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 .
Applicant’s response to the Non-final Office Action dated 08/14/2025, filed with the office on 11/12/2025, has been entered and made of record.
Status of Claims
Claims 1, 3-10, 12-19 and 21-23 are pending. Claims 1, 10 and 19 are amended. Claims 2, 11 and 20 are cancelled.
Response to Arguments
Applicant’s amendment of independent Claims 1, 10 and 19, which has altered the scope of the claims of the instant application, has necessitated the new ground(s) of rejection presented in this office action with respect to claims of the instant application. Accordingly, the amended claim set has rendered moot Applicant’s arguments with respect to the rejections of record under 35 U.S.C. 103.
Consequently, THIS ACTION IS MADE FINAL.
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.
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.
Claims 1, 3, 7-9, 10, 12, 16-19 and 21 are rejected under 35 U.S.C. 103 as being unpatentable over Cherubini et al. (US 2020/0272895 A1), in view of Said (US 2023/0262222 A1) and in further view of Liu et al. (US 2022/0172805 A1).
Regarding claim 1, Cherubini teaches, A system comprising: one or more computer processors; one or more computer readable storage media; (Cherubini, ¶0113: “The components of computer system/server 900 may include, but are not limited to, one or more processors or processing units 902, a system memory 904”) and program instructions stored on at least one of the one or more computer readable storage media for execution by (Cherubini, ¶0121: “The computer readable storage medium can be a tangible device that can retain and store instructions for use by an instruction execution device”) at least one of the one or more computer processors, (Cherubini, ¶0125: “instructions, which execute via the processor of the computer”) the stored program instructions comprising instructions to: (Cherubini, ¶0127: “portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s)”) receive image data (Cherubini, ¶0089: “sensor input signals may be typically regarded as image data”) associated with an artificial intelligence (AI) task; automatically process the image data (Cherubini, ¶0047: “perform tasks by considering examples, generally without being programmed with any task-specific rules, e.g., by automatically generating identifying characteristics from the learning material that is processed”) using a frontend that comprises an artificial neural network (ANN) (Cherubini, ¶0022: “feeding the sensor input data to input layers of a plurality of artificial neural networks”) and a vector-symbolic architecture (VSA), (Cherubini, ¶0076: “vector symbolic architecture (VSA)”) wherein processing the image data using the frontend comprises: using the VSA to define possible nested compositional structures (Cherubini, ¶0005: “Vector symbolic architectures are a class of distributed representational schemes that have shown to be able to represent and manipulate cognitive structures”) that may be depicted in the image data, using the ANN (Cherubini, ¶0075: “use information of sensor data from different layers of an artificial neural network… sensor data representing cognitive elements”) to transform the image data (Cherubini, ¶0013: “sensor data that may represent elements from the same class may be transformed”) to a hierarchy of objects depicted in the image data, (Cherubini, ¶0075: “sensor data representing cognitive elements from the same class translate into binary information”; the hierarchy of objects has been interpreted as class of elements) according to the possible nested compositional structures defined by the VSA, (Cherubini, ¶0076: “vector symbolic architecture (VSA) which is based on a set of operators on high-dimensional vectors of fixed length, i.e., the mapping vectors, representing a reduced description of a full concept”). However, Cherubini does not explicitly teach, producing probability mass functions (PMFs) for the image data; and automatically process an output of the frontend using a backend that comprises a symbolic logical reasoning engine, to solve the AI task, wherein processing the output of the frontend using the backend comprises: transforming the PMFs into vectors using a codebook, computing rule probabilities for each of a plurality of rules by using VSA operators on the vectors to implement functions embedded in each of the plurality of rules and calculating similarity between expected and actual vectors, wherein each of the plurality of rules is associated with a different set of VSA operators to apply the vectors to implement the functions embedded in a respective rule, applying one or more VSA operators to implement a selected rule from the plurality of rules on the vectors to generate a resulted vectorized representation, wherein the selected rule is selected based on the computed rule probability, and transforming the resulted vectorized representation into an output PMF using the codebook.
In an analogous field of endeavor, Said teaches, producing probability mass functions (PMFs) (Said, ¶0087: “probability mass function (PMF) defined by symbols after quantization”) for the image data; (Said, ¶0075: “for neural-based image and video compression adapts the variational autoencoder architecture, where the latent variables are quantized”) and automatically process an output of the frontend (Said, ¶0007: “determining a probability distribution function parameter for a data element of a data stream coded by a neural-based media compression technique”) using a backend that comprises (Said, ¶0007: “determining a code vector based on the probability distribution function parameter”) a symbolic logical reasoning engine, to solve the AI task, (Said, ¶0087: “probability mass function (PMF) defined by symbols after quantization”) wherein processing the output of the frontend using the backend comprises: transforming the PMFs into vectors using a codebook, (Said, ¶0087: “the code vector can be a vector with the cumulative distribution function (CDF) corresponding to the probability mass function (PMF)”) one or more VSA operators to implement a selected rule from the plurality of rules on the vectors (Said, ¶0094: “additional tasks may be performed to create vector c with the correct integer cumulative sums or binary probabilities”) to generate a resulted vectorized representation, wherein the selected rule is selected based on the computed rule probability, (Said, ¶0007: “a code vector based on the probability distribution function parameter, and entropy coding the data element using the code vector) and transforming the resulted vectorized representation (Said, ¶0170: “a discrete set of points, which can be used together with an interpolation method like cubic splines, for a precise computation of the transformation function”) into an output PMF using the codebook. (Said, ¶0080: “After entropy decoding, image synthesis neural network 426 processes the decoded data to produce output image 428”).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Cherubini using the teachings of Said to introduce transformation of probability mass functions to code vectors. A person skilled in the art would be motivated to combine the known elements as described above and achieve the predictable result of encoding objects structures into vectors. Therefore, it would have been obvious to combine the analogous arts Cherubini and Said to obtain the above-described limitations in claim 1. However, the combination of Cherubini and Said does not explicitly teach, computing rule probabilities for each of a plurality of rules by using VSA operators on the vectors to implement functions embedded in each of the plurality of rules and calculating similarity between expected and actual vectors, wherein each of the plurality of rules is associated with a different set of VSA operators to apply the vectors to implement the functions embedded in a respective rule.
In another analogous field of endeavor, Liu teaches, computing rule probabilities for each of a plurality of rules by using VSA operators on the vectors (Liu, ¶0053: “probability scores are compared to actual “serious” labels in the data for each parameter set described in operation 436”) to implement functions embedded in each of the plurality of rules and calculating similarity between expected and actual vectors, (Liu, ¶0053: “performance of the model is summarized for each set using two area under curve (AUC) metrics… These areas are used to compare performance and select the best model with the highest AUCs”) wherein each of the plurality of rules is associated with a different set of VSA operators (Liu, ¶0053: “For each hyper-parameter set, a model is trained… resulting model is used to generate an SAE probability score for each AE in the validation set”) to apply the vectors to implement the functions embedded in a respective rule. (Liu, ¶0053: “model architecture that is ultimately selected, as well as hyper-parameters for the model, are assessed based on the validation set performance”).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Cherubini in view of Said using the teachings of Liu to introduce validating different parameter sets. A person skilled in the art would be motivated to combine the known elements as described above and achieve the predictable result of selecting the most suitable set of parameters based on computed probability. Therefore, it would have been obvious to combine the analogous arts Cherubini, Said and Liu to obtain invention in claim 1.
Regarding claim 3, Cherubini in view of Said and in further view of Liu teaches, The system of claim 1, wherein: each of the objects is represented by performing a binding operation between attributes of the respective object; (Cherubini, ¶0085: “AMU including the SDM enables non-commutative binding of compositional structures which makes it possible to predict novel patterns”) and a scene comprising the objects is represented by performing a bundling operation between the object representations (Cherubini, ¶0077: “Reduced representations and cognitive models are essentially manipulated with two operators, performing “binding” and “bundling” functions”) of the objects comprised in the scene. (Cherubini, ¶0019: “using a binding operator and/or bundling operator—different hyper-vectors stored in the associative memory for deriving the cognitive query and candidate answers, each of which is obtained from the sensor input signals”).
Regarding claim 7, Cherubini in view of Said and in further view of Liu teaches, The system of claim 1, wherein the program instructions are further executable to automatically learn weights of the ANN using an additive cross-entropy loss (Cherubini, ¶0049: “Artificial neurons may have a threshold such that the signal may only be sent if the aggregate signal crosses that threshold”) that is optimized by updating trainable parameters of the ANN (Cherubini, ¶0049: “Edges connecting artificial neurons typically have a weight that adjusts as learning proceeds”) while a dictionary of the possible nested compositional structures frozen is maintained frozen. (Cherubini, ¶0057: “The term ‘candidate answer’ may denote one response pattern out of a group of response pattern”; dictionary/set of possible solutions does not affect neural network weights).
Regarding claim 8, Cherubini in view of Said and in further view of Liu teaches, The system of claim 1, wherein the AI task is an abstract visual (Cherubini, ¶0046: “The term ‘sensor input signal(s)’ may denote data directly derived from a sensor for visual data”) reasoning task. (Cherubini, ¶0045: “A series of candidate answers is tested against the cognitive query using the machine-learning system”).
Regarding claim 9, Cherubini in view of Said and in further view of Liu teaches, The system of claim 1, wherein: a combination of the frontend and the backend is differentiable; (Said, ¶0121: “this function is differentiable almost everywhere, to preserve automatic gradient back propagation”) and the program instructions are further executable to automatically perform end-to-end training (Cherubini, ¶0014: “two different time frames are used. The first time frame may be defined by the end of the training of the neural network”) of the combination of the frontend (Said, ¶0007: “determining a probability distribution function parameter for a data element of a data stream coded by a neural-based media compression technique”) and the backend. (Said, ¶0007: “determining a code vector based on the probability distribution function parameter”)
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Cherubini in view of Said and in further view of Liu using the additional teachings of Said to introduce differentiable processing steps. A person skilled in the art would be motivated to combine the known elements as described above and achieve the predictable result of preserving automatic gradient back propagation. Therefore, it would have been obvious to combine the analogous arts Cherubini, Said and Liu to obtain the invention in claim 9.
Regarding claim 10, it recites a method with steps corresponding to the elements of the system recited in claim 1. Therefore, the recited steps of method claim 10 are mapped to the proposed combination in the same manner as the corresponding elements in system claim 1. Additionally, the rationale and motivation to combine Cherubini, Said and Liu presented in rejection of claim 1, apply to this claim. Cherubini further teaches, A method comprising: receiving (Cherubini, ¶0008: “The method may comprise feeding sensor input signals to an input layer of an artificial neural network”).
Regarding claim 12, it recites a method with steps corresponding to the elements of the system recited in claim 3. Therefore, the recited steps of method claim 12 are mapped to the proposed combination in the same manner as the corresponding elements in system claim 3. Additionally, the rationale and motivation to combine Cherubini, Said and Liu presented in rejection of claim 1, apply to this claim.
Regarding claim 16, it recites a method with steps corresponding to the elements of the system recited in claim 7. Therefore, the recited steps of method claim 16 are mapped to the proposed combination in the same manner as the corresponding elements in system claim 7. Additionally, the rationale and motivation to combine Cherubini, Said and Liu presented in rejection of claim 1, apply to this claim.
Regarding claim 17, it recites a method with steps corresponding to the elements of the system recited in claim 8. Therefore, the recited steps of method claim 17 are mapped to the proposed combination in the same manner as the corresponding elements in system claim 8. Additionally, the rationale and motivation to combine Cherubini, Said and Liu presented in rejection of claim 1, apply to this claim.
Regarding claim 18, it recites a method with steps corresponding to the elements of the system recited in claim 9. Therefore, the recited steps of method claim 18 are mapped to the proposed combination in the same manner as the corresponding elements in system claim 9. Additionally, the rationale and motivation to combine Cherubini, Said and Liu presented in rejection of claim 9, apply to this claim.
Regarding claim 19, it recites a computer program product including instructions corresponding to the elements of the system recited in claim 1. Therefore, the recited instructions of the computer program product of claim 15 are mapped to the proposed combination in the same manner as the corresponding elements of the system claim 1. Additionally, the rationale and motivation to combine Cherubini, Said and Liu presented in rejection of claim 1, apply to this claim. In addition, Cherubini teaches, A computer program product comprising: one or more computer readable storage media and program instructions (Cherubini, ¶0030: “embodiments may take the form of a related computer program product, accessible from a computer-usable or computer-readable medium providing program code for use”).
Regarding claim 21, it recites a computer program product including instructions corresponding to the elements of the system recited in claim 3. Therefore, the recited instructions of the computer program product of claim 21 are mapped to the proposed combination in the same manner as the corresponding elements of the system claim 3. Additionally, the rationale and motivation to combine Cherubini, Said and Liu presented in rejection of claim 1, apply to this claim.
Claims 4, 13 and 22 are rejected under 35 U.S.C. 103 as being unpatentable over Cherubini et al. (US 2020/0272895 A1), in view of Said (US 2023/0262222 A1), in further view of Liu et al. (US 2022/0172805 A1) and still in further view of Zablotskaia et al. (US 2021/0374416 A1).
Regarding claim 4, Cherubini in view of Said and in further view of Liu teaches, The system of claim 3, wherein: a query vector (Cherubini, ¶0084: “a query (query vector)”) of the ANN resembles a bundling of vectorized object representations from a dictionary of the possible nested compositional structures; (Cherubini, ¶0094: “data sets relating to the domain of the query as well as the candidate answers. Thereby, both, the query as well as the candidate answers, are represented by hyper-vectors”). However, the combination of Cherubini, Said and Liu does not explicitly teach, and the processing of the image data by the frontend further comprises: decomposing the query vector into its constituent vectorized object representations; inferring the attributes of the objects based on the decomposed query vector; and producing the PMFs based on the inferred attributes of the objects.
In an analogous field of endeavor, Zablotskaia teaches, and the processing of the image data by the frontend further comprises: decomposing the query vector into its constituent vectorized object representations; (Zablotskaia, ¶0017: “the processor decomposes a static scene into multiple objects and represents each object by a latent vector capturing the object's unique appearance to encode visual properties”) inferring the attributes of the objects based on the decomposed query vector; (Zablotskaia, ¶0017: “decoder generates… appearance of a pixel for an object”; interpreted according to applicant’s specification ¶0060: “inferred (decoded) attributes of the detected objects”) and producing the PMFs based on the inferred attributes of the objects. (Zablotskaia, ¶0017: “for each latent vector, a broadcast decoder generates pixelwise pairs of assignment probability and appearance of a pixel for an object”).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Cherubini in view of Said and in further view of Liu using the teachings of Zablotskaia to introduce vector decomposition. A person skilled in the art would be motivated to combine the known elements as described above and achieve the predictable result of decomposing object representation vectors to identify more specific features of the object encoded by the vector. Therefore, it would have been obvious to combine the analogous arts Cherubini, Said, Liu and Zablotskaia to obtain the invention in claim 4.
Regarding claim 13, it recites a method with steps corresponding to the elements of the system recited in claim 4. Therefore, the recited steps of method claim 13 are mapped to the proposed combination in the same manner as the corresponding elements in system claim 4. Additionally, the rationale and motivation to combine Cherubini, Said, Liu and Zablotskaia presented in rejection of claim 4, apply to this claim.
Regarding claim 22, it recites a computer program product including instructions corresponding to the elements of the system recited in claim 4. Therefore, the recited instructions of the computer program product of claim 22 are mapped to the proposed combination in the same manner as the corresponding elements of the system claim 4. Additionally, the rationale and motivation to combine Cherubini, Said, Liu and Zablotskaia presented in rejection of claim 4, apply to this claim.
Claims 5, 14 and 23 are rejected under 35 U.S.C. 103 as being unpatentable over Cherubini et al. (US 2020/0272895 A1), in view of Said (US 2023/0262222 A1), in further view of Liu et al. (US 2022/0172805 A1), still in further view of Zablotskaia et al. (US 2021/0374416 A1) and yet in further view of Danihelka et al. (US 2017/0228642 A1).
Regarding claim 5, Cherubini in view of Said, in further view of Liu and still in further view of Zablotskaia teaches, The system of claim 4, wherein; transforming the PMFs into vectors comprises (Cherubini, ¶0103: “sensor input signals that represent cognitive queries and candidate answers are fed to the system”) and the program instructions are further executable to select the rule with a highest probability as a solution to the AI task. (Cherubini, ¶0094: “select the candidate answer having the highest probability of being the correct answer”). However, the combination of Cherubini, Said and Zablotskaia does not explicitly teach, transforming the PMFs into Fourier holographic reduced representations (FHRRs).
In an analogous field of endeavor, Danihelka teaches, transforming the PMFs into Fourier holographic reduced representations (FHRRs) (Danihelka, ¶0051: “representation data structure include a complex vector generated based on Holographic Reduced Representation).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Cherubini in view of Said, in further view of Liu and still in further view of Zablotskaia using the teachings of Danihelka to introduce a holographic reduced representation. A person skilled in the art would be motivated to combine the known elements as described above and achieve the predictable result of representing the probability value pairs in a fixed size vector. Therefore, it would have been obvious to combine the analogous arts Cherubini, Said, Liu, Zablotskaia and Danihelka to obtain the invention in claim 5.
Regarding claim 14, it recites a method with steps corresponding to the elements of the system recited in claim 5. Therefore, the recited steps of method claim 14 are mapped to the proposed combination in the same manner as the corresponding elements in system claim 5. Additionally, the rationale and motivation to combine Cherubini, Said, Liu, Zablotskaia and Danihelka presented in rejection of claim 5, apply to this claim.
Regarding claim 23, it recites a computer program product including instructions corresponding to the elements of the system recited in claim 5. Therefore, the recited instructions of the computer program product of claim 23 are mapped to the proposed combination in the same manner as the corresponding elements of the system claim 5. Additionally, the rationale and motivation to combine Cherubini, Said, Liu, Zablotskaia and Danihelka presented in rejection of claim 5, apply to this claim.
Claims 6 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Cherubini et al. (US 2020/0272895 A1), in view of Said (US 2023/0262222 A1), in further view Liu et al. (US 2022/0172805 A1), still in further view of Zablotskaia et al. (US 2021/0374416 A1), yet in further view of Danihelka et al. (US 2017/0228642 A1), and even in further view of Voelker et al. (US 2021/0133190 A1).
Regarding claim 6, Cherubini in view of Said, in further view of Liu and still in further view of Zablotskaia and yet in further view of Danihelka teaches, The system of claim 5, wherein the computation of the rule probability comprises: the combination of Cherubini, Said, Liu, Zablotskaia, and Danihelka does not explicitly teach, performing binding and unbinding operations.
In an analogous field of endeavor, Voelkar teaches, performing binding and unbinding operations. (Voelker, ¶0009: “subsystems that perform binding, unbinding, and cleanup with vector representations”).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Cherubini in view of Said, in further view of Liu, still in further view of Zablotskaia, and yet in further view of Danihelka using the teachings of Volker to introduce binding and unbinding operations. A person skilled in the art would be motivated to combine the known elements as described above and achieve the predictable result of retrieving the attributes of an object by unbinding the vector that represents the object of interest. Therefore, it would have been obvious to combine the analogous arts Cherubini, Said, Liu, Zablotskaia, Danihelka, and Voelker to obtain the invention in claim 6.
Regarding claim 15, it recites a method with steps corresponding to the elements of the system recited in claim 6. Therefore, the recited steps of method claim 15 are mapped to the proposed combination in the same manner as the corresponding elements in system claim 6. Additionally, the rationale and motivation to combine Cherubini, Said, Liu, Zablotskaia, Danihelka and Voelker presented in rejection of claim 6, apply to this claim.
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MEHRAZUL ISLAM whose telephone number is (571)270-0489. The examiner can normally be reached Monday-Friday: 8am-5pm.
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, Saini Amandeep can be reached on (571) 272-3382. 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.
/MEHRAZUL ISLAM/Examiner, Art Unit 2662
/AMANDEEP SAINI/Supervisory Patent Examiner, Art Unit 2662
Prosecution Insights