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 .
DETAILED ACTION
This office action is in response to the claims filed on 03/11/2022.
Claims 16-35 are presented for examination.
Priority
The following claimed benefit is acknowledged: the instant application, filed 03/11/2022 claims priority from foreign application GR20190100406, filed 09/20/2019.
Information Disclosure Statement
The information disclosure statements (IDS) filed 03/11/2022 is in compliance with the provisions of 37 CFR 1.97 and 1.98. Accordingly, the information disclosure statement is being considered by the examiner.
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.
Claims 16-35 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 analysis:
In the instant case, the claims are directed to a system (claims 16-21), method (claims 22-34) and non-transitory computer-readable media (claim 35). Thus, each of the claims falls within one of the four statutory categories (i.e., process, machine, manufacture, or composition of matter).
Step 2A analysis:
Based on the claims being determined to be within of the four categories (Step 1), it must be determined if the claims are directed to a judicial exception (i.e., law of nature, natural phenomenon, and abstract idea), in this case the claims fall within the judicial exception of an abstract idea. Specifically the abstract idea of “Mental Processes/Concepts performed in the human mind (including an observation, evaluation, judgment, opinion)” and mathematical concept.
The claim 16 recites.
Step 2A: prong 1 analysis:
- “initialize the ordered plurality of inducing inputs;” This is a mental process, the human mind can set the initial value or order of the inducing input , (Observation/Evaluation).
- “initialize the parameters of the multivariate Gaussian distribution” this is a mental process, the human mind can set the initial value of the parameter of the multivariate Gaussian distribution, (observation/Evaluation).
“and predict, the state of the physical system at a further time.” This is a mental process, the human mind can predict the state of the physical system, such as can tell what happen to the system at the future time based on the current status of the physical system, (Observation/Evaluation).
a) Step 2A: Prong 2 analysis:
-“ a data interface configured to receive data representing observations of a state of a physical system at a plurality of times;” These/this limitation(s) are/is recited at a high-level of generality such that it amounts to necessary data gathering. As described in MPEP 2106.05(g), limitations that amount to merely adding insignificant extra-solution activity of data gathering to a judicial exception do not amount to significantly more (than the judicial exception and cannot integrate a judicial exception into a practical application.
-“ a memory configured to store: the data” These/this limitation(s) are/is recited at a high-level of generality such that it amounts to necessary data storing. As described in MPEP 2106.05(g), limitations that amount to merely adding insignificant extra-solution activity of data storing to a judicial exception do not amount to significantly more (than the judicial exception and cannot integrate a judicial exception into a practical application.
-“and parameters of a multivariate Gaussian distribution over a set of inducing states, each inducing state having components corresponding to a Markovian Gaussian process (GP) and one or more derivatives of the Markovian GP at a respective inducing time of a plurality of inducing times, wherein the parameters comprise a mean vector and a lower block-banded Cholesky factor of a precision matrix for the multivariate Gaussian distribution”, “ using the modified parameters of the multivariate Gaussian distribution,”, “the objective function being a function of the lower block-banded Cholesky factor of the precision matrix” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (See MPEP 2106.05(h)). As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception and that it does not integrate the judicial exception into a practical application.
-“and one or more processors configured to:”, “ iteratively modify the parameters of the multivariate Gaussian distribution to increase an objective function corresponding to a variational lower bound of a marginal log-likelihood of the observations under the Markovian GP” These additional limitations are recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)).
b) Step 2B analysis:
-“ a data interface configured to receive data representing observations of a state of a physical system at a plurality of times;” These/this limitation(s) are/is recited at a high-level of generality such that it amounts to necessary data gathering. As described in MPEP 2106.05(g), limitations that amount to merely adding insignificant extra-solution activity of data gathering to a judicial exception do not amount to significantly more than the judicial exception itself.
The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory")
-“ a memory configured to store: the data” These/this limitation(s) are/is recited at a high-level of generality such that it amounts to necessary data storing. As described in MPEP 2106.05(g), limitations that amount to merely adding insignificant extra-solution activity of data storing to a judicial exception do not amount to significantly more than the judicial exception itself .
The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory")
-“and parameters of a multivariate Gaussian distribution over a set of inducing states, each inducing state having components corresponding to a Markovian Gaussian process (GP) and one or more derivatives of the Markovian GP at a respective inducing time of a plurality of inducing times, wherein the parameters comprise a mean vector and a lower block-banded Cholesky factor of a precision matrix for the multivariate Gaussian distribution”, “ using the modified parameters of the multivariate Gaussian distribution,”, “the objective function being a function of the lower block-banded Cholesky factor of the precision matrix” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the exception itself.
-“and one or more processors configured to:”, “ iteratively modify the parameters of the multivariate Gaussian distribution to increase an objective function corresponding to a variational lower bound of a marginal log-likelihood of the observations under the Markovian GP” These additional limitations are recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)).
The claim 17 recites.
a) Step 2A: Prong 2 analysis:
-“ wherein the further time is later than any of the plurality of times.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (See MPEP 2106.05(h)). As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception and that it does not integrate the judicial exception into a practical application.
b) Step 2B analysis:
-“ wherein the further time is later than any of the plurality of times.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the exception itself.
The claim 18 recites.
Step 2A: prong 1 analysis:
-“determining hyperparameters for the Markovian GP;” this is a mental process, the human mind can determinate the hyperparameter for the particular function (Markovian GP), (observation/Evaluation).
-“ and deriving one or more physical properties of the physical system from the determined hyperparameters for the Markovian GP.” This is a mental process, the human mind can deriving one or more physical properties from the determined hyperparameter, for example, the human can derive each portion of the manufacture system, associates with the determined hyperparameter values, (observation/Evaluation).
Step 2A: Prong 2 analysis and Step 2B analysis
No additional element that provides a practical application or amount to significantly more than the abstract idea.
The claim 19 recites.
a) Step 2A: Prong 2 analysis:
-“wherein the operations comprise initializing the inducing inputs sequentially and concurrently with the receiving of the data.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (See MPEP 2106.05(h)). As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception and that it does not integrate the judicial exception into a practical application.
b) Step 2B analysis:
-“wherein the operations comprise initializing the inducing inputs sequentially and concurrently with the receiving of the data.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the exception itself.
The claim 20 recites.
a) Step 2A: Prong 2 analysis:
-“ initializing the parameters of the multivariate Gaussian distribution comprises allocating a first region of the memory to store a dense matrix comprising in-band elements of the lower block-banded Cholesky factor of the precision matrix.” These/this limitation(s) are/is recited at a high-level of generality such that it amounts to necessary data storing. As described in MPEP 2106.05(g), limitations that amount to merely adding insignificant extra-solution activity of data storing to a judicial exception do not amount to significantly more (than the judicial exception and cannot integrate a judicial exception into a practical application.
b) Step 2B analysis:
-“ initializing the parameters of the multivariate Gaussian distribution comprises allocating a first region of the memory to store a dense matrix comprising in-band elements of the lower block-banded Cholesky factor of the precision matrix.” These/this limitation(s) are/is recited at a high-level of generality such that it amounts to necessary data storing. As described in MPEP 2106.05(g), limitations that amount to merely adding insignificant extra-solution activity of data storing to a judicial exception do not amount to significantly more than the judicial exception itself .
The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory").
The claim 21 recites.
a) Step 2A: Prong 2 analysis:
-“ wherein the number of inducing inputs is less than the number of observations in the plurality of observations.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (See MPEP 2106.05(h)). As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception and that it does not integrate the judicial exception into a practical application.
b) Step 2B analysis:
-“ wherein the number of inducing inputs is less than the number of observations in the plurality of observations.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the exception itself.
The claims 22 is rejected for the same reason as the claim 16, since these claims recite the same limitation.
The claim 23 is rejected for the same reason as the claim 20, since these claims recite the same limitation.
The claim 24 recites:
a) Step 2A: Prong 2 analysis:
-“ iteratively modifying the inducing inputs to increase or decrease the objective function.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (See MPEP 2106.05(h)). As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception and that it does not integrate the judicial exception into a practical application.
b) Step 2B analysis:
-“ iteratively modifying the inducing inputs to increase or decrease the objective function.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the exception itself.
The claim 25 recites:
a) Step 2A: Prong 2 analysis:
-“ receiving a data stream comprising the plurality of observations;” These/this limitation(s) are/is recited at a high-level of generality such that it amounts to necessary data gathering. As described in MPEP 2106.05(g), limitations that amount to merely adding insignificant extra-solution activity of data gathering to a judicial exception do not amount to significantly more (than the judicial exception and cannot integrate a judicial exception into a practical application.
-“ and initializing the inducing inputs sequentially and concurrently with the receiving of the data stream.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (See MPEP 2106.05(h)). As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception and that it does not integrate the judicial exception into a practical application.
b) Step 2B analysis:
-“ receiving a data stream comprising the plurality of observations;” These/this limitation(s) are/is recited at a high-level of generality such that it amounts to necessary data gathering. As described in MPEP 2106.05(g), limitations that amount to merely adding insignificant extra-solution activity of data gathering to a judicial exception do not amount to significantly more than the judicial exception itself .
The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory").
-“ and initializing the inducing inputs sequentially and concurrently with the receiving of the data stream.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the exception itself.
The claim 26 recites:
Step 2A: prong 1 analysis:
-“ initializing first inducing inputs within the first interval;” this is a mental process, the human mind can set the initial value of the input at the first time, (observation/Evaluation).
“Initializing first parameters of the multivariate Gaussian distribution corresponding to first inducing states associated with the first inducing inputs;” this is a mental process, the human mind can set the initial values of the first parameter based on the initial state of the input, (observation/Evalution).
initializing second parameters of the multivariate Gaussian distribution corresponding to second inducing states associated with the second inducing inputs; this is a mental process, the human mind can set the initial value of the parameter based on the second state of the second input, (observation/Evaluation).
a) Step 2A: Prong 2 analysis:
-“ wherein first input values associated with first observations of the plurality of observations lie within a first interval, and second input values associated with second observations of the plurality of observations lie within a second interval different from the first interval,”, “and iteratively modifying the second parameters of the multivariate Gaussian distribution to increase or decrease an objective function for the second interval.” -“iteratively modifying the first parameters of the multivariate Gaussian distribution to increase or decrease an objective function for the first interval;”, “ initializing second parameters of the multivariate Gaussian distribution corresponding to second inducing states associated with the second inducing inputs; and iteratively modifying the second parameters of the multivariate Gaussian distribution to increase or decrease an objective function for the second interval.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (See MPEP 2106.05(h)). As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception and that it does not integrate the judicial exception into a practical application.
-“ receiving the first observations;”, “receiving the second observations;” These/this limitation(s) are/is recited at a high-level of generality such that it amounts to necessary data gathering. As described in MPEP 2106.05(g), limitations that amount to merely adding insignificant extra-solution activity of data gathering to a judicial exception do not amount to significantly more (than the judicial exception and cannot integrate a judicial exception into a practical application.
b) Step 2B analysis:
--“iteratively modifying the first parameters of the multivariate Gaussian distribution to increase or decrease an objective function for the first interval;”, “ initializing second parameters of the multivariate Gaussian distribution corresponding to second inducing states associated with the second inducing inputs; and iteratively modifying the second parameters of the multivariate Gaussian distribution to increase or decrease an objective function for the second interval.”,“ wherein first input values associated with first observations of the plurality of observations lie within a first interval, and second input values associated with second observations of the plurality of observations lie within a second interval different from the first interval,”, “and iteratively modifying the second parameters of the multivariate Gaussian distribution to increase or decrease an objective function for the second interval.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the exception itself.
-“ receiving the first observations;”, “receiving the second observations;” These/this limitation(s) are/is recited at a high-level of generality such that it amounts to necessary data gathering. As described in MPEP 2106.05(g), limitations that amount to merely adding insignificant extra-solution activity of data gathering to a judicial exception do not amount to significantly more than the judicial exception itself.
The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory").
The claim 27 is rejected for the same reason as the claim 21, since these claims recite the same limitation.
The claim 28 recites:
a) Step 2A: Prong 2 analysis:
-“wherein iteratively modifying the parameters of the multivariate Gaussian distribution comprises performing a natural gradient update.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (See MPEP 2106.05(h)). As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception and that it does not integrate the judicial exception into a practical application.
b) Step 2B analysis:
-“wherein iteratively modifying the parameters of the multivariate Gaussian distribution comprises performing a natural gradient update.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the exception itself.
The claim 29 recites:
a) Step 2A: Prong 2 analysis:
-“ wherein the data is time-series data and the ordered input values correspond to times.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (See MPEP 2106.05(h)). As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception and that it does not integrate the judicial exception into a practical application.
b) Step 2B analysis:
-“ wherein the data is time-series data and the ordered input values correspond to times..” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the exception itself.
The claim 30 recites:
a) Step 2A: Prong 2 analysis:
-“ wherein each of the observations corresponds to a sample from an audio file.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (See MPEP 2106.05(h)). As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception and that it does not integrate the judicial exception into a practical application.
b) Step 2B analysis:
-“ wherein each of the observations corresponds to a sample from an audio file.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the exception itself.
The claim 31 recites:
a) Step 2A: Prong 2 analysis:
-“ wherein each of the observations corresponds to a neural activation measurement.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (See MPEP 2106.05(h)). As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception and that it does not integrate the judicial exception into a practical application.
b) Step 2B analysis:
-“ wherein each of the observations corresponds to a neural activation measurement.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the exception itself.
The claim 32 recites:
a) Step 2A: Prong 2 analysis:
-“ wherein each of the observations corresponds to a measurement of a radio frequency signal.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (See MPEP 2106.05(h)). As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception and that it does not integrate the judicial exception into a practical application.
b) Step 2B analysis:
-“ wherein each of the observations corresponds to a measurement of a radio frequency signal.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the exception itself.
The claim 33 recites:
a) Step 2A: Prong 2 analysis:
-“ wherein the Markovian GP is a component GP in a composite GP comprising a plurality of further component GPs.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (See MPEP 2106.05(h)). As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception and that it does not integrate the judicial exception into a practical application.
b) Step 2B analysis:
-“ wherein the Markovian GP is a component GP in a composite GP comprising a plurality of further component GPs.” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the exception itself.
The claim 34 recites:
a) Step 2A: Prong 2 analysis:
-“ The computer-implemented method of claim 31, wherein the composite GP is an additive GP and each of the component GPs of the composite GP represents a source underlying the plurality of observations,” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (See MPEP 2106.05(h)). As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception and that it does not integrate the judicial exception into a practical application.
-“ the method comprising training the Markovian GP and the plurality of further GPs to determine a distribution of each of the sources underlying the plurality of observations.” The additional limitation is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)).
b) Step 2B analysis:
-“ The computer-implemented method of claim 31, wherein the composite GP is an additive GP and each of the component GPs of the composite GP represents a source underlying the plurality of observations,” This/these limitation(s) is/are amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the exception itself.
-“ the method comprising training the Markovian GP and the plurality of further GPs to determine a distribution of each of the sources underlying the plurality of observations.” The additional limitation is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)).
The claim 35 is rejected for the same reason as the claim 16, since these claims recite the same limitation.
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 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.
Claims 16, 17, 21, 22, 24, 26, 27, 28, 29, 33, 35 are rejected under 35 U.S.C. 103 as being unpatentable over Marti et al. , (PUB: No. US20140156180-hereinafter, Marti) and further in view of Titsias et al. (NPL: Variational Learning of Inducing Variables in Sparse Gaussian Processes-hereinafter, Titsias) and further in view of Durrande, et al. (NPL: Banded Matrix Operators for Gaussian Markov Models in the Automatic Differentiation Era, hereinafter, Durrande).
Regarding claim 16, Marti teaches a system comprising: a data interface configured to receive data representing observations of a state of a physical system at a plurality of times (Marti, [Par. 0004], “In one aspect, a method for determining a location of a mobile device in a venue is provided. The venue includes a space accessible by a movable body carrying the mobile device and one or more constraints of movement of the movable body. A state space estimator determines candidate locations of the mobile device at a first time point based on candidate positions determined at a previous time point conditioned upon an observation of one or more environmental variables provided at the first time point. A second observation of the one or more environment variables is received at a second time point, and the state space estimator performs a propagation step to determine the candidate locations of the mobile device at the second time point based on the candidate locations at the first time point and the second observation of the one or more environmental variable”));
a memory configured to store: the data (Marti, [Par.00006], “In one aspect, a mobile device includes a storage configured to store map data associated with a venue comprising a space accessible by a movable body and one or more constraints of movement of the movable body; and a processor configured to implement a state space estimator to determine candidate locations of the mobile device at a first time point based on candidate positions determined at a previous time point conditioned upon an observation of one or more environmental variables provided at the first time point.”);
and predict, using the modified parameters, the state of the physical system at the further time (Marti, [Par.0004], “In one aspect, a method for determining a location of a mobile device in a venue is provided. The venue includes a space accessible by a movable body carrying the mobile device and one or more constraints of movement of the movable body. A state space estimator determines candidate locations of the mobile device at a first time point based on candidate positions determined at a previous time point conditioned upon an observation of one or more environmental variables provided at the first time point. A second observation of the one or more environment variables is received at a second time point, and the state space estimator performs a propagation step to determine the candidate locations of the mobile device at the second time point based on the candidate locations at the first time point and the second observation of the one or more environmental variables” Examiner’s note, the determining the location of the mobile device based on the second observation at the second time t (further time).).
However, Marti does not teach and parameters of a multivariate Gaussian distribution over a set of inducing states, each inducing state having components corresponding to a Markovian Gaussian process (GP) and one or more derivatives of the Markovian GP at a respective inducing time of a plurality of inducing times, wherein the parameters comprise a mean vector
and a lower block-banded Cholesky factor of a precision matrix for the multivariate Gaussian distribution and one or more processors configured to: initialize the ordered plurality of inducing inputs, initialize the parameters of the multivariate Gaussian distribution iteratively modify the parameters of the multivariate Gaussian distribution to increase an objective function corresponding to a variational lower bound of a marginal log-likelihood of the observations under the Markovian GP, the objective function being a function of the lower block-banded Cholesky factor of the precision matrix, The modified parameter of the multivariate Gaussian distribution
On the other hand, Titsias teaches and parameters of a multivariate Gaussian distribution over a set of inducing states (Titsias, section 1 p.567 "In this paper we introduce a variational method that jointly selects the inducing inputs and the hyperpaprameters by maximizing a lower bound to the exact marginal likelihood. The important difference between this formulation and previous methods is that here the inducing inputs are defined to be variational parameters which are selected by minimizing the Kullback-Leibler (KL) divergence between a variational GP and the true posterior GP. This allows i) to avoid overfitting and ii) to rigorously approximate the exact GP model by minimizing a distance between the sparse model and the exact one. The selection of the inducing inputs and hyperparameters is achieved either by applying continuous optimization over all unknown quantities or by using a variational EM algorithm where at the E step we greedily select the inducing inputs from the training data and at the M step we update the hyperparameters.") ,
wherein the parameters comprise a mean vector (Titsias, [Sec. 2, pages 568], “
PNG
media_image1.png
631
636
media_image1.png
Greyscale
“
and one or more processors configured to: initialize the ordered plurality of inducing inputs (Titsias, [Sec. 1, pages 568], “Our method is most closely related to the variational sparse GP method described in (Csato and Opper, 2002; Seeger, 2003) that is applied to GP classification (Seeger, 2003). The main difference between our formulation and these techniques is that we maximize a variational lower bound in order to select the inducing inputs, while these methods use variational bounds for estimating only the kernel hyperparameters.”);
initialize the parameters of the multivariate Gaussian distribution (Titsias, section 1 p.567 "In this paper we introduce a variational method that jointly selects the inducing inputs and the hyperpaprameters by maximizing a lower bound to the exact marginal likelihood. The important difference between this formulation and previous methods is that here the inducing inputs are defined to be variational parameters which are selected by minimizing the Kullback-Leibler (KL) divergence between a variational GP and the true posterior GP. This allows i) to avoid overfitting and ii) to rigorously approximate the exact GP model by minimizing a distance between the sparse model and the exact one. The selection of the inducing inputs and hyperparameters is achieved either by applying continuous optimization over all unknown quantities or by using a variational EM algorithm where at the E step we greedily select the inducing inputs from the training data and at the M step we update the hyperparameters." Examiner’s note, the hyperparameter is updated at the M step that corresponds to the initialize the parameters) ;
iteratively modify the parameters of the multivariate Gaussian distribution (Titsias, Sec.3], “
PNG
media_image2.png
752
530
media_image2.png
Greyscale
“
to increase an objective function corresponding to a variational lower bound of a marginal log-likelihood of the observations under the Markovian GP (Titsias, section 1 p.567 "In this paper we introduce a variational method that jointly selects the inducing inputs and the hyperpaprameters by maximizing a lower bound to the exact marginal likelihood. The important difference between this formulation and previous methods is that here the inducing inputs are defined to be variational parameters which are selected by minimizing the Kullback-Leibler (KL) divergence between a variational GP and the true posterior GP.”),
The modified parameter of the multivariate Gaussian distribution (Titsias, section 1 p.567 "In this paper we introduce a variational method that jointly selects the inducing inputs and the hyperparameters by maximizing a lower bound to the exact marginal likelihood. The important difference between this formulation and previous methods is that here the inducing inputs are defined to be variational parameters which are selected by minimizing the Kullback-Leibler (KL) divergence between a variational GP and the true posterior GP.”), and [Titsias, Sec.3], “
PNG
media_image2.png
752
530
media_image2.png
Greyscale
”).
Marti and Titsias are analogous in arts because they have the same field of endeavor of training data by using the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the a data interface configured to receive data representing observations of a state of a physical system at a plurality of times, as taught by Marti, to include the and parameters of a multivariate Gaussian distribution over a set of inducing states, wherein the parameters comprise a mean vector and one or more processors configured to: initialize the ordered plurality of inducing inputs, initialize the parameters of the multivariate Gaussian distribution
iteratively modify the parameters of the multivariate Gaussian distribution
to increase an objective function corresponding to a variational lower bound of a marginal log-likelihood of the observations under the Markovian GP, the modified parameter of the multivariate Gaussian distribution, as taught by Titsias. The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the subset of data prediction, (Titsias, Sec.5, , “In this section we compare the variational lower bound (VAR), the projected process approximate log likelihood (PP) and the sparse pseudo-inputs GP (SPGP) log likelihood in four real datasets. As a baseline technique, we use the subset of data (SD) method…optimizes the lower bound over the initial values of the inducing inputs, while RSPP just keep them fixed. Clearly RSPP significantly improves over the SD prediction, and VAR significantly improves over RSP…”).
However, neither Marti nor Titsias teaches each inducing state having components corresponding to a Markovian Gaussian process (GP) and one or more derivatives of the Markovian GP at a respective inducing time of a plurality of inducing times and a lower block-banded Cholesky factor of a precision matrix for the multivariate Gaussian distribution the objective function being a function of the lower block-banded Cholesky factor of the precision matrix,
On the other hand, Durrande teaches each inducing state having components corresponding to a Markovian Gaussian process (GP) and one or more derivatives of the Markovian GP at a respective inducing time of a plurality of inducing times (Durrande, Fig.1. sec. 5.2,
“
PNG
media_image3.png
330
402
media_image3.png
Greyscale
…
PNG
media_image4.png
217
756
media_image4.png
Greyscale
“
and a lower block-banded Cholesky factor of a precision matrix for the multivariate Gaussian distribution (Durrande, [Sec.2], “
PNG
media_image5.png
722
336
media_image5.png
Greyscale
“
the objective function being a function of the lower block-banded Cholesky factor of the precision matrix (Durrande, [Sec.3.3], “
PNG
media_image6.png
457
420
media_image6.png
Greyscale
“;
Marti, Titsias and Durrande are analogous in arts because they have the same field of endeavor of training data by using the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the combined teaching of Marti and Titsias of a data interface configured to receive data representing observations of a state of a physical system at a plurality of times, and parameters of a multivariate Gaussian distribution over a set of inducing states, wherein the parameters comprise a mean vector and one or more processors configured to: initialize the ordered plurality of inducing inputs, initialize the parameters of the multivariate Gaussian distribution iteratively modify the parameters of the multivariate Gaussian distribution to increase an objective function corresponding to a variational lower bound of a marginal log-likelihood of the observations under the Markovian GP, the modified parameter of the multivariate Gaussian distribution, as set forth above, to include the each inducing state having components corresponding to a Markovian Gaussian process (GP) and one or more derivatives of the Markovian GP at a respective inducing time of a plurality of inducing times and a lower block-banded Cholesky factor of a precision matrix for the multivariate Gaussian distribution the objective function being a function of the lower block-banded Cholesky factor of the precision matrix, as taught by Durrenda. The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the prediction ability, (Durrande, [Sec. 5.2], “Finally, Figure 1 (right) shows the model predictions (J = 2, l = 11) for the years from 2015 to 2020. We use the implementation based on our custom operators and we learn the kernel parameters by optimizing the marginal likelihood of the model given the data
from 1958 until 2015. This illustrates that the model can account for complex patterns even with a small bandwidth. The mean test log-likelihood of this model
is -1.75 whereas we obtain -1.56 with the reference implementation (Rasmussen and Williams, 2006, Eq. 5.19). Although, this is slightly to the advantage of the later, it means that a model with small bandwidth can have good prediction abilities, even when it is not finely tuned for the dataset at hand.”)
Regarding claim 17, Marti teaches Marti teaches the system of claim 16, wherein the further time is later than any of the plurality of times, (Marti, [Par.0004], “In one aspect, a method for determining a location of a mobile device in a venue is provided. The venue includes a space accessible by a movable body carrying the mobile device and one or more constraints of movement of the movable body. A state space estimator determines candidate locations of the mobile device at a first time point based on candidate positions determined at a previous time point conditioned upon an observation of one or more environmental variables provided at the first time point. A second observation of the one or more environment variables is received at a second time point, and the state space estimator performs a propagation step to determine the candidate locations of the mobile device at the second time point based on the candidate locations at the first time point and the second observation of the one or more environmental variables.”).
Regarding claim 21, The combined of Marti, Titsias and Durrande teaches the method of the claim 16, however, neither Marti nor Titsias teaches wherein the number of inducing inputs is less than the number of observations in the plurality of observations ,
On the other hand, Durrande teaches wherein the number of inducing inputs is less than the number of observations in the plurality of observations (Durrande, [Sec.4.3], “
PNG
media_image7.png
342
358
media_image7.png
Greyscale
”
The total number of the observation is divided into the two subsets of data based on the lower bandwidth and upper bandwidth.).
Marti, Titsias and Durrande are analogous in arts because they have the same field of endeavor of training data by using the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the receiving the observation data, as taught by Marti to include , to include the the number of inducing inputs is less than the number of observations in the plurality of observations, as taught by Durrenda. The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the prediction ability, (Durrande, [Sec. 5.2], “Finally, Figure 1 (right) shows the model predictions (J = 2, l = 11) for the years from 2015 to 2020. We use the implementation based on our custom operators and we learn the kernel parameters by optimizing the marginal likelihood of the model given the data
from 1958 until 2015. This illustrates that the model can account for complex patterns even with a small bandwidth. The mean test log-likelihood of this model
is -1.75 whereas we obtain -1.56 with the reference implementation (Rasmussen and Williams, 2006, Eq. 5.19). Although, this is slightly to the advantage of the later, it means that a model with small bandwidth can have good prediction abilities, even when it is not finely tuned for the dataset at hand.”)
Regarding claim 22, The claims 22 is rejected for the same reason as the claim 16, since these claims recite the same limitation.
Regarding claim 24, Marti teaches the method of the claim 22, but it does not teach the comprising iteratively modifying the inducing inputs to increase or decrease the objective function,
On the other hand, Titsias teaches the computer-implemented method of claim 22, comprising iteratively modifying the inducing inputs to increase or decrease the objective function (Titsias, section 1 p.567 "In this paper we introduce a variational method that jointly selects the inducing inputs and the hyperpaprameters by maximizing a lower bound to the exact marginal likelihood. The important difference between this formulation and previous methods is that here the inducing inputs are defined to be variational parameters which are selected by minimizing the Kullback-Leibler (KL) divergence between a variational GP and the true posterior GP.”), and
Titsias, Sec.3], “
PNG
media_image2.png
752
530
media_image2.png
Greyscale
“
Marti and Titsias are analogous in arts because they have the same field of endeavor of training data by using the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the inducing input, as taught by Marti, to include the iteratively modifying the inducing inputs to increase or decrease the objective function, as taught by Titsias. The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the subset of data prediction, (Titsias, Sec.5, , “In this section we compare the variational lower bound (VAR), the projected process approximate log likelihood (PP) and the sparse pseudo-inputs GP (SPGP) log likelihood in four real datasets. As a baseline technique, we use the subset of data (SD) method…optimizes the lower bound over the initial values of the inducing inputs, while RSPP just keep them fixed. Clearly RSPP significantly improves over the SD prediction, and VAR significantly improves over RSP…”).
Regarding claim 26, Marti teaches the computer-implemented method of claim 25, wherein first input values associated with first observations of the plurality of observations lie within a first interval, and second input values associated with second observations of the plurality of observations lie within a second interval different from the first interval (Marti, [Par.0004], “In one aspect, a method for determining a location of a mobile device in a venue is provided. The venue includes a space accessible by a movable body carrying the mobile device and one or more constraints of movement of the movable body. A state space estimator determines candidate locations of the mobile device at a first time point based on candidate positions determined at a previous time point conditioned upon an observation of one or more environmental variables provided at the first time point. A second observation of the one or more environment variables is received at a second time point, and the state space estimator performs a propagation step to determine the candidate locations of the mobile device at the second time point based on the candidate locations at the first time point and the second observation of the one or more environmental variables.”). ,
the method comprising: receiving the first observations; initializing first inducing inputs within the first interval (Marti, [Par.0004], “In one aspect, a method for determining a location of a mobile device in a venue is provided. The venue includes a space accessible by a movable body carrying the mobile device and one or more constraints of movement of the movable body. A state space estimator determines candidate locations of the mobile device at a first time point based on candidate positions determined at a previous time point conditioned upon an observation of one or more environmental variables provided at the first time point. A second observation of the one or more environment variables is received at a second time point, and the state space estimator performs a propagation step to determine the candidate locations of the mobile device at the second time point based on the candidate locations at the first time point and the second observation of the one or more environmental variables.”),
receiving the second observations (Marti, [Par.0004], “In one aspect, a method for determining a location of a mobile device in a venue is provided. The venue includes a space accessible by a movable body carrying the mobile device and one or more constraints of movement of the movable body. A state space estimator determines candidate locations of the mobile device at a first time point based on candidate positions determined at a previous time point conditioned upon an observation of one or more environmental variables provided at the first time point. A second observation of the one or more environment variables is received at a second time point, and the state space estimator performs a propagation step to determine the candidate locations of the mobile device at the second time point based on the candidate locations at the first time point and the second observation of the one or more environmental variables.”).;
However, Marti does not teach initializing first parameters of the multivariate Gaussian distribution corresponding to first inducing states associated with the first inducing inputs, iteratively modifying the first parameters of the multivariate Gaussian distribution to increase or decrease an objective function for the first interval, initializing second parameters of the multivariate Gaussian distribution corresponding to second inducing states associated with the second inducing inputs and iteratively modifying the second parameters of the multivariate Gaussian distribution to increase or decrease an objective function for the second interval,
On the other hand, Titsias teaches initializing first parameters of the multivariate Gaussian distribution corresponding to first inducing states associated with the first inducing inputs (Tatsis, [Sec.1, page 567], “In this paper we introduce a variational method that jointly selects the inducing inputs and the hyperparameters by maximizing a lower bound to the exact marginal likelihood. The important difference between this formulation and previous methods is that here the inducing inputs are defined to be variational parameters which are selected by minimizing the KullbackLeibler (KL) divergence between a variational GP and the true posterior GP. This allows i) to avoid overfitting and ii) to rigorously approximate the exact GP model by minimizing a distance between the sparse model and the exact one. The selection of the inducing inputs and hyperparameters is achieved either by applying continuous optimization over all unknown quantities or by using a variational EM algorithm where at the E step we greedily select the inducing inputs from the training data and at the M step we update the hyperparameters.” Examiner’s note, the selecting the initial hyperparameter for generate the Gaussian process, which is considered as the first hyperparameter.
iteratively modifying the first parameters of the multivariate Gaussian distribution to increase or decrease an objective function for the first interval (Titsias, section 1 p.567 "In this paper we introduce a variational method that jointly selects the inducing inputs and the hyperparameters by maximizing a lower bound to the exact marginal likelihood. The important difference between this formulation and previous methods is that here the inducing inputs are defined to be variational parameters which are selected by minimizing the Kullback-Leibler (KL) divergence between a variational GP and the true posterior GP.”), and
Titsias, Sec.3], “
PNG
media_image2.png
752
530
media_image2.png
Greyscale
“
initializing second parameters of the multivariate Gaussian distribution corresponding to second inducing states associated with the second inducing inputs (Tatsias, [Sec.1, page 567], “In this paper we introduce a variational method that
jointly selects the inducing inputs and the hyperparameters by maximizing a lower bound to the exact marginal likelihood. The important difference between this formulation and previous methods is that here the inducing inputs are defined to be variational parameters which are selected by minimizing the KullbackLeibler (KL) divergence between a variational GP and the true posterior GP. This allows i) to avoid overfitting and ii) to rigorously approximate the exact GP model by minimizing a distance between the sparse model and the exact one. The selection of the inducing
inputs and hyperparameters is achieved either by applying continuous optimization over all unknown quantities or by using a variational EM algorithm where at the E step we greedily select the inducing inputs from the training data and at the M step we update the hyperparameters.” Examiner’s note, selecting the inducing input at the E step and update the hyperparameter at the M step. );
and iteratively modifying the second parameters of the multivariate Gaussian distribution to increase or decrease an objective function for the second interval (Tatsis, [Sec.1, page 567], “In this paper we introduce a variational method that jointly selects the inducing inputs and the hyperparameters by maximizing a lower bound to the exact marginal likelihood. The important difference between this formulation and previous methods is that here the inducing inputs are defined to be variational parameters which are selected by minimizing the KullbackLeibler (KL) divergence between a variational GP and the true posterior GP. This allows i) to avoid overfitting and ii) to rigorously approximate the exact GP model by minimizing a distance between the sparse model and the exact one. The selection of the inducing
inputs and hyperparameters is achieved either by applying continuous optimization over all unknown quantities or by using a variational EM algorithm where at the E step we greedily select the inducing inputs from the training data and at the M step we update the hyperparameters.”).
Marti and Titsias are analogous in arts because they have the same field of endeavor of training data by using the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the first input values associated with first observations of the plurality of observations lie within a first interval, and second input values associated with second observations of the plurality of observations lie within a second interval different from the first interval, as taught by Marti, to include the initializing first parameters of the multivariate Gaussian distribution corresponding to first inducing states associated with the first inducing inputs, iteratively modifying the first parameters of the multivariate Gaussian distribution to increase or decrease an objective function for the first interval, initializing second parameters of the multivariate Gaussian distribution corresponding to second inducing states associated with the second inducing inputs and iteratively modifying the second parameters of the multivariate Gaussian distribution to increase or decrease an objective function for the second interval, as taught by Titsias. The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the subset of data prediction, (Titsias, Sec.5, , “In this section we compare the variational lower bound (VAR), the projected process approximate log likelihood (PP) and the sparse pseudo-inputs GP (SPGP) log likelihood in four real datasets. As a baseline technique, we use the subset of data (SD) method…optimizes the lower bound over the initial values of the inducing inputs, while RSPP just keep them fixed. Clearly RSPP significantly improves over the SD prediction, and VAR significantly improves over RSP…”).
Regarding claim 27, The claims 27 is rejected for the same reason as the claim 21, since these claims recite the same limitation.
Regarding claim 28, Marti teaches the claim 22, but it does not teach wherein iteratively modifying the parameters of the multivariate Gaussian distribution comprises performing a natural gradient update,
On the hand, Titsias teaches wherein iteratively modifying the parameters of the multivariate Gaussian distribution comprises performing a natural gradient update (Tatsis, [Sec. 3, pages 570], “So far we assumed that the inducing inputs are selected by applying gradient-based optimization. However, this can be difficult in high dimensional input spaces as the number of variables becomes very large. Further, the kernel function might not be differentiable with respect to the inputs. In such cases we can still apply the variational method by selecting the inducing inputs from the training inputs. An important property of this discrete optimization scheme is that FV monotonically increases when we greedily select inducing inputs and adapt the hyperparameters. Next we discuss this greedy selection method…”)
Marti and Titsias are analogous in arts because they have the same field of endeavor of training data by using the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the receiving the observation, as taught by Marti, to include the wherein iteratively modifying the parameters of the multivariate Gaussian distribution comprises performing a natural gradient update, as taught by Titsias. The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the subset of data prediction, (Titsias, Sec.5, , “In this section we compare the variational lower bound (VAR), the projected process approximate log likelihood (PP) and the sparse pseudo-inputs GP (SPGP) log likelihood in four real datasets. As a baseline technique, we use the subset of data (SD) method…optimizes the lower bound over the initial values of the inducing inputs, while RSPP just keep them fixed. Clearly RSPP significantly improves over the SD prediction, and VAR significantly improves over RSP…”).
Regarding claim 29, Marti teaches the computer-implemented method of claim 22, wherein the data is time-series data and the ordered input values correspond to times (Marti, [Par. 0004], “In one aspect, a method for determining a location of a mobile device in a venue is provided. The venue includes a space accessible by a movable body carrying the mobile device and one or more constraints of movement of the movable body. A state space estimator determines candidate locations of the mobile device at a first time point based on candidate positions determined at a previous time point conditioned upon an observation of one or more environmental variables provided at the first time point. A second observation of the one or more environment variables is received at a second time point, and the state space estimator performs a propagation step to determine the candidate locations of the mobile device at the second time point based on the candidate locations at the first time point and the second observation of the one or more environmental variable”)).
Regarding claim 33, Marti teaches the method of claim 22, but it does not teach wherein the Markovian GP is a component GP in a composite GP comprising a plurality of further component GPs,
On the other hand, Durrenda teaches wherein the Markovian GP is a component GP in a composite GP comprising a plurality of further component GPs (Durrande, Sec. 5.3, “In this section we illustrate our ability to perform inference on a GMRF with non-conjugate likelihood. To this aim, we consider the Porto dataset that gathers the GPS locations of taxi pick-ups in the city of Porto for the period July 2013 - June 2014. We use the first three weeks of the data as our training set and the following three weeks as our test set. This dataset has already been modeled successfully with GP based Cox. processes by John and Hensman (2018) but we choose a different approach here: we consider a GP based Cox process model defined on a graph representing the road network and each data point is projected onto the closest node (if it is within a 10m radius). The main advantage of this approach is that the GP covariances are using the graph distance, which are more meaningful that the Euclidian distance (think about two locations separated by the river).”
Marti, Titsias and Durrande are analogous in arts because they have the same field of endeavor of training data by using the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the receiving the observation data, as taught by Marti, to include the Markovian GP is a component GP in a composite GP comprising a plurality of further component GPs, as taught by Durrenda. The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the prediction ability, (Durrande, [Sec. 5.2], “Finally, Figure 1 (right) shows the model predictions (J = 2, l = 11) for the years from 2015 to 2020. We use the implementation based on our custom operators and we learn the kernel parameters by optimizing the marginal likelihood of the model given the data
from 1958 until 2015. This illustrates that the model can account for complex patterns even with a small bandwidth. The mean test log-likelihood of this model
is -1.75 whereas we obtain -1.56 with the reference implementation (Rasmussen and Williams, 2006, Eq. 5.19). Although, this is slightly to the advantage of the later, it means that a model with small bandwidth can have good prediction abilities, even when it is not finely tuned for the dataset at hand.”)
Regrading claim 35, the claim 35 is rejected for the same reason as the claim 16, since these claims recite the same limitation.
Claim 18 is rejected under 35 U.S.C. 103 as being unpatentable over Marti et al. , (PUB: No. US20140156180-hereinafter, Marti) and further in view of Titsias et al. (NPL: Variational Learning of Inducing Variables in Sparse Gaussian Processes-hereinafter, Titsias) and further in view of Durrande, et al. (NPL: Banded Matrix Operators for Gaussian Markov Models in the Automatic Differentiation Era, hereinafter, Durrande) and further in view of Ephraim (Patent: No. US 6202047-hereinafter, Ephraim).
Regarding claim 18, Marti teaches the system of claim 16, but it does not teach wherein the operations further comprise: determining hyperparameters for the Markovian GP, and deriving one or more physical properties of the physical system from the determined hyperparameters for the Markovian GP
On the other hand, Ephraim teaches wherein the operations further comprise: determining hyperparameters for the Markovian GP (Ephraim, [Col.5, lines 52-63], “This can significantly reduce the number of parameters that need to be estimated from the training data. For example, if a diagonal covariance matrix is assumed and estimated from training data for the cepstral vectors, then the number of parameters of the Gaussian pdf that need to be estimated from the training data can be reduced by a factor of two.”;
and deriving one or more physical properties of the physical system from the determined hyperparameters for the Markovian GP (Ephraim, [Col.1 ,lines 43-56], “The extracted features can then be processed by the speech recognizer 300 to produce the answer. This is done by statistically modeling the cepstral vectors representing speech signal vectors for a given word in the vocabulary using a Hidden Markov Model (HMM). The HMM provides a parametric representation for the probability density function (pdf) of the cepstral vectors for a given word. It assumes that cepstral vectors can emerge from several Markovian states, where each state represents a Gaussian vector source with a given mean and covariance matrix. The parameters of the HMM, which consist of initial state probabilities, state transition probabilities, mixture gains, mean vectors and covariance matrices of different states and mixture components, are estimated from training data.”). .
Marti and Ephraim are analogous in arts because they have the same field of endeavor of training data by using the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the determining hyperparameters, as taught by Marti, to include the determining hyperparameters for the Markovian GP, and deriving one or more physical properties of the physical system from the determined hyperparameters for the Markovian GP, as taught by Ephraim. The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the capability of a recognizer to perform online adaption (Ephraim, Col.5, lines 7-21, “This can significantly reduce the number of parameters that need to be estimated from the training data. For example, if a diagonal covariance matrix is assumed and estimated from training data for the cepstral vectors, then the number of parameters of the Gaussian pdf that need to be estimated from the training data can be reduced by a factor of two. Using the fixed theoretically calculated covariance of cepstral components, rather then estimating this covariance from training data, had no effect on the performance of the speech recognition system according to an embodiment of the present invention. Furthermore, since now the HMM has less data dependent parameters, it is less susceptible to the effects of noise in the input signal. In addition, the reduced number of parameters can improve the capability of a recognizer to perform on-line adaptation.”).
Claim 19 is rejected under 35 U.S.C. 103 as being unpatentable over Marti et al. , (PUB: No. US20140156180-hereinafter, Marti) and further in view of Titsias et al. (NPL: Variational Learning of Inducing Variables in Sparse Gaussian Processes-hereinafter, Titsias) and further in view of Durrande, et al. (NPL: Banded Matrix Operators for Gaussian Markov Models in the Automatic Differentiation Era, hereinafter, Durrande) and further in view of Masayuki (PUB: No. US 20040189795-hereinafter, Masayuki).
Regarding claim 19, Marti teaches the system of the claim 16, but it does not teach wherein the operations comprise initializing the inducing inputs sequentially and concurrently with the receiving of the data,
On the other hand, Masayuki teaches wherein the operations comprise initializing the inducing inputs sequentially and concurrently with the receiving of the data (Masayuki, [Par.0110], “More specifically, when the three-dimensional processing section 28 converts the reduced image data into the three-dimensional data, the three-dimensional processing section 28 sequentially receives the induced image data corresponding to the input image data from the respective image pickup devices 11a, 11b, and 11c. The three-dimensional processing section 28 prepares the three-dimensional image data while storing in the frame memory 32 the reduced image data thus received.”).
Marti and Masayuki are analogous in arts because they have the same field of endeavor of processing the input data by computer system.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the inducing the input, as taught by Marti, to include the operations comprise initializing the inducing inputs sequentially and concurrently with the receiving of the data as taught by Masayuki. The modification would have been obvious because one of the ordinary skills in art would be motivated to sequentially receiving the input (Masayuki, [Par.0110], “ More specifically, when the three-dimensional processing section 28 converts the reduced image data into the three-dimensional data, the three-dimensional processing section 28 sequentially receives the induced image data corresponding to the input image data from the respective image pickup devices 11a, 11b, and 11c. The three-dimensional processing section 28 prepares the three-dimensional image data while storing in the frame memory 32 the reduced image data thus received. Upon completion of receiving the n reduced image data that have been reduced the number to be 1/n in a lateral direction by the image data reduction section 23, the three-dimensional image data is finished up. The three-dimensional image data thus finished up is outputted to the image display apparatus 12 via the selector 19.”).
Claims 20, 23 are rejected under 35 U.S.C. 103 as being unpatentable over Marti et al. , (PUB: No. US20140156180-hereinafter, Marti) and further in view of Titsias et al. (NPL: Variational Learning of Inducing Variables in Sparse Gaussian Processes-hereinafter, Titsias) and further in view of Durrande, et al. (NPL: Banded Matrix Operators for Gaussian Markov Models in the Automatic Differentiation Era, hereinafter, Durrande) and further in view of WEI-Hao (PUB: No. US 20170104499-hereinafter, WeiHao).
Regarding claim 20, Marti teaches the system of the claim16, however, Marti does not teach wherein initializing the parameters of the multivariate Gaussian distribution comprises allocating a first region of the memory to store a dense matrix comprising in-band elements of the lower block-banded Cholesky factor of the precision matrix.
On the other hand, Titsias teaches wherein initializing the parameters of the multivariate Gaussian distribution (Titsias, section 1 p.567 "In this paper we introduce a variational method that jointly selects the inducing inputs and the hyperpaprameters by maximizing a lower bound to the exact marginal likelihood. The important difference between this formulation and previous methods is that here the inducing inputs are defined to be variational parameters which are selected by minimizing the Kullback-Leibler (KL) divergence between a variational GP and the true posterior GP. This allows i) to avoid overfitting and ii) to rigorously approximate the exact GP model by minimizing a distance between the sparse model and the exact one. The selection of the inducing inputs and hyperparameters is achieved either by applying continuous optimization over all unknown quantities or by using a variational EM algorithm where at the E step we greedily select the inducing inputs from the training data and at the M step we update the hyperparameters." Examiner’s note, the hyperparameter is updated at the M step that corresponds to the initialize the parameters) ;
Marti and Titsias are analogous in arts because they have the same field of endeavor of training data by using the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the re initializing the parameters, as taught by Marti, to include the initializing the parameters of the multivariate Gaussian distribution, as taught by Titsias. The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the subset of data prediction, (Titsias, Sec.5, , “In this section we compare the variational lower bound (VAR), the projected process approximate log likelihood (PP) and the sparse pseudo-inputs GP (SPGP) log likelihood in four real datasets. As a baseline technique, we use the subset of data (SD) method…optimizes the lower bound over the initial values of the inducing inputs, while RSPP just keep them fixed. Clearly RSPP significantly improves over the SD prediction, and VAR significantly improves over RSP…”).
However, neither Marti nor Titsias teaches comprises allocating a first region of the memory to store a dense matrix comprising in-band elements of the lower block-banded Cholesky factor of the precision matrix.
On the other hand, Durrande teaches a dense matrix comprising in-band elements of the lower block-banded Cholesky factor of the precision matrix (Durrande, [Sec.4.3], “
PNG
media_image7.png
342
358
media_image7.png
Greyscale
”
).
Marti, Titsias and Durrande are analogous in arts because they have the same field of endeavor of generating the input data.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the initializing the parameters of the multivariate Gaussian distribution, as taught by Titsias, to include the and a dense matrix comprising in-band elements of the lower block-banded Cholesky factor of the precision matrix, as tuaght by Durrande . The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the prediction ability, (Durrande, [Sec. 5.2], “Finally, Figure 1 (right) shows the model predictions (J = 2, l = 11) for the years from 2015 to 2020. We use the implementation based on our custom operators and we learn the kernel parameters by optimizing the marginal likelihood of the model given the data from 1958 until 2015. This illustrates that the model can account for complex patterns even with a small bandwidth. The mean test log-likelihood of this model is -1.75 whereas we obtain -1.56 with the reference implementation (Rasmussen and Williams, 2006, Eq. 5.19). Although, this is slightly to the advantage of the later, it means that a model with small bandwidth can have good prediction abilities, even when it is not finely tuned for the dataset at hand.”)
However, Marti, Titsias and Durrande do not teach comprises allocating a first region of the memory to store a dense matrix,
On the other hand, WEI-Hao teaches comprises allocating a first region of the memory to store a dense matrix (Wei-Hao, [Par.0028], “ Memory location 390 in FIG. 3 stores dense matrix F 396 which is given by: F=(ET.sup.−1B+D).sup.−1 (3)
and where T is an identity matrix, dense matrix F can be represented as:
F=(E*B+D).sup.−1 (4)
Dense matrix F can be pre-calculated to lighten the processing burden of the encoder, such as encoder 110.”
Marti, Titsias, Durrande and Wei-hao are analogous in arts because they have the same field of endeavor of generating the input data.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the combined teaching of Marti and Titsias of initializing the parameters of the multivariate Gaussian distribution, and a dense matrix comprising in-band elements of the lower block-banded Cholesky factor of the precision matrix, as set forth above, to include the memory to store a dense matrix c, as taught by Wei-Hao. The modification would have been obvious because one of the ordinary skills in art would be motivated to reduce the memory burden, (Wei-Hao, [Par.0028], “Dense matrix F can be pre-calculated to lighten the processing burden of the encoder, such as encoder 110. Having portion T 215 be an identity matrix allows equation (3) to be simplified to equation (4), which may reduce the processing burden. Also, having portion T 215 be an identity matrix is potentially advantageous since it need not be stored in memory since it can be easily generated. Memory location 390 has a width 392 of n-k-k1 and a height of n-k-k1. Memory location 390 is SRAM, in one embodiment.”).
Regarding claim 23, the claim 23 is rejected for the same reason as the claim 20, since these claims recite the same limitation.
Claim 25 is rejected under 35 U.S.C. 103 as being unpatentable over Marti et al. , (PUB: No. US20140156180-hereinafter, Marti) and further in view of Titsias et al. (NPL: Variational Learning of Inducing Variables in Sparse Gaussian Processes-hereinafter, Titsias) and further in view of Durrande, et al. (NPL: Banded Matrix Operators for Gaussian Markov Models in the Automatic Differentiation Era, hereinafter, Durrande) and further in view of Kim (PUB: No. US 20170011771-hereinafter, Kim).
Regarding claim 25, Marti teaches the method of The computer-implemented method of claim 22, but it does not teach receiving a data stream comprising the plurality of observations,
On the other hand, Kim teaches receiving a data stream comprising the plurality of observations [ Kim, Claim 25, “An electronic device comprising: an image sensor; a display; a transceiver; at least one processor; and a memory configured to store instructions that, when executed by the at least one processor, cause the at least one processor to control to: establish a communication group of devices including the electronic device and an external electronic device, capture images via the image sensor to generate a first live video stream, receive a second live video stream from the external electronic device via the transceiver,”)
Marti and Kim are analogous in arts because they have the same field of endeavor of generating the input data.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the receiving the observation data, as taught by Marti, to include the receiving a data stream comprising the plurality of observations, as taught by KIm. The modification would have been obvious because one of the ordinary skills in art would be motivated to receive the streaming data, [ Kim, Claim 25, “An electronic device comprising: an image sensor; a display; a transceiver; at least one processor; and a memory configured to store instructions that, when executed by the at least one processor, cause the at least one processor to control to: establish a communication group of devices including the electronic device and an external electronic device, capture images via the image sensor to generate a first live video stream, receive a second live video stream from the external electronic device via the transceiver,”).
Claims 30-32, 34 is rejected under 35 U.S.C. 103 as being unpatentable over Marti et al. , (PUB: No. US20140156180-hereinafter, Marti) and further in view of Titsias et al. (NPL: Variational Learning of Inducing Variables in Sparse Gaussian Processes-hereinafter, Titsias) and further in view of Durrande, et al. (NPL: Banded Matrix Operators for Gaussian Markov Models in the Automatic Differentiation Era, hereinafter, Durrande) and further in view of Brandon (Patent: No. US 10766137-hereinafter, Brandon).
Regrading claim 30, Marti teaches the computer-implemented method of claim 29, but it does not teach wherein each of the observations corresponds to a sample from an audio file,
On the other hand, Brandon teaches wherein each of the observations corresponds to a sample from an audio file (Brandon, [Col.6, lines 43-53], “Recorded observations can include audio signals, still images, video images sequences, electromagnetic tracking data, and textual information, depending upon the nature of a particular task. Recorded observations can additionally or alternatively include data from sensors on the robot or the target object of the task, for example data from strain gauges, torque sensors (e.g., back EMF sensors), inertial sensors (e.g., gyroscopes, accelerometers), optical sensors, radio frequency sensors, magnetic wave detectors, haptic sensors, air pressure sensors, and piezoelectric sensors..”).
Marti and Brandon are analogous in arts because they have the same field of endeavor of generating the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the receiving the observation data, as taught by Marti, to include wherein each of the observations corresponds to a sample from an audio file, as taught by Brandon. The modification would have been obvious because one of the ordinary skills in art would be motivated to detect the robot task performing, (Brandon, [Col. 3, lines 5-10], “Beneficially, the disclosed techniques can also be leveraged to recalibrate robotic systems by using a trained machine learning classifier to detect when robot task performance becomes less successful, and by using the output of the classifier to recalibrate the reward function and policy of the robotic system.”).
Regrading claim 31, Marti teaches the computer-implemented method of claim 29, but it does not teach wherein each of the observations corresponds to a neural activation measurement,
On the other hand, Brandon teaches wherein each of the observations corresponds to a neural activation measurement (Brandon, [Col.10, lines 7-24], “The machine learning classifier 234 is a module configured to learn the parameters that enable it to evaluate the level of success at a particular task that is represented by input data. The training data repository 242 stores training data that can be used to learn these parameters, and as illustrated can include one or both of external training data 241 and data received from observation system 215. External training data 241 can include examples of task performance by a human, computer simulation, or a different robotic system, and preferably is in the same format as the data that is received from the observation system 215. Data received from the observation system 215 may depict one or more robotic systems 210 performing the task. In some embodiments, the external training data 241 can include human-provided labels (e.g., A or B preferences, success level scores, or binary success/failure indications) on data received from the observation system 215.” ).
Marti and Brandon are analogous in arts because they have the same field of endeavor of generating the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the receiving the observation data, as taught by Marti, to include wherein each of the observations corresponds to a neural activation measurement, as taught by Brandon. The modification would have been obvious because one of the ordinary skills in art would be motivated to detect the robot task performing, (Brandon, [Col. 3, lines 5-10], “Beneficially, the disclosed techniques can also be leveraged to recalibrate robotic systems by using a trained machine learning classifier to detect when robot task performance becomes less successful, and by using the output of the classifier to recalibrate the reward function and policy of the robotic system.”).
Regrading claim 32, Marti teaches the computer-implemented method of claim 29, but it does not teach wherein each of the observations corresponds to a measurement of a radio frequency signal.
On the other hand, Brandon teaches wherein each of the observations corresponds to a measurement of a radio frequency signal (Brandon, [Col.6, lines 43-53], “Recorded observations can include audio signals, still images, video images sequences, electromagnetic tracking data, and textual information, depending upon the nature of a particular task. Recorded observations can additionally or alternatively include data from sensors on the robot or the target object of the task, for example data from strain gauges, torque sensors (e.g., back EMF sensors), inertial sensors (e.g., gyroscopes, accelerometers), optical sensors, radio frequency sensors, magnetic wave detectors, haptic sensors, air pressure sensors, and piezoelectric sensors..”).
Marti and Brandon are analogous in arts because they have the same field of endeavor of generating the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the receiving the observation data, as taught by Marti, to include wherein each of the observations corresponds to a measurement of a radio frequency signal, as taught by Brandon. The modification would have been obvious because one of the ordinary skills in art would be motivated to detect the robot task performing, (Brandon, [Col. 3, lines 5-10], “Beneficially, the disclosed techniques can also be leveraged to recalibrate robotic systems by using a trained machine learning classifier to detect when robot task performance becomes less successful, and by using the output of the classifier to recalibrate the reward function and policy of the robotic system.”).
Regarding claim 34, Marti as modified in view of Durrande teaches the computer-implemented method of claim 31, wherein the composite GP is an additive GP and each of the component GPs of the composite GP represents a source underlying the plurality of observations (Durrande, Sec. 5.3, “In this section we illustrate our ability to perform inference on a GMRF with non-conjugate likelihood. To this aim, we consider the Porto dataset that gathers the GPS locations of taxi pick-ups in the city of Porto for the period July 2013 - June 2014. We use the first three weeks of the data as our training set and the following three weeks as our test set. This dataset has already been modeled successfully with GP based Cox. processes by John and Hensman (2018) but we choose a different approach here: we consider a GP based Cox process model defined on a graph representing the road network and each data point is projected onto the closest node (if it is within a 10m radius). The main advantage of this approach is that the GP covariances are using the graph distance, which are more meaningful that the Euclidian distance (think about two locations separated by the river).”,
the method comprising training the Markovian GP and the plurality of further GPs to determine a distribution of each of the sources underlying the plurality of observations (Durrande, Sec. 5.3, “In this section we illustrate our ability to perform inference on a GMRF with non-conjugate likelihood. To this aim, we consider the Porto dataset that gathers the GPS locations of taxi pick-ups in the city of Porto for the period July 2013 - June 2014. We use the first three weeks of the data as our training set and the following three weeks as our test set. This dataset has already been modeled successfully with GP based Cox. processes by John and Hensman (2018) but we choose a different approach here: we consider a GP based Cox process model defined on a graph representing the road network and each data point is projected onto the closest node (if it is within a 10m radius). The main advantage of this approach is that the GP covariances are using the graph distance, which are more meaningful that the Euclidian distance (think about two locations separated by the river).”
Marti and Durrande are analogous in arts because they have the same field of endeavor of generating the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the receiving the observation data, as taught by Marti, to include wherein the composite GP is an additive GP and each of the component GPs of the composite GP represents a source underlying the plurality of observations, the method comprising training the Markovian GP and the plurality of further GPs to determine a distribution of each of the sources underlying the plurality of observations., as taught by Durrande. The modification would have been obvious because one of the ordinary skills in art would be motivated to detect the robot task performing, (Brandon, [Col. 3, lines 5-10], “Beneficially, the disclosed techniques can also be leveraged to recalibrate robotic systems by using a trained machine learning classifier to detect when robot task performance becomes less successful, and by using the output of the classifier to recalibrate the reward function and policy of the robotic system.”).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to EM N TRIEU whose telephone number is (571)272-5747. The examiner can normally be reached on Mon-Fri from 9:00-5:00.
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 on (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.
/E.T./Examiner, Art Unit 2128
/OMAR F FERNANDEZ RIVAS/Supervisory Patent Examiner, Art Unit 2128
Prosecution Insights