914240DETAILED ACTION
1. This office action is in response to the Application No. 18914240 filed on 05/20/2026. Claims 2 and 6 has been cancelled. Claims 1, 3-5, 7 and 8 are presented for examination and are currently pending.
Notice of Pre-AIA or AIA Status
2. The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Continued Examination Under 37 CFR 1.114
3. A request for continued examination under 37 CFR 1.114, including the fee set
forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this
application is eligible for continued examination under 37 CFR 1.114, and the fee set
forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action
has been withdrawn pursuant to 37 CFR 1.114. Applicant’s submission filed on
05/20/2026 has been entered.
Priority
4. The Examiner notes that the following applications 15975741 filed 05/09/2018, 16200466 filed 11/26/2018, 16455655 filed 06/27/2019, 16716098 filed 12/16/2019, 16923039 filed 05/19/2020, 16923039 filed 07/07/2020, 17180439 filed 01/21/2021, 17180439 filed 02/19/2021, 17234007 filed 04/19/2021, 18305305 filed 08/11/2021, 17404699 filed 08/17/2021, 17404699 filed 08/17/2021, 17458747 filed 08/27/2021, 17514913 filed 10/29/2021, 17727913 filed 04/25/2022, 18190044 filed 07/12/2022, 17875201 filed 07/27/2022, 18305305 filed 02/16/2023, 18190044 filed 03/24/2023, 18305305 filed 04/21/2023, 18503135 filed 11/06/2023 has no support for the following limitations:
- “deep learning architecture”.
- “homomorphically compressed and encrypted data”.
- “quantize the input data into discrete intervals”.
- “latent transformer”.
- “plurality of inference or prediction”.
Furthermore, the above listed applications appears to disclose “codewords”, but it does not appear to disclose “codewords corresponding to the discrete intervals”.
However, application 18755627 filed 06/26/2024 discloses “deep learning architecture”, “homomorphically compressed and encrypted data”, “quantize the input data into discrete intervals”, “compress the input data into a plurality of compressed codewords by dividing the input data into sections”, but does not disclose “inference” or prediction and has no support for “transformer” and “latent transformer”.
Application 18770652 filed 07/12/2024 discloses “homomorphically compressed and encrypted data” but does not disclose “deep learning architecture”, “quantize the input data into discrete intervals”. The application has no support for “codewords corresponding to the discrete intervals”. Although it discloses “codewords”, it does not appear to disclose “codewords corresponding to the discrete intervals”.
However, application 18737906 filed 06/07/2024 has support for “deep learning architecture”, “homomorphically compressed and encrypted data”, “quantize the input data into discrete intervals”, “codewords corresponding to the discrete intervals”, latent transformer architecture and produces a plurality of inferences.
As a result, for the purpose of prosecution, the effective filling date for application 18737906 filed 06/07/2024 has been used for the prior art rejection.
Response to Arguments
5. The Applicant’s arguments regarding the prior art have been considered and the Examiner is withdrawing the rejections in the previous Office Action because Applicant’s amendment necessitated new grounds of rejection presented in this Office Action.
It is noted that arguments regarding independent claims 1 and 5 have been considered but are moot because new references have now been applied to remap the independent claims.
The dependent claims 3, 4 and 7, 8 which depend directly or indirectly from independent claims 1 and 5 are not patentable because the instant claims are still obvious over the prior art of record.
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, 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.
6. Claims 1, 3-5, 7 and 8 are rejected under 35 U.S.C. 103 as being unpatentable over Horesh et al. (US20210241166) in view of Patil et al. (US20230084202 filed 09/14/2021)
Regarding claim 1, Horesh teaches a system for operating a deep learning architecture on homomorphically compressed and encrypted data (embodiments provide a computer-implemented method for generating predictions from a machine learning model trained using a reduced-size homomorphically encrypted data set [0009]; Training data set compressor 122 may encrypt the reduced bit-size representation using an encryption key and a homomorphic encryption scheme, and the encrypted, reduced bit-size representation may be output by training data set compressor 122 to machine learning model trainer 124 [0030]; the machine learning model to be trained (e.g., an equation defining a regression model, a specification of a neural network including a number of layers, input features, output features, etc., and the like) [0031]),
comprising one or more computers with executable instructions that, when executed, cause a deep learning system to (a system comprising a processor and a non-transitory computer-readable memory storing instructions that, when executed by the processor, perform an operation for adaptively reducing the bit size of features in a training data set used to train a machine learning mode [0008]; structure of the machine learning model to be trained (e.g., an equation defining a regression model, a specification of a neural network including a number of layers, input features, output features, etc., and the like) [0031]):
sense input data from a data source (receives a training data set from a training data store 140 [0030]);
quantize the input data into discrete intervals (For example, assume that an input value to execute an operation on is 169. The system can determine that the input value can be converted to the bit pattern of 1000, since the input value is between 150 and 170 [0067]; The Examiner notes the input data 169 is quantized into discrete interval between 150 and 170, and discrete intervals are between 150 and 170, between 170 and 185, between 185 and 200 [0067]. According to the instant specification: “ In a step 3002, the system quantizes the input data into discrete intervals using the data quantization subsystem ... Under this scheme, a credit score of 742 would be quantized into the interval [740, 750)” [0251]);
generate a codebook comprising codewords (Using the example illustrated in FIG. 3, if the input value is less than 150, the input value can be converted to a bit pattern of 0000. If the input value is between 150 and 170, the input value can be converted to a bit pattern of 1000. If the input value is between 170 and 185, the input value can be converted to a bit pattern of 1100. If the input value is between 185 and 200, the input value can be converted to a bit pattern of 1110 [0067]. The Examiner notes codebook comprising codewords are bit pattern of 0000, bit pattern of 1000, bit pattern of 1100, bit pattern of 1100) corresponding to the discrete intervals (The system can determine that the input value can be converted to the bit pattern of 1000, since the input value is between 150 and 170 [0067]. The Examiner notes codeword 1000 corresponds to discrete interval between 150 and 170. According to the instant specification: “In a step 3003, the system generates a codebook and assigns unique codewords to the quantized intervals ... For example, if credit scores in the 700-750 range are most common in the dataset, they might be assigned shorter codewords like ‘101’ or ‘110’” [0252]);
compress the input data into a plurality of compressed codewords by dividing the input data into sections and allocating codewords from the codebook to sections of the input data (the feature may be compressed into a four-bit representation ... the bit patterns used to represent the different paths, where the bit patterns of 0000, 1000, 1100, 1110, and 1111 [0033]);
encrypt the plurality of compressed codewords into a plurality of encrypted and compressed codewords using a homomorphic encryption algorithm (Training data set compressor 122 generally receives a training data set from a training data store 140 and compresses the training data set into a reduced bit-size representation, as discussed in further detail below. Training data set compressor 122 may encrypt the reduced bit-size representation using an encryption key and a homomorphic encryption scheme [0030]);
process the plurality of encrypted and compressed codewords through a machine learning core of the deep learning architecture (Model training system 120 generally compresses a training data set into a reduced bit-size representation, encrypts the reduced bit-size representation of the training data set, and uses the encrypted reduced bit-size representation to train a machine learning model [0029]; a specification of a neural network including a number of layers [0031]; As illustrated, application server 130 includes an application 132, which includes a machine learning model 134 trained using an encrypted reduced bit-size training data set. [0045]),
and performs homomorphic operations on the plurality of encrypted and compressed codewords without decrypting the plurality of encrypted and compressed codewords to produce a plurality of inferences based on the encrypted and compressed codewords (As discussed, because the encrypted, reduced bit-size training data set may be encrypted using a homomorphic encryption scheme, the training data set need not be decrypted prior to use in training the machine learning model or in performing other mathematical operations in respect of the training data set [0088]; The reduced bit-size prediction may be an encrypted value, as the machine learning model may be trained to generate predictions or inferences that are encrypted using a homomorphic encryption scheme based on homomorphically encrypted data [0079]); and
Horesh does not explicitly teach wherein the machine learning core comprises a latent transformer architecture, modify the data source or the input data based on the plurality of inferences.
Patil teaches wherein the machine learning core comprises a latent transformer
architecture (various types of ML algorithms, including (but not limited to): deep learning models, neural network models, ... transformers and the like [0028]),
modify the data source or the input data (the client device 122 can receive encrypted output data 120 generated by the AI algorithm orchestration component 106 [0038]; the server security component 112 can employ one or more mechanisms to ensure the inference output data is provided to the client device 122 [0039]; local memory of the client device 122 [0044]. The Examiner notes client device 122 is a data source that stores data in local memory and the data is modified with the inference output data) based on the plurality of inferences (AI algorithm orchestration component 106 that facilitates ...distributing inference output data 111 generated therefrom in an encrypted/encoded form, represented in FIG. 1 as encrypted output data 120 [0035]; and distributing inferences generated, abstract).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have modified the method of Horesh to incorporate the teachings of Patil for the benefit of securely deploying AI models and distributing inferences generated therefrom thereby preventing its usage by unauthorized entities (Patil [0021])
Regarding claim 3, Horesh and Patil teaches the system of claim 1, Horesh teaches wherein the machine learning core produces a plurality of predictions based on the plurality of encrypted and compressed codewords (As illustrated, application server 130 includes an application 132, which includes a machine learning model 134 trained using an encrypted reduced bit-size training data set [0045]; Application 132 generally receives prediction requests including encrypted data from client-side application 112. The prediction requests generally include one or more input features, encrypted using a homomorphic encryption scheme, and an identification of the feature for which a prediction is to be made based on the encrypted input features [0046])
Regarding claim 4, Horesh and Patil teaches the system of claim 1, Horesh teaches wherein the input data is quantized before allocating codewords to sections of the input data (If the input value is between 150 and 170, the input value can be converted to a bit pattern of 1000. If the input value is between 170 and 185, the input value can be converted to a bit pattern of 1100 [0067]. The Examiner notes that the input value is quantized into discrete interval between 150 and 170 before converting to bit pattern 1000 (codewords); According to the instant specification: “ In a step 3002, the system quantizes the input data into discrete intervals using the data quantization subsystem ... Under this scheme, a credit score of 742 would be quantized into the interval [740, 750)” [0251]).
Regarding claim 5, claim 5 is similar to claim 1. It is rejected in the same manner and reasoning applying.
Regarding claim 7, claim 7 is similar to claim 3. It is rejected in the same manner and reasoning applying.
Regarding claim 8, claim 8 is similar to claim 4. It is rejected in the same manner and reasoning applying.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MORIAM MOSUNMOLA GODO whose telephone number is (571)272-8670. The examiner can normally be reached Monday-Friday 8:00am-5:00pm 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, Michelle T. Bechtold can be reached on (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.
/M.G./Examiner, Art Unit 2148
Prosecution Insights