NON-FINAL REJECTION, FIRST DETAILED ACTION
Status of Prosecution
The present application, 18/421,231 filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
The application was filed in the Office on January 24, 2024.
Claims 1-20 are pending and all are rejected. Claims 1, 9 and 17 are independent claims.
Status of Claims
Claims 2, 10 and 18 are rejected under 35 U.S.C. § 112(b).
Claims 1-20 are rejected under 35 U.S.C. § 101 because the claimed invention is directed to an abstract idea without significantly more.
Claims 1-3, 6-11 and 14-19 are rejected under 35 USC. § 103 as being unpatentable over Yoon et al. (“Yoon”), United States Patent Application Publication 2024/0256895, published on Aug. 1, 2024 in view of Korean patent application publication Kim et al. (“Kim”), KR20240103473A published on July 4, 2024.
Claims 4, 12 and 20 are rejected under 35 USC. § 103 as being unpatentable over Yoon in view of Kim in further view of Yvinec et al. (“Yvinec”), United States Patent Application Publication 2024/0378433, published on Nov. 14, 2024.
Claims 5 and 13 are rejected under 35 USC. § 103 as being unpatentable over Yoon in view of Kim in further view of Marwah et al. (“Marwah”), United States Patent Application Publication 2024/0144075, published on May 2, 2024.
Claim Rejections -- §112(b)
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.
Claims 2, 10 and 18 are rejected under 35 U.S.C. § 112(b) as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim 2 recites in part, “aggregate the de-quantized models with the weights” Examiner queries as to whether this aggregate step is different from the parent claims step of “aggregate the dequantized models” or is a distinct step instead. The claim is rendered indefinite. For purposes of examination here, the claim will be construed as further limiting of parent claim’s aggregating step. Correction or clarification is requested. Claims 10 and 18 are similarly rejected.
Claim Rejections – § 101 Subject Matter Eligibility
Claims 1-20 are rejected under 35 U.S.C. § 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding representative claim 1, at step 1, the claim recites a system with hardware components, and therefore is an apparatus, or manufacture, which is a statutory category of invention. See MPEP § 2106.03.
At step 2A, prong one, the claim recites a computer-implemented method for multimodal response generation through a virtual agent.
The following limitations are the abstract idea of mathematical concepts. See MPEP § 2106.04(a)(2)(I):
de-quantize the quantized models by using global unlabeled data to run self-supervised learning (SSL);
aggregate the de-quantized models;
re-quantize the aggregated models based on the SSL and the global unlabeled data; Therefore, the claim recites at least one abstract idea per this part of the analysis.
At step 2A prong 2, the claim language is analyzed to determine whether it recites additional elements that integrate the judicial exception into a practical application. See MPEP § 2106.04(d).
The limitations:
obtain quantized models under different bitwidth generated by client devices;
and transmit the re-quantized models to the client devices.
are additional elements that generally links the use of the judicial exception to a particular technological environment or field of use, specifically artificial intelligence systems and obtaining and sending models. See MPEP §§ 2106.04(d), 2106.05(h).
Accordingly, these additional elements do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea. The claim is therefore directed to an abstract idea.
Next, at step 2B of the analysis, the claim is considered if it recites additional elements that amount to significantly more than the judicial exception. See MPEP § 2106.05.
As discussed above with respect to integration of the abstract idea into a practical application, the additional element do nothing more than linking the use of the judicial exception to a particular technological environment or field of use. See MPEP § 2106.05(h).
Therefore, claim 1 is ineligible.
As to dependent claims 2-8 the analysis of the parent claim is incorporated. In the step 2A, prong 2 analysis, the additional limitations are additional element that are steps under broadest reasonable interpretations, are additional elements that generally link the use of the judicial exception to a particular technological application. See MPEP § 2106.05(h). The claims are also ineligible.
Claims 9-20 are similarly rejected.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. § 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102 of this title, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
A.
Claims 1-3, 6-11 and 14-19 are rejected under 35 USC. § 103 as being unpatentable over Yoon et al. (“Yoon”), United States Patent Application Publication 2024/0256895, published on Aug. 1, 2024 in view of Korean patent application publication Kim et al. (“Kim”), KR20240103473A published on July 4, 2024.
As to Claim 1, Yoon teaches: A system comprising: one or more processors programmed to:
obtain quantized models under different bitwidth generated by client devices (Yoon: Fig. 2, [210], pars. 0045-47, different client devices have different precision and thus different bitwidths; par. 0052, in [210], the weights (i.e. corresponding to quantized models) are received by a server [110]);
de-quantize the quantized models (Yoon: par. 0054, at step [220], the weights are dequantized);
aggregate the de-quantized models (Yoon: par. 0054-56, the weights may be integrated (i.e. aggregated) to generate a singly high-precision weight);
re-quantize the aggregated models (Yoon: par. 0057, the server may requantize the integrated weights to weights precisions corresponding to the original inherent precisions),; and
transmit the re-quantized models to the client devices (Yoon: Fig. 2, par. 0058, the weights are transmitted to the client devices).
PNG
media_image1.png
798
670
media_image1.png
Greyscale
Yoon may not explicitly teach: de-quantize the quantized models by using global unlabeled data to run self-supervised learning (SSL);
re-quantize the aggregated models based on the SSL and the global unlabeled data.
Kim teaches in general concepts related to a federated learning device for chest disease detection (Kim: Abstract and Title). Specifically, Kim teaches that unlabeled data is received as input and is then transformed to allow for self-supervised learning (SSL) (Kim, par. 0010, this may be used for performing federated learning using a global model to which the learned self-supervised learning expression has been transferred).
It would have been obvious to a person having ordinary skill in the art at a time before the effective filing date of the application to have further modified the Yoon disclosures and teachings by performing the dequantization and requantization using the SSL of global unlabeled data as taught and suggested by Kim. Such a person would have been motivated to do so with a reasonable expectation of success to allow for addressing different distribution of classes or data sizes held by each client by using unlabeled SSL in a federated learning environment (Yoon: pars. 0004-07).
As to Claim 2, Yoon and Kim teach the limitations of claim 1.
Yoon further teaches: wherein the one or more processors are further programmed to:
determine weights for the quantized models based on training loss of de-quantizing the quantized models (Yoon: par. 0012, “obtaining, for each of the blocks, a first loss that is determined based on a difference between intermediate output weight data predicted from a block and quantized weight data corresponding to the block; obtaining a second loss that is determined based on a difference between final output weight data output from the dequantizer receiving the learning weight data and true weight data corresponding to the learning weight data; and training the dequantizer based on the first loss and the second loss”); and
aggregate the de-quantized models with the weights (Yoon: par. 0054-56, the weights may be integrated (i.e. aggregated) to generate a singly high-precision weight).
As to Claim 3, Yoon and Kim teach the limitations of claim 1.
Yoon and Kim as combined further teaches: wherein the quantized models under different bitwidth are trained using the SSL.
As to Claim 6, Yoon and Kim teach the limitations of claim 1.
Yoon further teaches: wherein the de-quantization converts the quantized models from the different bitwidth to a full precision bitwidth greater than the different bitwidth (Yoon: par. 0063, equation 1 allows for converting he weight value to a higher precision (i.e. a full precision bitwidth greater than the different, or less, bitwidth weight).
As to Claim 7, Yoon and Kim teach the limitations of claim 6.
Yoon further teaches: wherein the re-quantization quantizes the aggregated models from the full precision bitwidth to the different bitwidth (Yoon: par. 0058, the integrated weight is qunatize to the appropriate precision to each of the clients).
As to Claim 8, Yoon and Kim teach the limitations of claim 1.
Yoon further teaches: wherein the client devices are autonomous driving vehicles or edge devices (Yoon: a client device may be a terminal (i.e. an edge device)).
As to Claim 9, it is rejected for similar reasons as claim 1.
As to Claim 10, it is rejected for similar reasons as claim 2.
As to Claim 11, it is rejected for similar reasons as claim 3.
As to Claim 14, it is rejected for similar reasons as claim 6.
As to Claim 15, it is rejected for similar reasons as claim 7.
As to Claim 16, it is rejected for similar reasons as claim 8.
As to Claim 17, it is rejected for similar reasons as claim 1.
Yoon further teaches a computer readable medium to perform the steps recited in claim 1 (Yoon: par. 0109).
As to Claim 18, it is rejected for similar reasons as claim 2.
As to Claim 19, it is rejected for similar reasons as claim 3.
B.
Claims 4, 12 and 20 are rejected under 35 USC. § 103 as being unpatentable over Yoon et al. (“Yoon”), United States Patent Application Publication 2024/0256895, published on Aug. 1, 2024 in view of Korean patent application publication Kim et al. (“Kim”), KR20240103473A published on July 4, 2024 in further view of Yvinec et al. (“Yvinec”), United States Patent Application Publication 2024/0378433, published on Nov. 14, 2024.
As to Claim 4, Yoon and Kim teach the limitations of claim 1.
Yoon further teaches: the client devices have different computing resources(Yoon: par. 0050, “each local face recognition model (or other type of model) may be trained with a precision corresponding to a specification (or performance) of each respective client.”).
Yoon and Kim may not explicitly teach: the quantized models under different bitwidth are quantized based on non-uniform quantization.
Yvinec teaches in general concepts related to neural network quantization (Yvinec: Abstract). Specifically, Yvinec teaches that non-uniform quantization can preserve network accuracy better (Yvinec: par. 0004).
It would have been obvious to a person having ordinary skill in the art at a time before the effective filing date of the application to have further modified the Yoon-Kim disclosures and teachings by performing the quantization based on non-uniform processes as taught and suggested by Yvinec. Such a person would have been motivated to do so with a reasonable expectation of success to allow for providing a closer fit to the network weight distributions (Yvinec: pars. 0004).
As to Claim 12, it is rejected for similar reasons as claim 4.
As to Claim 20, it is rejected for similar reasons as claim 4.
C.
Claims 5 and 13 are rejected under 35 USC. § 103 as being unpatentable over Yoon et al. (“Yoon”), United States Patent Application Publication 2024/0256895, published on Aug. 1, 2024 in view of Korean patent application publication Kim et al. (“Kim”), KR20240103473A published on July 4, 2024 in further view of Marwah et al. (“Marwah”), United States Patent Application Publication 2024/0144075, published on May 2, 2024.
As to Claim 5, Yoon and Kim teach the limitations of claim 1.
Yoon and Kim may not explicitly teach: one or more memories storing the global unlabeled data with uniform distribution.
Marwah teaches in general on iterative calculation for label probability distribution (Marwah: Abstract). Specifically, Marwah teaches for anomaly detection functions, machine learning, including self-supervised techniques are used (Marwah: par. 0025). A calculation of the the labels to be used is used, particularly the entropy thereof (Marwah: par. 0055, the unlabeled data entropy may be calculated once to reduce computation costs).
It would have been obvious to a person having ordinary skill in the art at a time before the effective filing date of the application to have further modified the Yoon-Kim disclosures and teachings by storing the unlabeled data with uniform distribution per the entropy calculated as taught and suggested by Marwah. Such a person would have been motivated to do so with a reasonable expectation of success to allow for a reduction in entropy and optimizing efficiency.
As to Claim 13, it is rejected for similar reasons as claim 5.
Conclusion
Prior art not relied upon but relevant to Applicant’s disclosure is made of the record:
Yoon et al. (“Yoon”), “Bitwidth Heterogeneous Federated Learning with Progressive Weight Dequantization”, published on Sept. 14, 2023.
Subramanya et al. (“Subramanya”), United States Patent Application Publication 2025/0356176, published on Nov. 20, 2025
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JAMES T TSAI whose telephone number is (571)270-3916. The examiner can normally be reached M-F 8-5 Eastern.
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, Viker Lamardo can be reached at 571-270-5871. 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.
/JAMES T TSAI/ Primary Examiner, Art Unit 2147
Prosecution Insights