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 . This action is made non-final.
This action is in response to the application and claims filed August 21, 2024. Claims 1-20 are pending in the case and have been examined. Claims 1-20 are rejected.
Priority
Acknowledgment is made of applicant's claim for foreign priority based on an application filed in The Republic of Korea on August 24, 2023. The examiner acknowledges that a certified copy of Republic of Korea Patent Application No. KR10-2023-0111513 has been retrieved (on February 2, 2026, in Korean), as required by 37 CFR 1.55. The examiner notes that a translation of Republic of Korea Patent Application No. KR10-2023-0111513 does not appear to have been furnished to-date.
Although a certified copy of the foreign priority application was retrieved, a translation of said application has not yet been made of record in accordance with 37 CFR 1.55. See MPEP §§ 215 and 216. Applicant is reminded of requirements set forth in 37 CFR 1.55(g)(3)-(4) Claim for foreign priority:
“(3) An English language translation of a non-English language foreign application is not required except:
(i) When the application is involved in an interference (see § 41.202 of this chapter) or derivation (see part 42 of this chapter) proceeding;
(ii) When necessary to overcome the date of a reference relied upon by the examiner; or
(iii) When specifically required by the examiner.
(4) If an English language translation of a non-English language foreign application is required, it must be filed together with a statement that the translation of the certified copy is accurate” (emphasis added).
Since an English language translation of Republic of Korea Patent Application No. KR10-2023-0111513 has not been made of record to-date, the Examiner notes that prior art references with a filing date or a publication date prior to the instant Application’s filing date of August 21, 2024, are considered applicable prior art references.
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.
To determine if a claim is directed to patent ineligible subject matter, the Court has guided the Office to apply the Alice/Mayo test, which requires:
Step 1: Determining if the claim falls within a statutory category.
Step 2A: Determining if the claim is directed to a patent ineligible judicial exception consisting of a law of nature, a natural phenomenon, or abstract idea; and Step 2A is a two prong inquiry. MPEP 2106.04(II)(A). Under the first prong, examiners evaluate whether a law of nature, natural phenomenon, or abstract idea is set forth or described in the claim. Abstract ideas include mathematical concepts, certain methods of organizing human activity, and mental processes. MPEP 2104.04(a)(2). The second prong is an inquiry into whether the claim integrates a judicial exception into a practical application. MPEP 2106.04(d).
Step 2B: If the claim is directed to a judicial exception, determining if the claim recites limitations or elements that amount to significantly more than the judicial exception. (See MPEP 2106).
Claims 1-20 is/are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1: Claims 1-7 are directed to a method (a process), Claims 8-14 are directed to a system comprising one or more processors (a machine), and Claims 15-20 are directed to an apparatus (a manufacture). Therefore, Claims 1-20 are directed to a process, machine or manufacture or composition of matter.
Regarding claim 1
Step 2A Prong 1
Claim 1 recites the following mathematical concepts, that in each case under the broadest reasonable interpretation, covers performance of mathematical relationships, mathematical formulas or equations, and mathematical calculations but for recitation of generic computer components (e.g., “artificial neural network”) [see MPEP 2106.04(a)(2)(I)].
estimating sample scale factors of first sample parameters, the first sample parameters being part of first parameters within the artificial neural network (e.g., calculating numerical skill factor values from numerical parameter values, such as by determining maximum and/or minimum parameter values and mathematically driving a scale factor from them)
determining a prediction scale factor based on the sample scale factors (e.g., Mathematically deriving in numerical prediction scale factor from the sample scale factors using averaging, weighted averaging, maximum/minimum selection, or other mathematical relationships)
quantizing the first parameters based on the prediction scale factor (e.g., Mathematically converting or mapping numerical parameter values to quantize numerical values using the prediction scale factor, including scaling, around and, and/or clipping relationships)
Accordingly, at Step 2A, prong one, the claim recites an abstract idea.
Step 2A Prong 2
The judicial exception is not integrated into a practical application. In particular, the claim recites the additional elements of “artificial neural network” which are recited at a high-level of generality such that they amount to no more than mere instructions to apply the exception using generic computer components (See MPEP 2106.05(f)). The Examiner notes that this is used throughout the claim limitations and is rejected thusly for each claim which recites the same language.
The examiner notes that the specification describes a reduction in computational complexity resulting from estimating scale factors for only a portion of the parameters and using a prediction scale factor such that computation of scale factors for the remaining parameters may be omitted. However, claim 1 does not require that scale factor determinations for the remaining first parameters are emitted, that scale factors are determined only for the sample parameters, or that the prediction scale factory places otherwise required individual scale factor determinations. Rather, the claim broadly requires estimating sample scale factors, determining a prediction scale factor based there on, and quantizing the first parameters based on the prediction scale factor. Thus, under the broadest reasonable interpretation, the claim does not require the particular computation reducing mechanism identified in the specification as providing the asserted technological improvement.
Accordingly, at Step 2A, prong two, the additional elements individually or in combination do not integrate the judicial exception into a practical application.
Step 2B
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception. As discussed above, the additional element of a “artificial neural network” which are recited at a high-level of generality such that they amount to no more than mere instructions to apply the exception using a generic computer component (See MPEP 2106.05(f)).
Accordingly, at Step 2B, the additional element individually or in combination does not amount to significantly more than the judicial exception.
Regarding claim 2
Step 2A Prong 1
Claim 2 does not introduce any new abstract ideas, but recites the abstract idea identified in its parent claims.
Accordingly, at Step 2A, prong one, the claim recites an abstract idea.
Step 2A Prong 2
The judicial exception is not integrated into a practical application. In particular, the claim recites the additional elements of “wherein the first parameters comprise input variables that are input to a first layer and weights of the first layer within the artificial neural network” which is recited at a high-level of generality such that they amount to no more than generally linking the use of abstract idea to a particular technological environment or field of use using a generic computer component (See MPEP 2106.05(h)). In particular it is merely specifying the particular neural network data to which the mathematical operations are applied.
Accordingly, at Step 2A, prong two, the additional elements individually or in combination do not integrate the judicial exception into a practical application.
Step 2B
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception. As discussed above, the additional element of a “wherein the first parameters comprise input variables that are input to a first layer and weights of the first layer within the artificial neural network” which is recited at a high-level of generality such that they amount to no more than generally linking the use of abstract idea to a particular technological environment or field of use using a generic computer component (See MPEP 2106.05(h)). In particular it is merely describing how the data is labeled for use in the claimed process.
Accordingly, at Step 2B, the additional element individually or in combination does not amount to significantly more than the judicial exception.
Regarding claim 3
Step 2A Prong 1
Claim 3 recites the following mathematical concepts, that in each case under the broadest reasonable interpretation, covers performance of mathematical relationships, mathematical formulas or equations, and mathematical calculations but for recitation of generic computer components (e.g., “artificial neural network”) [see MPEP 2106.04(a)(2)(I)].
“wherein the determining the prediction scale factor comprises determining the prediction scale factor based on an exponential moving average of the sample scale factors” (e.g., mathematically combining current and previous numerical skill factor values using exponential weighting to calculate a moving average value)
Accordingly, at Step 2A, prong one, the claim recites an abstract idea.
Step 2A Prong 2
In accordance with Step 2A, Prong 2, the claim does not include any additional elements, and the judicial exception is not integrated into a practical application.
Step 2B
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
Regarding claim 4
Step 2A Prong 1
Claim 4 recites the following mathematical concepts, that in each case under the broadest reasonable interpretation, covers performance of mathematical relationships, mathematical formulas or equations, and mathematical calculations but for recitation of generic computer components (e.g., “artificial neural network”) [see MPEP 2106.04(a)(2)(I)].
“wherein the determining the prediction scale factor comprises determining the prediction scale factor based on an arithmetic mean of the sample scale factors” (e.g., Summing numerical sample scale factor values and dividing the sum by the number of values)
Accordingly, at Step 2A, prong one, the claim recites an abstract idea.
Step 2A Prong 2
In accordance with Step 2A, Prong 2, the claim does not include any additional elements, and the judicial exception is not integrated into a practical application.
Step 2B
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
Regarding claim 5
Step 2A Prong 1
Claim 5 recites the following mathematical concepts, that in each case under the broadest reasonable interpretation, covers performance of mathematical relationships, mathematical formulas or equations, and mathematical calculations but for recitation of generic computer components (e.g., “artificial neural network”) [see MPEP 2106.04(a)(2)(I)].
“wherein the determining the prediction scale factor comprises determining the prediction scale factor based on a maximum value or a minimum value of the sample scale factors” (e.g., Mathematically comparing numerical sample scale factor values and selecting the greatest or least numerical value)
Accordingly, at Step 2A, prong one, the claim recites an abstract idea.
Step 2A Prong 2
In accordance with Step 2A, Prong 2, the claim does not include any additional elements, and the judicial exception is not integrated into a practical application.
Step 2B
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
Regarding claim 6
Step 2A Prong 1
Claim 6 recites the following mathematical concepts, that in each case under the broadest reasonable interpretation, covers performance of mathematical relationships, mathematical formulas or equations, and mathematical calculations but for recitation of generic computer components (e.g., “artificial neural network”) [see MPEP 2106.04(a)(2)(I)].
“quantizing the first parameters through Affine quantization or scale quantization” (e.g., Mathematically mapping numerical parameter values to quantize numerical values using scaling and/or and offset values, including multiplication, rounding, and clipping operations)
Accordingly, at Step 2A, prong one, the claim recites an abstract idea.
Step 2A Prong 2
In accordance with Step 2A, Prong 2, the claim does not include any additional elements, and the judicial exception is not integrated into a practical application.
Step 2B
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
Regarding claim 7
Step 2A Prong 1
Claim 7 recites the following mathematical concepts, that in each case under the broadest reasonable interpretation, covers performance of mathematical relationships, mathematical formulas or equations, and mathematical calculations but for recitation of generic computer components (e.g., “artificial neural network”) [see MPEP 2106.04(a)(2)(I)].
“quantizing the first parameters by using the prediction scale factor instead of respective scale factors of the first parameters” (e.g., Mathematically mapping numerical parameter values using a common calculated prediction scale factor in place of parameter specific numerical scale factors)
Accordingly, at Step 2A, prong one, the claim recites an abstract idea.
Step 2A Prong 2
In accordance with Step 2A, Prong 2, the claim does not include any additional elements, and the judicial exception is not integrated into a practical application.
Step 2B
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
Regarding claims 8-13
Claims 8-13 recites a system. Each of these claims corresponds to the method steps of claims 1-6, respectively, with the addition of generic hardware components such as a processor which are insufficient to render the claims subject matter eligible for the same reasons as described above.
Regarding claim 14
Step 2A Prong 1
Claim 14 does not introduce any new abstract ideas, but recites the abstract idea identified in its parent claims.
Accordingly, at Step 2A, prong one, the claim recites an abstract idea.
Step 2A Prong 2
The judicial exception is not integrated into a practical application. In particular, the claim recites the additional elements of “wherein the at least one processor is configured to perform the quantization process of the artificial neural network in real time as initial data is input to the at least one processor” which is recited at a high-level of generality such that they amount to no more than mere instructions to apply the exception using a generic computer component (See MPEP 2106.05(f)).
Accordingly, at Step 2A, prong two, the additional elements individually or in combination do not integrate the judicial exception into a practical application.
Step 2B
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception. As discussed above, the additional element of a “wherein the at least one processor is configured to perform the quantization process of the artificial neural network in real time as initial data is input to the at least one processor” which is recited at a high-level of generality such that they amount to no more than mere instructions to apply the exception using a generic computer component (See MPEP 2106.05(f)).
Accordingly, at Step 2B, the additional element individually or in combination does not amount to significantly more than the judicial exception.
Regarding claims 15-20
Claims 15-20 recites an apparatus. Each of these claims corresponds to the method steps of claims 1-7 (Claim 17 combines claims 3 and 4), respectively, which are insufficient to render the claims subject matter eligible for the same reasons as described above.
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.
Claim(s) 1, 2, 4-9, 11-13, 15, 16, and 18-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Khailany et al . (US 20220067530 A1, referred to as Khailany), in view of Chaudhuri et al. (US 20050025076 A1, referred to as Chaudhuri).
Regarding claim 1 Khailany teaches a quantization method for an artificial neural network for interference acceleration, the quantization method([0003-0006]: Describes quantization of parameters, including weights and/or activities, of a deep neural network to enable efficient acceleration of the neural network.) comprising:
estimating sample scale factors of first sample parameters, the first sample parameters being part of first parameters within the artificial neural network ([0031-0032], and [0063-0066]: Describes computing a scale factor for each vector of parameters within a single dimension of a multidimensional neural network parameter tensor, wherein each vector contains a portion of the parameters of the tensor. The multidimensional parameter tensor corresponds to the first parameters, while each vector of parameters is a portion of those, and corresponds to the first sample parameters. The respective per-vector scale factors correspond to sample scale factors.);
determining a prediction scale factor based on the sample scale factors([0081-0083]: Describes determining a coarse grained scale factor based on the previously determined per-vector scale factors, including determining a maximum of the per-vector scale factors and using the maximum to determine the coarse grained scale factor.);
Although Khailany teaches quantizing the first parameters ([0063-0066]: Describes computing a per-vector scale factor and the reduces the precision of the parameters in each vector according to that scale factor.). It does not teach doing so based on the prediction scale factor.
Chaudhuri teaches …based on the prediction scale factor([0012-0013, and [0067-0073]: Describes determining a common scale factor based on a plurality of individual quantization scale factors and suing the common scale factor to rescale the values, which are subsequently requantized.)
Khailany, in view of Chaudhuri teaches quantizing the first parameters based on the prediction scale factor.
It would have been obvious to one of ordinary skill in the art at the time of the claimed invention to have combined the per-vector ANN quantization of Khailany, with the common scale approach of Chaudhuri. Doing so would have enabled the system to improve performance of subsequent quantized processing.
Regarding claim 2, Khailany, in view of Chaudhuri teaches the quantization method of claim 1.
Khailany teaches wherein the first parameters comprise input variables that are input to a first layer and weights of the first layer within the artificial neural network ([0023], [0028], and [0055-0056]: Describes identifying neural network parameters as weights and/or activations and shows a convolutional layer receiving an input activation tensor and a weight tensor. The input activations correspond to input variables input to a first layer, and the weight tensor corresponds to weights of the
Regarding claim 4, Khailany, in view of Chaudhuri teaches the quantization method of claim 1.
Chaudhuri further teaches wherein the determining the prediction scale factor comprises determining the prediction scale factor based on an arithmetic mean of the sample scale factors ([0068-0070], and Fig. 7: Describes determining a common scale factor based on an average of a plurality of individual quantization scale factors (sample scale factors), including determining an average quantization scale factor for the plurality of bursts and determining the common scale factor (prediction scale factor) based thereon.).
Regarding claim 5, Khailany, in view of Chaudhuri teaches the quantization method of claim 1.
Khailany further teaches wherein the determining the prediction scale factor comprises determining the prediction scale factor based on a maximum value or a minimum value of the sample scale factors ([0083]: Describes determining a coarse grained scale factor based on a maximum value of the previously determined per-vector scale factors, including taking an absolute maximum of the per-vector scale factors in determining the coarse grained scale factor.).
Regarding claim 6, Khailany, in view of Chaudhuri teaches the quantization method of claim 1.
Khailany further teaches, wherein the quantizing the first parameters comprises: quantizing the first parameters through Affine quantization or scale quantization ([0036]: Describes quantizing neural network parameters using symmetric scale only quantization, wherein a scale factor is sued to map real valued parameters to reduced precision integer values).
Regarding claim 7, Khailany, in view of Chaudhuri teaches the quantization method of claim 6.
Chaudhuri further teaches wherein the quantizing of the first parameters comprises: quantizing the first parameters by using the prediction scale factor instead of respective scale factors of the first parameters ([0066-0078], and Fig. 7: Describes replacing the respective scaling bases associated with individually quantized groups with a common scale factor for subsequent requantization, including determining a common scale factor form the individual quantization scale factors, rescaling the individually quantized values based on the common scale factor, and requantizing the recalled values.).
Regarding claims 8, 9 and 11-13 which recite a quantization system for an artificial neural network for interference acceleration, the quantization system comprising: at least one processor; and a storage medium (Khailany [0054], and [0093-0094]: Describes a system including a processor coupled to memory, to perform quantization functions by the system.) to execute the method steps of claims 1, 2, and 4-6, respectively, and are rejected for the same reasons as describe above.
Regarding claims 15, 16, and 18-20 which recite a quantization apparatus (Khailany [0061-0064]: Describes a neural network quantization apparatus including a quantization unit configured to determine scale factors and a parameter quantization unit configured to quantize neural network parameters based on the determined scale factors.) to execute the method steps of claims 1, 2, and 5-7, respectively, and are rejected for the same reasons as describe above.
Claim(s) 3, 10 and 17 is/are rejected under 35 U.S.C. 103 as being unpatentable over Khailany et al . (US 20220067530 A1, referred to as Khailany), in view of Chaudhuri et al. (US 20050025076 A1, referred to as Chaudhuri), in view of Bai et al. ("BatchQuant: Quantized-for-all Architecture Search with Robust Quantizer", referred to as Bai).
Regarding claim 3, Khailany, in view of Chaudhuri teaches the quantization method of claim 1.
Although Khailany, in view of Chaudhuri teaches the quantization method of claim 1. They do not teach, wherein the determining the prediction scale factor comprises determining the prediction scale factor based on an exponential moving average of the sample scale factors.
wherein the determining the prediction scale factor comprises determining the prediction scale factor based on an exponential moving average of the sample scale factors (Bai Section .1 Describes using an exponential moving average in a neural network quantization to generate an estimated quantization scale, wherein an EMA of previously observed quantization statistics is maintained to generate a scale estimator.)
It would have been obvious to one of ordinary skill in the art at the time of the claimed invention to have combined the quantization method of Khailany, in view of Chaudhuri, with the exponential moving average of Bai. Doing so would have enabled thee system to generate a smooth prediction of the scale factor for subsequent quantization.
Regarding claim 10 which recites a quantization system for an artificial neural network for interference acceleration, the quantization system comprising: at least one processor; and a storage medium (Khailany [0054], and [0093-0094]: Describes a system including a processor coupled to memory, to perform quantization functions by the system.) to execute the method steps of claim 3, respectively, and is rejected for the same reasons as describe above.
Regarding claims 17 which recites a quantization apparatus (Khailany [0061-0064]: Describes a neural network quantization apparatus including a quantization unit configured to determine scale factors and a parameter quantization unit configured to quantize neural network parameters based on the determined scale factors.) to execute the method steps of claims 3 and 4 and is rejected for the same reasons as describe above.
Claim(s) 14 is/are rejected under 35 U.S.C. 103 as being unpatentable over Khailany et al . (US 20220067530 A1, referred to as Khailany), in view of Chaudhuri et al. (US 20050025076 A1, referred to as Chaudhuri), in view of Deisher (US 20190042935 A1, referred to as Deisher).
Regarding claim 14, Khailany, in view of Chaudhuri teaches the quantization system of claim 8.
Although Khailany, in view of Chaudhuri teaches the quantization system of claim 8, they do not teach wherein the at least one processor is configured to perform the quantization process of the artificial neural network in real time as initial data is input to the at least one processor.
Deisher teaches wherein the at least one processor is configured to perform the quantization process of the artificial neural network in real time as initial data is input to the at least one processor ([0012], [0026-0037]: Describes performing neural network quantization dynamically in real time as input data is received, including obtaining input data, determining a dynamic range of the input data, dynamically calculating scaling factors, and quantizing/requantizing the neural network as new input data is received).
It would have been obvious to one of ordinary skill in the art at the time of the claimed invention to have combined the quantization method of Khailany, in view of Chaudhuri, with the real time quantization of Deisher. Doing so would have enabled thee system to dynamically perform quantization as input data is received and adapt quantization to the actual input dynamic range without requiring the input dynamic range to be known in advance, improving quantization accuracy across varying inputs.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. See attached PTO-892 for additional art including.
US 20210406682 A1 : sampling/subset, neural network parameter, scale factor, and quantization concepts
US 20220044109 A1 : exponential moving average smoothing
US 20190385050 A1: scale estimation form sampled/statistical information by quantization
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DONALD T RODEN whose telephone number is (571)272-6441. The examiner can normally be reached Mon-Thur 8:00-5:00 EST.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Omar Fernandez Rivas can be reached at (571) 272-2589. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/D.T.R./Examiner, Art Unit 2128
/OMAR F FERNANDEZ RIVAS/Supervisory Patent Examiner, Art Unit 2128
