is thus an evaluation.)
Therefore, claim 11 recites an abstract idea.
Subject Matter Eligibility Analysis Step 2A Prong 2:
Claim 11 further recites additional elements of:
A non-transitory storage medium having stored therein instructions that are executable by one or more hardware processors to perform operations comprising (This element does not integrate the abstract idea into a practical application because it amounts to mere "apply it on a computer" (see MPEP 2106.05(f)).)
wherein the compressor is configured to compress the input data based on the compression quality (This element does not integrate the abstract idea into a practical application because it amounts to mere “apply it on a computer” (see MPEP 2106.05(f)).)
Therefore, claim 11 is not integrated into a practical application.
Subject Matter Eligibility Analysis Step 2B:
The additional elements of claim 11 do not provide significantly more than the abstract
idea itself, taken alone and in combination because
A non-transitory storage medium having stored therein instructions that are executable by one or more hardware processors to perform operations comprising uses a hardware device and instructions as tools to perform the abstract idea and cannot provide significantly more (see MPEP 2106.05(f)).)
wherein the compressor is configured to compress the input data based on the compression quality uses a computer as a tool to perform the abstract idea and cannot provide significantly more (see MPEP 2106.05(f)).
Therefore, claim 11 is subject-matter ineligible.
Claim(s) 12 recites substantially similar limitations to claim(s) 2, and is/are therefore rejected under the same analysis.
Claim 13 recites substantially similar limitations to claim 3, and is therefore rejected under the same analysis.
Claim(s) 14 recites substantially similar limitations to claim(s) 4, and is/are therefore rejected under the same analysis.
Claim(s) 15 recites substantially similar limitations to claim(s) 5, and is/are therefore rejected under the same analysis.
Claim 16 recites substantially similar limitations to claim 6, and is therefore rejected under the same analysis.
Claim 17 recites substantially similar limitations to claim 7, and is therefore rejected under the same analysis.
Claim 18 recites substantially similar limitations to claim 8, and is therefore rejected under the same analysis.
Claim(s) 19 recites substantially similar limitations to claim(s) 9, and is/are therefore rejected under the same analysis.
Claim(s) 20 recites substantially similar limitations to claim(s) 10, and is/are therefore rejected under the same analysis.
Claim Rejections - 35 USC § 102
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claim(s) 1 and 9 is/are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Rahman et al. (“Dynamic Error-bounded Lossy Compression (EBLC) to Reduce the Bandwidth Requirement for Real-time Vision based Pedestrian Safety Applications”).
Regarding Claim 1, Rahman teaches
a method comprising: determining a global divergence threshold for input data based on sample data, a compressor, and a tolerance parameter (Rahman, Page 3, “The edge computing infrastructure selects an appropriate model from the set of pre-trained and calibrated models. In addition, it determines the corresponding PSNR for the model that yields the largest reductions in bandwidth while still maintaining the same detection accuracy” Rahman, Page 3, “The video compression unit compresses the raw video stream using a set tolerance level. In our experiments, we set the tolerance based on the PSNR ratio between the raw video and the resulting compressed video.” Examiner notes that the global divergence threshold is the PSNR, the sample data is the video stream, the compressor is the video compression unit, and the tolerance parameter is the set tolerance level.);
determining a relationship between a compression quality and divergences generated using the sample data that is processed by a classifier and using perturbed sample data that is processed by the classifier (Rahman, page 4, “Using field-collected data, we compress each video using different Constant Rate Factor (CRF) values using the FFmpeg video compression tool [27]. The video compression level is controlled by the CRF value, and the CRF range is from 0 to 51; where 0 indicates no compression, and 51 indicates maximum compression level. After that, we calculate the PSNR by comparing the original video file and compressed video file.” Rahman, page 4, Fig. 3. Examiner notes that the sample data is the original video file (compressed using a CRF value of 0) and the perturbed sample data is the compressed video file (compressed using a CRF value greater than 0), both of which are processed by the environment classification model in Figure 3. Examiner further notes that the compression quality is the CRF value for each compressed video, and the divergence is the PSNR.)
and determining a compression quality for the input data based on the global divergence threshold and the relationship, (Rahman, Page 4, “To construct the reference table and the catalog of corresponding models, we train and evaluate a model on data compressed with a CRF of 10 (highly accurate) along with computing the PSNR. Next, we increase the CRF by 10 (degrading video quality and improving compression) until the new model’s detection accuracy drops below the minimum threshold. At this point, we vary the CRF by 1 to fully explore the range between the last valid CRF and the first invalid CRF. Again, we evaluate each model to determine if it meets our quality-of-service standards; rejecting any models that do not. After exploring each CRF in the interval, we have a table that allows us to select a trained model given a requested CRF or PSNR value.”)
wherein the compressor is configured to compress the input data based on the compression quality (Rahman, Page 4, “Using field-collected data, we compress each video using different Constant Rate Factor (CRF) values using the FFmpeg video compression tool [27].”)
Regarding Claim 9, Rahman teaches
the method of claim 1, further comprising compressing the input data based on the compression quality (Rahman, Page 4, “Using field-collected data, we compress each video using different Constant Rate Factor (CRF) values using the FFmpeg video compression tool [27].”);
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claim(s) 2-4, 6, 9, 11, 12-14, 16, and 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rahman in view of Cooper et al. (US 11385794 B2) (hereafter referred to as Cooper).
Regarding Claim 2, Rahman teaches the method of claim 1. Rahman does not teach the usage of a KL (Kullback-Leibler) method. However, Cooper teaches
wherein the global divergence threshold comprises a global KL (Kullback-Leibler) threshold (Cooper, Paragraph 92, “Once stored, the test dataset may be forwarded to a statistical analysis engine 2920 which may utilize one or more algorithms to determine the probability distribution of the test dataset. Best-practice probability distribution algorithms such as Kullback-Leibler divergence, adaptive windowing, and Jensen-Shannon divergence may be used to compute the probability distribution of training and test datasets.” Cooper, paragraph 93, “After statistical analysis engine 2920 calculates the probability distribution of the test dataset it may retrieve from monitor database 2930 the calculated and stored probability distribution of the current training dataset. It may then compare the two probability distributions of the two different datasets in order to verify if the difference in calculated distributions exceeds a predetermined difference threshold.”)
Cooper and Rahman are considered analogous to the claimed invention because they deal with optimizing accuracy in data compression. It would have been obvious to one with ordinary skill in the art at the time of the effective filing date to modify Rahman to substitute Kullback-Leibler divergence as a performance metric in place of PSNR like Cooper as it “provides substantial improvements in data compaction” (Cooper, paragraph 41).
Regarding Claim 3, Rahman and Cooper teach the method of claim 2. Rahman further teaches
further comprising determining an individual KL threshold for each sample in the sample data and each sample in the perturbed sample data. (Rahman, Page 4, “Using field-collected data, we compress each video using different Constant Rate Factor (CRF) values using the FFmpeg video compression tool [27]. The video compression level is controlled by the CRF value, and the CRF range is from 0 to 51; where 0 indicates no compression, and 51 indicates maximum compression level. After that, we calculate the PSNR by comparing the original video file and compressed video file.” Examiner notes that the original video files are the samples and the compressed video files are the perturbed samples. Examiner further notes that the PSNR is the metric calculated for each of the compressed and uncompressed video files.)
Rahman does not teach “determining an individual KL threshold.” However, Cooper teaches
determining an individual KL threshold (Cooper, Paragraph 92, “Once stored, the test dataset may be forwarded to a statistical analysis engine 2920 which may utilize one or more algorithms to determine the probability distribution of the test dataset. Best-practice probability distribution algorithms such as Kullback-Leibler divergence, adaptive windowing, and Jensen-Shannon divergence may be used to compute the probability distribution of training and test datasets.” Cooper, paragraph 93, “After statistical analysis engine 2920 calculates the probability distribution of the test dataset it may retrieve from monitor database 2930 the calculated and stored probability distribution of the current training dataset. It may then compare the two probability distributions of the two different datasets in order to verify if the difference in calculated distributions exceeds a predetermined difference threshold.”)
Cooper and Rahman are considered analogous to the claimed invention because they deal with optimizing accuracy in data compression. It would have been obvious to one with ordinary skill in the art to modify Rahman to substitute Kullback-Leibler divergence as a performance metric in place of PSNR like Cooper as it “provides substantial improvements in data compaction” (Cooper, paragraph 41).
Regarding Claim 4, Rahman and Cooper teach the method of claim 3. Rahman further teaches
wherein an accuracy of the classifier is determined according to the tolerance parameter (Rahman, page 3, “The video compression unit compresses the raw video stream using a set tolerance level. In our experiments, we set the tolerance based on the PSNR ratio between the raw video and the resulting compressed video.” Rahman, Page 4, “We calculate the accuracy of the pedestrian detection model by comparing it with manually annotated ground truth data. To establish a baseline accuracy, we perform pedestrian detection on the uncompressed video feed coming from traffic cameras for all scenarios and calculate the accuracy based on a manually annotated ground truth. For a compression baseline, we compress the video stream to a fixed quality level using standard image difference metric, PSNR, and use a pedestrian detection model with weights calibrated for the compressed data.” Examiner notes that the compression baseline accuracy of the classifier is dependent on the compression of the data, which is based on the tolerance level.)
Rahman does not, but Cooper teaches
if a perturbation applied to the perturbed sample data satisfies a relationship such that a divergence between results of the classifier applied to the sample data and results of the classifier applied to the perturbed data is smaller than the global divergence threshold for substantially all samples in the sample data (Cooper, paragraph 91, “Based on the results of the analyses, the codebook training module 2830 may create a new training dataset from a subset of the requested data in order to counteract the effects of data drift on the encoding/decoding models, and then publish updated 2850 codebooks to both the encoding machine 2810 and decoding machine 2820.” Cooper, Paragraph 92, “a data collector 2910 is present which may send requests for incoming data 2905 to a data deconstruction engine 102 which may receive the request and route incoming data to codebook training module 2900 where it may be received by data collector 2910. Data collector 2910 may be configured to request data periodically such as at schedule time intervals, or for example, it may be configured to request data after a certain amount of data has been processed through the encoding machine 2810 or decoding machine 2820. The received data may be a plurality of sourceblocks, which are a series of binary digits, originating from a source packet otherwise referred to as a datagram. The received data may compiled into a test dataset and temporarily stored in a cache 2970. Once stored, the test dataset may be forwarded to a statistical analysis engine 2920 which may utilize one or more algorithms to determine the probability distribution of the test dataset. Best-practice probability distribution algorithms such as Kullback-Leibler divergence, adaptive windowing, and Jensen-Shannon divergence may be used to compute the probability distribution of training and test datasets.” Cooper, paragraph 93, “After statistical analysis engine 2920 calculates the probability distribution of the test dataset it may retrieve from monitor database 2930 the calculated and stored probability distribution of the current training dataset. It may then compare the two probability distributions of the two different datasets in order to verify if the difference in calculated distributions exceeds a predetermined difference threshold. If the difference in distributions does not exceed the difference threshold, that indicates the test dataset, and therefore the incoming data, has not experienced enough data drift to cause the encoding/decoding system performance to degrade significantly, which indicates that no updates are necessary to the existing codebooks. However, if the difference threshold has been surpassed, then the data drift is significant enough to cause the encoding/decoding system performance to degrade to the point where the existing models and accompanying codebooks need to be updated.” Examiner notes that the sample data is the test dataset, the perturbed sample data is the training dataset (created by the training module), the classifier is the encoding/decoding machine that the requested data must go through, the global divergence is the difference in probability distributions between the two datasets, and the global divergence threshold is the difference threshold, wherein the difference must be equal to or less than the difference threshold.)
Cooper and Rahman are considered analogous to the claimed invention because they deal with optimizing accuracy in data compression. It would have been obvious to one with ordinary skill in the art at the time of the effective filing date to modify Rahman to use Kullback-Leibler divergence as a performance metric and implement a maximum threshold value like Cooper in order to optimize compression while preserving perturbed data similarity to the original uncompressed data. Doing so would be advantageous because it “provides substantial improvements in data compaction” (Cooper, paragraph 41).
Regarding Claim 6, Rahman teaches
The method of claim 1, further comprising determining average KL divergences between classifying the original sample data with the classifier and classifying the decompressed perturbed sample data with the classifier for each of multiple compression quality values applied to the compressor. (Rahman, page 4, “Using field-collected data, we compress each video using different Constant Rate Factor (CRF) values using the FFmpeg video compression tool [27]. The video compression level is controlled by the CRF value, and the CRF range is from 0 to 51; where 0 indicates no compression, and 51 indicates maximum compression level. After that, we calculate the PSNR by comparing the original video file and compressed video file.” Examiner notes "the decompressed perturbed sample data" is interpreted as "the perturbed sample data" from Claim 1 for lack of antecedent basis. Examiner further notes the original sample data is the original video file, the decompressed perturbed sample data is the compressed video file, and the multiple compression quality values are the CRF values from 0 to 51.)
Rahman does not teach “determining average KL divergences”. However, Cooper teaches
determining average KL divergences (Cooper, Paragraph 0149, “Best-practice probability distribution metrics such as Kullback-Leibler divergence, adaptive windowing, and Jensen-Shannon divergence may be used to compute and/or estimate the probability distribution of training and test datasets. These metrics may also be used to estimate the probability distribution from the current run-time data. In some implementations, the estimate of the training data may be compared against the estimate of the run-time data to verify if the difference in calculated distributions exceeds a predetermined difference threshold.”)
Cooper and Rahman are considered analogous to the claimed invention because they deal with optimizing accuracy in data compression. It would have been obvious to one with ordinary skill in the art at the time of the effective filing date to modify Rahman to use Kullback-Leibler divergence as a performance metric rather than PSNR as it “provides substantial improvements in data compaction” (Cooper, paragraph 41).
Regarding Claim 11, Rahman teaches
determining a global divergence threshold for input data based on sample data, a compressor, and a tolerance parameter (Rahman, Page 3, “The edge computing infrastructure selects an appropriate model from the set of pre-trained and calibrated models. In addition, it determines the corresponding PSNR for the model that yields the largest reductions in bandwidth while still maintaining the same detection accuracy” Rahman, Page 3, “The video compression unit compresses the raw video stream using a set tolerance level. In our experiments, we set the tolerance based on the PSNR ratio between the raw video and the resulting compressed video.” Examiner notes that the global divergence threshold is the PSNR, the sample data is the video stream, the compressor is the video compression unit, and the tolerance parameter is the set tolerance level.);
determining a relationship between a compression quality and divergences generated using the sample data that is processed by a classifier and using perturbed sample data that is processed by the classifier (Rahman, page 4, “Using field-collected data, we compress each video using different Constant Rate Factor (CRF) values using the FFmpeg video compression tool [27]. The video compression level is controlled by the CRF value, and the CRF range is from 0 to 51; where 0 indicates no compression, and 51 indicates maximum compression level. After that, we calculate the PSNR by comparing the original video file and compressed video file.” Rahman, page 4, Fig. 3. Examiner notes that the sample data is the original video file (compressed using a CRF value of 0) and the perturbed sample data is the compressed video file (compressed using a CRF value greater than 0), both of which are processed by the environment classification model in Figure 3. Examiner further notes that the compression quality is the CRF value for each compressed video, and the divergence is the PSNR.)
and determining a compression quality for the input data based on the global divergence threshold and the relationship (Rahman, Page 4, “To construct the reference table and the catalog of corresponding models, we train and evaluate a model on data compressed with a CRF of 10 (highly accurate) along with computing the PSNR. Next, we increase the CRF by 10 (degrading video quality and improving compression) until the new model’s detection accuracy drops below the minimum threshold. At this point, we vary the CRF by 1 to fully explore the range between the last valid CRF and the first invalid CRF. Again, we evaluate each model to determine if it meets our quality-of-service standards; rejecting any models that do not. After exploring each CRF in the interval, we have a table that allows us to select a trained model given a requested CRF or PSNR value.”)
wherein the compressor is configured to compress the input data based on the compression quality. (Rahman, Page 4, “Using field-collected data, we compress each video using different Constant Rate Factor (CRF) values using the FFmpeg video compression tool [27].”)
Rahman does not teach “a non-transitory storage medium having stored therein instructions that are executable by one or more hardware processors to perform operations”. However, Cooper teaches
A non-transitory storage medium having stored therein instructions that are executable by one or more hardware processors to perform operations (Cooper, Paragraph 9, “According to a preferred embodiment, a system for encoding data using a plurality of codebooks is disclosed, comprising: a computing device comprising a processor, a memory, and a non-volatile data storage device; a codebook selector comprising a first plurality of programming instructions stored in the memory and operable on the processor, wherein the first plurality of programming instructions, when operating on the processor, causes the processor to”)
Rahman teaches a method for optimizing data compression based on PSNR, an accuracy metric, for real world data. Cooper teaches a computer medium with instructions to run a method on adaptive data compression optimization based on one or more accuracy metrics. Rahman and Cooper both teach optimizing data compression based on an accuracy metric and thus are considered analogous to the claimed invention. It would have been obvious to one with ordinary skill in the art at the time of the effective filing date to modify Rahman to implement the method on a software medium as this would be applying a known technique to a known device to yield the predictable result of running software instructions on a computer medium successfully (see MPEP 2143(d)).
Regarding Claim 12, Rahman and Cooper teach the non-transitory storage medium of claim 11. Rahman does not teach “wherein the global divergence threshold comprises a global KL (Kullback-Leibler) threshold.” However, Cooper teaches
wherein the global divergence threshold comprises a global KL (Kullback-Leibler) threshold (Cooper, Paragraph 92, “Once stored, the test dataset may be forwarded to a statistical analysis engine 2920 which may utilize one or more algorithms to determine the probability distribution of the test dataset. Best-practice probability distribution algorithms such as Kullback-Leibler divergence, adaptive windowing, and Jensen-Shannon divergence may be used to compute the probability distribution of training and test datasets.” Cooper, paragraph 93, “After statistical analysis engine 2920 calculates the probability distribution of the test dataset it may retrieve from monitor database 2930 the calculated and stored probability distribution of the current training dataset. It may then compare the two probability distributions of the two different datasets in order to verify if the difference in calculated distributions exceeds a predetermined difference threshold.”)
Cooper and Rahman are considered analogous to the claimed invention because they deal with optimizing accuracy in data compression. It would have been obvious to one with ordinary skill in the art at the time of the effective filing date to modify Rahman to substitute Kullback-Leibler divergence as a performance metric in place of PSNR like Cooper as it “provides substantial improvements in data compaction” (Cooper, paragraph 41).
Regarding Claim 13, Rahman and Cooper teach the non-transitory storage medium of claim 12. Rahman further teaches
further comprising determining an individual KL threshold for each sample in the sample data and each sample in the perturbed sample data (Rahman, Page 4, “Using field-collected data, we compress each video using different Constant Rate Factor (CRF) values using the FFmpeg video compression tool [27]. The video compression level is controlled by the CRF value, and the CRF range is from 0 to 51; where 0 indicates no compression, and 51 indicates maximum compression level. After that, we calculate the PSNR by comparing the original video file and compressed video file.” Examiner notes that the original video files are the samples and the compressed video files are the perturbed samples. Examiner further notes that the PSNR is the metric calculated for each of the compressed and uncompressed video files.)
Rahman does not teach “determining an individual KL threshold.” However, Cooper teaches
determining an individual KL threshold (Cooper, Paragraph 92, “Once stored, the test dataset may be forwarded to a statistical analysis engine 2920 which may utilize one or more algorithms to determine the probability distribution of the test dataset. Best-practice probability distribution algorithms such as Kullback-Leibler divergence, adaptive windowing, and Jensen-Shannon divergence may be used to compute the probability distribution of training and test datasets.” Cooper, paragraph 93, “After statistical analysis engine 2920 calculates the probability distribution of the test dataset it may retrieve from monitor database 2930 the calculated and stored probability distribution of the current training dataset. It may then compare the two probability distributions of the two different datasets in order to verify if the difference in calculated distributions exceeds a predetermined difference threshold.”)
Cooper and Rahman are considered analogous to the claimed invention because they deal with optimizing accuracy in data compression. It would have been obvious to one with ordinary skill in the art at the time of the effective filing date to modify Rahman to substitute Kullback-Leibler divergence as a performance metric in place of PSNR like Cooper. This would have been advantageous as it “provides substantial improvements in data compaction” (Cooper, paragraph 41).
Regarding Claim 14, Rahman and Cooper teach the non-transitory storage medium of claim 13. Rahman further teaches
wherein an accuracy of the classifier is determined according to the tolerance parameter (Rahman, page 3, “The video compression unit compresses the raw video stream using a set tolerance level. In our experiments, we set the tolerance based on the PSNR ratio between the raw video and the resulting compressed video.” Rahman, Page 4, “We calculate the accuracy of the pedestrian detection model by comparing it with manually annotated ground truth data. To establish a baseline accuracy, we perform pedestrian detection on the uncompressed video feed coming from traffic cameras for all scenarios and calculate the accuracy based on a manually annotated ground truth. For a compression baseline, we compress the video stream to a fixed quality level using standard image difference metric, PSNR, and use a pedestrian detection model with weights calibrated for the compressed data.” Examiner notes that the compression baseline accuracy of the classifier is dependent on the compression of the data, which is based on the tolerance level.)
Rahman does not, but Cooper teaches
if a perturbation applied to the perturbed sample data satisfies a relationship such that a divergence between results of the classifier applied to the sample data and results of the classifier applied to the perturbed data is smaller than the global divergence threshold for substantially all samples in the sample data (Cooper, paragraph 91, “Based on the results of the analyses, the codebook training module 2830 may create a new training dataset from a subset of the requested data in order to counteract the effects of data drift on the encoding/decoding models, and then publish updated 2850 codebooks to both the encoding machine 2810 and decoding machine 2820.” Cooper, Paragraph 92, “a data collector 2910 is present which may send requests for incoming data 2905 to a data deconstruction engine 102 which may receive the request and route incoming data to codebook training module 2900 where it may be received by data collector 2910. Data collector 2910 may be configured to request data periodically such as at schedule time intervals, or for example, it may be configured to request data after a certain amount of data has been processed through the encoding machine 2810 or decoding machine 2820. The received data may be a plurality of sourceblocks, which are a series of binary digits, originating from a source packet otherwise referred to as a datagram. The received data may compiled into a test dataset and temporarily stored in a cache 2970. Once stored, the test dataset may be forwarded to a statistical analysis engine 2920 which may utilize one or more algorithms to determine the probability distribution of the test dataset. Best-practice probability distribution algorithms such as Kullback-Leibler divergence, adaptive windowing, and Jensen-Shannon divergence may be used to compute the probability distribution of training and test datasets.” Cooper, paragraph 93, “After statistical analysis engine 2920 calculates the probability distribution of the test dataset it may retrieve from monitor database 2930 the calculated and stored probability distribution of the current training dataset. It may then compare the two probability distributions of the two different datasets in order to verify if the difference in calculated distributions exceeds a predetermined difference threshold. If the difference in distributions does not exceed the difference threshold, that indicates the test dataset, and therefore the incoming data, has not experienced enough data drift to cause the encoding/decoding system performance to degrade significantly, which indicates that no updates are necessary to the existing codebooks. However, if the difference threshold has been surpassed, then the data drift is significant enough to cause the encoding/decoding system performance to degrade to the point where the existing models and accompanying codebooks need to be updated.” Examiner notes that the sample data is the test dataset, the perturbed sample data is the training dataset (created by the training module), the classifier is the encoding/decoding machine that the requested data must go through, the global divergence is the difference in probability distributions between the two datasets, and the global divergence threshold is the difference threshold, wherein the difference must be equal to or less than the difference threshold.)
Cooper and Rahman are considered analogous to the claimed invention because they deal with optimizing accuracy in data compression. It would have been obvious to one with ordinary skill in the art at the time of the effective filing date to modify Rahman to use Kullback-Leibler divergence as a performance metric and implement a maximum threshold value like Cooper in order to optimize compression while preserving perturbed data similarity to the original uncompressed data. Doing so would be advantageous because it “provides substantial improvements in data compaction” (Cooper, paragraph 41).
Regarding Claim 16, Rahman and Cooper teach the non-transitory storage medium of claim 11. Rahman further teaches
further comprising determining average KL divergences between classifying the original sample data with the classifier and classifying the decompressed perturbed sample data with the classifier for each of multiple compression quality values applied to the compressor. (Rahman, page 4, “Using field-collected data, we compress each video using different Constant Rate Factor (CRF) values using the FFmpeg video compression tool [27]. The video compression level is controlled by the CRF value, and the CRF range is from 0 to 51; where 0 indicates no compression, and 51 indicates maximum compression level. After that, we calculate the PSNR by comparing the original video file and compressed video file.” Examiner notes "the decompressed perturbed sample data" is interpreted as "the perturbed sample data" from Claim 1 for lack of antecedent basis. Examiner further notes the original sample data is the original video file, the decompressed perturbed sample data is the compressed video file, and the multiple compression quality values are the CRF values from 0 to 51.)
Rahman does not teach “determining average KL divergences”. However, Cooper teaches
determining average KL divergences (Cooper, Paragraph 92, “the test dataset may be forwarded to a statistical analysis engine 2920 which may utilize one or more algorithms to determine the probability distribution of the test dataset. Best-practice probability distribution algorithms such as Kullback-Leibler divergence, adaptive windowing, and Jensen-Shannon divergence may be used to compute the probability distribution of training and test datasets.”)
Cooper and Rahman are considered analogous to the claimed invention because they deal with optimizing accuracy in data compression. It would have been obvious to one with ordinary skill in the art at the time of the effective filing date to modify Rahman to use Kullback-Leibler divergence as a performance metric rather than PSNR as it “provides substantial improvements in data compaction” (Cooper, paragraph 41).
Regarding Claim 19, Rahman and Cooper teach the non-transitory storage medium of claim 11. Rahman further teaches
further comprising compressing the input data based on the compression quality (Rahman, Page 4, “Using field-collected data, we compress each video using different Constant Rate Factor (CRF) values using the FFmpeg video compression tool [27].”);
Claim(s) 10 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rahman in view of Choi et al. (“Delay-Sensitive and Power-Efficient Quality Control of Dynamic Video Streaming using Adaptive Super-Resolution”) (hereafter referred to as Choi).
Regarding Claim 10, Rahman teaches the method of claim 1.
Rahman does not teach “dynamically adjusting the compression quality based on network conditions.” However, Choi teaches
further comprising dynamically adjusting the compression quality based on network conditions (Choi, page 2, “This paper proposes the adaptive quality control of video chunks depending on the time-varying network condition and both the transmitter and the receiver states. We allow the transmitter to determine the video quality enhancement rate at the receiver as well as the compression rate at the transmitter, and it is beneficial to control a variety of performance metrics while improving the average quality measure.”)
Rahman and Choi are considered analogous to the claimed invention because they deal with optimizing video compression to maximize quality in real time. It would have been obvious to one with ordinary skill in the art at the time of the effective filing date to modify Rahman to further adjust the compression quality based on network conditions as doing so “strikes a balance among a variety of conflicting performance metrics of video streaming carefully, i.e., video quality, queuing delay, chunk processing delay, playback stall, power consumption, and CPU usage” (Choi, page 13).
Claim(s) 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rahman in view of Cooper in further view of Choi.
Regarding Claim 20, Rahman and Cooper teach the non-transitory storage medium of claim 11.
Rahman and Cooper do not teach “dynamically adjusting the compression quality based on network conditions.” However, Choi teaches
further comprising dynamically adjusting the compression quality based on network conditions (Choi, page 2, “This paper proposes the adaptive quality control of video chunks depending on the time-varying network condition and both the transmitter and the receiver states. We allow the transmitter to determine the video quality enhancement rate at the receiver as well as the compression rate at the transmitter, and it is beneficial to control a variety of performance metrics while improving the average quality measure.”)
Rahman, Cooper, and Choi are considered analogous to the claimed invention because they deal with optimizing video compression to maximize quality. It would have been obvious to one with ordinary skill in the art at the time of the effective filing date to modify Rahman and Cooper to further adjust the compression quality based on network conditions as doing so “strikes a balance among a variety of conflicting performance metrics of video streaming carefully, i.e., video quality, queuing delay, chunk processing delay, playback stall, power consumption, and CPU usage” (Choi, page 13).
Claim(s) 5 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rahman in view of Shabara et al. (US 20240356608).
Regarding Claim 5, Rahman teaches
The method of claim 1, wherein determining the global divergence threshold includes evaluating the sample data with the classifier and evaluating the perturbed sample data with the classifier, wherein the perturbed sample data has been compressed and decompressed prior to processing by the classifier
(Rahman, page 4, “Using field-collected data, we compress each video using different Constant Rate Factor (CRF) values using the FFmpeg video compression tool [27]. The video compression level is controlled by the CRF value, and the CRF range is from 0 to 51; where 0 indicates no compression, and 51 indicates maximum compression level. After that, we calculate the PSNR by comparing the original video file and compressed video file.” Rahman, page 4, Fig. 3. Examiner notes that the sample data is the uncompressed video file (compressed using a CRF value of 0) and the perturbed sample data is the compressed video file (compressed using a CRF value greater than 0), both of which are processed by the environment classification model in Figure 3.)
Rahman does not, but Shabara teaches
wherein the perturbed sample data has been compressed and decompressed prior to processing by the classifier (Shabara, paragraph 0055, “a CSI classifier algorithm can be used to select a best pair to use. Each encoder-decoder pair can be specialized in compressing and decompressing a corresponding class of data that is classified by the CSI classifier algorithm.”)
Rahman and Shabara are considered analogous to the claimed invention because they deal with classifying compressed data with a neural network. It would have been obvious to one with ordinary skill in the art at the time of the effective filing date to modify Rahman to decompress the sample data before evaluating it with the classifier like Shabara because it is a well-known practice to reduce storage and transmission costs while maintaining the original data format the classifier was trained on.
Claim(s) 15 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rahman in view of Cooper in further view of Shabara.
Regarding Claim 15, Rahman and Cooper teach
The non-transitory storage medium of claim 11, wherein determining the global divergence threshold includes evaluating the sample data with the classifier and evaluating the perturbed sample data with the classifier, wherein the perturbed sample data has been compressed and decompressed prior to processing by the classifier (Rahman, page 4, “Using field-collected data, we compress each video using different Constant Rate Factor (CRF) values using the FFmpeg video compression tool [27]. The video compression level is controlled by the CRF value, and the CRF range is from 0 to 51; where 0 indicates no compression, and 51 indicates maximum compression level. After that, we calculate the PSNR by comparing the original video file and compressed video file.” Rahman, page 4, Fig. 3. Examiner notes that the sample data is the uncompressed video file (compressed using a CRF value of 0) and the perturbed sample data is the compressed video file (compressed using a CRF value greater than 0, both of which are processed by the environment classification model in Figure 3.)
Rahman and Cooper do not, but Shabara teaches
wherein the perturbed sample data has been compressed and decompressed prior to processing by the classifier (Shabara, paragraph 0055, “a CSI classifier algorithm can be used to select a best pair to use. Each encoder-decoder pair can be specialized in compressing and decompressing a corresponding class of data that is classified by the CSI classifier algorithm.”)
Rahman, Cooper, and Shabara are considered analogous to the claimed invention because they deal with classifying compressed data with a neural network. It would have been obvious to one with ordinary skill in the art at the time of the effective filing date to modify Rahman and Cooper to decompress the sample data before evaluating it with the classifier because it is a well-known practice to reduce storage and transmission costs while maintaining the original data format the classifier was trained on.
Claim(s) 7-8, 17-18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rahman in view of Cooper in further view of Chang et al. (US 20210357316 A1).
Regarding Claim 7, Rahman and Cooper teach the method of claim 6. Rahman and Cooper do not teach “wherein the relationship is defined as R(q) =Average ( kl(f(xi),f(x̃i)), i = 1,2, ...n).” However, Chang teaches
wherein the relationship is defined as R(q) =Average ( kl(f(xi),f(x̃i)), i = 1,2, ...n) (Chang, paragraph 0111, “At optional block 1108, method 1100 includes, for each synthesized data item in the second dataset, (i) associating the synthesized data item with a topic in the topic model based on the combination of one or more first-type codes and one or more second-type codes in the synthesized data item, and (ii) determining a metric distance between the topic associated with the synthesized data item and the topic associated with the original data item that was used to generate the synthesized data item.” Chang, paragraph 0113, “In some embodiments, determining a metric distance between the topic associated with the synthesized data item and the topic associated with the original data item that was used to generate the synthesized data item in block 1108 includes: (i) determining a first Kullback-Leibler (KL) Divergence from the topic associated with the synthesized data item to the topic associated with the original data item that was used to generate the synthesized data item; (ii) determining a second KL Divergence from the topic associated with the original data item that was used to generate the synthesized data item to the topic associated with the synthesized data item; and (iii) setting the metric distance between the topic associated with the synthesized data item and the topic associated with the original data item that was used to generate the synthesized data item equal to an average of the first KL Divergence and the second KL Divergence.” Examiner notes that f(xi) is the original data item and f(x̃i) is the synthesized data item. Examiner further notes that the metric is calculated for each data item i in the dataset containing n data items.)
Rahman, Cooper, and Chang are considered analogous to the claimed invention because they deal with creating synthesized datasets from large amounts of data to train a ML model. It would have been obvious to one with ordinary skill in the art at the time of the effective filing date to modify Rahman and Cooper to substitute using the average KL Divergence for each data item in the dataset for using PSNR as a distance metric between sample and perturbed sample data. Doing so would be advantageous because KL Divergence better captures distributional differences in synthetic data generation and thus would better preserve model training fidelity of the dataset.
Regarding Claim 8, Rahman, Cooper, and Chang teach
The method of claim 7, wherein determining the compression quality includes selecting a compression quality that ensures that the global divergence threshold is satisfied or such that R (q) < Lkl for all q > x (Rahman, Page 4, “To construct the reference table and the catalog of corresponding models, we train and evaluate a model on data compressed with a CRF of 10 (highly accurate) along with computing the PSNR. Next, we increase the CRF by 10 (degrading video quality and improving compression) until the new model’s detection accuracy drops below the minimum threshold. At this point, we vary the CRF by 1 to fully explore the range between the last valid CRF and the first invalid CRF. Again, we evaluate each model to determine if it meets our quality-of-service standards; rejecting any models that do not. After exploring each CRF in the interval, we have a table that allows us to select a trained model given a requested CRF or PSNR value.” Examiner notes that the compression quality q is the CRF value, and x is the minimum threshold.)
Claim(s) 17 recites substantially similar limitations to claim(s) 7, and is/are therefore rejected under the same analysis.
Claim(s) 18 recites substantially similar limitations to claim(s) 8, and is/are therefore rejected under the same analysis.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
Zamir et al. (US-20230305703-A1) also discloses using a divergence threshold metric with compression rollback.
Kloepper et al. (US 20230221684 A1) also discloses finding a relationship between sample data and perturbed sample data that has been classified by a model.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Zared O. Cohen whose telephone number is (571)270-0531. The examiner can normally be reached M-F, 9am to 5pm ET.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Michelle Bechtold can be reached at (571) 431-0762. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/Z.O.C./Examiner, Art Unit 2148 /MICHELLE T BECHTOLD/Supervisory Patent Examiner, Art Unit 2148
Prosecution Insights