DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Drawings
The drawings were received on 09/14/2023. These drawings are acceptable.
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 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e. an abstract idea) without significantly more.
Claim 1: Does claim fall within a statutory category? Yes.
Step 2A Prong 1: Evaluate whether the claim recites a judicial exception.
target value 1. (Considered directed to Mathematical concepts – mathematical relationships (see MPEP § 2106.04(a)(2), subsection I);)
Step 2A Prong 2: Evaluate whether the claim as a whole integrates the recited judicial exception into a practical application of the exception
The preamble is deemed insufficient to transform the judicial exception to a patentable invention because the preamble generally links the use of a judicial exception to a particular technological environment or field of use, see MPEP 2106.05(h).
receiving aggregated statistics objects, (Deemed insufficient to transform the judicial exception to a patentable invention because the recitation is directed to insignificant solution activity, e.g. receiving or transmitting data over a network)
wherein the aggregated statistics objects comprise: bin frequencies F0 and F1 (Deemed insufficient to transform the judicial exception to a patentable invention because the recitation is directed to generally linking the use of a judicial exception to a particular technological environment or field of use. See 2106.05(h).)
respectively; feeding the aggregate statistics objects into the classifier, (Deemed insufficient to transform the judicial exception to a patentable invention because the recitation is directed to insignificant solution activity, e.g. receiving or transmitting data over a network)
The additional elements do not appear to be sufficient to transform the judicial exception into a practical application at Step 2A as analyzed above.
Step 2B: Evaluates whether the claim as a whole/in combination integrates the recited judicial exception into a practical application of the exception
The claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception and fail to integrate the abstract into practical application.
First, the additional limitations, as noted above, are directed to elements that generally link the use of a judicial exception to a particular technological environment or field of use.
Secondly, the additional limitations, as noted above, are directed to insignificant solution activity that the courts have deemed well-known, routine and convectional, see evidences noted below:
Receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016) (using a telephone for image transmission); OIP Techs., Inc., v. Amazon.com, Inc., 788 F.3d 1359, 1363, 115 USPQ2d 1090, 1093 (Fed. Cir. 2015) (sending messages over a network); buySAFE, Inc. v. Google, Inc., 765 F.3d 1350, 1355, 112 USPQ2d 1093, 1096 (Fed. Cir. 2014) (computer receives and sends information over a network); but see DDR Holdings, LLC v. Hotels.com, L.P., 773 F.3d 1245, 1258, 113 USPQ2d 1097, 1106 (Fed. Cir. 2014) ("Unlike the claims in Ultramercial, the claims at issue here specify how interactions with the Internet are manipulated to yield a desired result‐‐a result that overrides the routine and conventional sequence of events ordinarily triggered by the click of a hyperlink." (emphasis added));
Storing and retrieving information in memory, Versata Dev. Group, Inc. v. SAP Am., Inc., 793 F.3d 1306, 1334, 115 USPQ2d 1681, 1701 (Fed. Cir. 2015); OIP Techs., 788 F.3d at 1363, 115 USPQ2d at 1092-93;
These types of claimed elements cannot transform the judicial exception into a practical application at Step 2A or provide an inventive concept in Step 2B.
Thus, considering the additional elements individually and in combination and the claims as a whole, the additional elements do not provide significantly more than the abstract idea. This claim is not patent eligible.
Claim 2: Does claim fall within a statutory category? Yes.
Step 2A Prong 1: Evaluate whether the claim recites a judicial exception.
wherein the aggregated statistics objects further comprise bin-level covariances C0 and C1, wherein the bin-level covariances C0 and C1 are calculated for each pair of bins, conditioned on the target value being 0 or 1, respectively. (Considered directed to Mathematical concepts – mathematical relationships (see MPEP § 2106.04(a)(2), subsection I);)
Step 2A Prong 2: Evaluate whether the claim as a whole integrates the recited judicial exception into a practical application of the exception
The preamble is deemed insufficient to transform the judicial exception to a patentable invention because the preamble generally links the use of a judicial exception to a particular technological environment or field of use, see MPEP 2106.05(h).
The additional elements do not appear to be sufficient to transform the judicial exception into a practical application at Step 2A as analyzed above.
Step 2B: Evaluates whether the claim as a whole/in combination integrates the recited judicial exception into a practical application of the exception
The claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception and fail to integrate the abstract into practical application.
The additional limitations, as noted above, are directed to elements that generally link the use of a judicial exception to a particular technological environment or field of use.
These types of claimed elements cannot transform the judicial exception into a practical application at Step 2A or provide an inventive concept in Step 2B.
Thus, considering the additional elements individually and in combination and the claims as a whole, the additional elements do not provide significantly more than the abstract idea. This claim is not patent eligible.
Claim 3: Does claim fall within a statutory category? Yes.
Step 2A Prong 1: Evaluate whether the claim recites a judicial exception.
wherein the shape function is a weighted linear combination of B-splines using B-spline coefficients. (Considered directed to Mathematical concepts – mathematical relationships (see MPEP § 2106.04(a)(2), subsection I);)
Step 2A Prong 2: Evaluate whether the claim as a whole integrates the recited judicial exception into a practical application of the exception
The preamble is deemed insufficient to transform the judicial exception to a patentable invention because the preamble generally links the use of a judicial exception to a particular technological environment or field of use, see MPEP 2106.05(h).
The additional elements do not appear to be sufficient to transform the judicial exception into a practical application at Step 2A as analyzed above.
Step 2B: Evaluates whether the claim as a whole/in combination integrates the recited judicial exception into a practical application of the exception
The claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception and fail to integrate the abstract into practical application.
The additional limitations, as noted above, are directed to elements that generally link the use of a judicial exception to a particular technological environment or field of use.
These types of claimed elements cannot transform the judicial exception into a practical application at Step 2A or provide an inventive concept in Step 2B.
Thus, considering the additional elements individually and in combination and the claims as a whole, the additional elements do not provide significantly more than the abstract idea. This claim is not patent eligible.
Claim 4: Does claim fall within a statutory category? Yes.
Step 2A Prong 1: Evaluate whether the claim recites a judicial exception.
wherein a fitting objective for fitting the plurality of shape functions is to determine the B-splines coefficients that maximize the divergence. (Considered directed to Mathematical concepts – mathematical relationships (see MPEP § 2106.04(a)(2), subsection I);)
Step 2A Prong 2: Evaluate whether the claim as a whole integrates the recited judicial exception into a practical application of the exception
The preamble is deemed insufficient to transform the judicial exception to a patentable invention because the preamble generally links the use of a judicial exception to a particular technological environment or field of use, see MPEP 2106.05(h).
The additional elements do not appear to be sufficient to transform the judicial exception into a practical application at Step 2A as analyzed above.
Step 2B: Evaluates whether the claim as a whole/in combination integrates the recited judicial exception into a practical application of the exception
The claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception and fail to integrate the abstract into practical application.
The additional limitations, as noted above, are directed to elements that generally link the use of a judicial exception to a particular technological environment or field of use.
These types of claimed elements cannot transform the judicial exception into a practical application at Step 2A or provide an inventive concept in Step 2B.
Thus, considering the additional elements individually and in combination and the claims as a whole, the additional elements do not provide significantly more than the abstract idea. This claim is not patent eligible.
Claim 5: Does claim fall within a statutory category? Yes.
Step 2A Prong 1: Evaluate whether the claim recites a judicial exception.
further comprising intervening by applying constraints on the B-spline coefficients of the plurality of shape functions. (Considered directed to Mathematical concepts – mathematical relationships (see MPEP § 2106.04(a)(2), subsection I);)
Step 2A Prong 2: Evaluate whether the claim as a whole integrates the recited judicial exception into a practical application of the exception
The preamble is deemed insufficient to transform the judicial exception to a patentable invention because the preamble generally links the use of a judicial exception to a particular technological environment or field of use, see MPEP 2106.05(h).
The additional elements do not appear to be sufficient to transform the judicial exception into a practical application at Step 2A as analyzed above.
Step 2B: Evaluates whether the claim as a whole/in combination integrates the recited judicial exception into a practical application of the exception
The claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception and fail to integrate the abstract into practical application.
The additional limitations, as noted above, are directed to elements that generally link the use of a judicial exception to a particular technological environment or field of use.
These types of claimed elements cannot transform the judicial exception into a practical application at Step 2A or provide an inventive concept in Step 2B.
Thus, considering the additional elements individually and in combination and the claims as a whole, the additional elements do not provide significantly more than the abstract idea. This claim is not patent eligible.
Claim 6: Does claim fall within a statutory category? Yes.
Step 2A Prong 1: Evaluate whether the claim recites a judicial exception.
further comprising intervening by rejecting one or more predictive features. (Considered directed to a Mental Process: Making evaluations and judgements of observations for formulating observations, evaluations and judgements as claimed; see MPEP § 2106.04(a)(2), subsection III)
Step 2A Prong 2: Evaluate whether the claim as a whole integrates the recited judicial exception into a practical application of the exception
The preamble is deemed insufficient to transform the judicial exception to a patentable invention because the preamble generally links the use of a judicial exception to a particular technological environment or field of use, see MPEP 2106.05(h).
The additional elements do not appear to be sufficient to transform the judicial exception into a practical application at Step 2A as analyzed above.
Step 2B: Evaluates whether the claim as a whole/in combination integrates the recited judicial exception into a practical application of the exception
The claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception and fail to integrate the abstract into practical application.
The additional limitations, as noted above, are directed to elements that generally link the use of a judicial exception to a particular technological environment or field of use.
These types of claimed elements cannot transform the judicial exception into a practical application at Step 2A or provide an inventive concept in Step 2B.
Thus, considering the additional elements individually and in combination and the claims as a whole, the additional elements do not provide significantly more than the abstract idea. This claim is not patent eligible.
Claim 7: Does claim fall within a statutory category? Yes.
Step 2A Prong 1: Evaluate whether the claim recites a judicial exception.
… adding additional predictive features and their associated shape functions to the classifier subject to a stopping criterion based on a predetermined threshold for score separation improvement. (Considered directed to Mathematical concepts – mathematical relationships (see MPEP § 2106.04(a)(2), subsection I);)
Step 2A Prong 2: Evaluate whether the claim as a whole integrates the recited judicial exception into a practical application of the exception
The preamble is deemed insufficient to transform the judicial exception to a patentable invention because the preamble generally links the use of a judicial exception to a particular technological environment or field of use, see MPEP 2106.05(h).
further comprising iteratively adding additional predictive features and their associated shape functions to the classifier subject to a stopping criterion based on a predetermined threshold for score separation improvement. (Deemed insufficient to transform the judicial exception to a patentable invention because the recitation is directed to insignificant solution activity, e.g. Performing repetitive calculations)
The additional elements do not appear to be sufficient to transform the judicial exception into a practical application at Step 2A as analyzed above.
Step 2B: Evaluates whether the claim as a whole/in combination integrates the recited judicial exception into a practical application of the exception
The claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception and fail to integrate the abstract into practical application.
The additional limitations, as noted above, are directed to elements that generally link the use of a judicial exception to a particular technological environment or field of use.
Secondly, the additional limitations, as noted above, are directed to insignificant solution activity that the courts have deemed well-known, routine and convectional, see evidences noted below:
Performing repetitive calculations, Flook, 437 U.S. at 594, 198 USPQ2d at 199 (recomputing or readjusting alarm limit values); Bancorp Services v. Sun Life, 687 F.3d 1266, 1278, 103 USPQ2d 1425, 1433 (Fed. Cir. 2012) ("The computer required by some of Bancorp’s claims is employed only for its most basic function, the performance of repetitive calculations, and as such does not impose meaningful limits on the scope of those claims.");
These types of claimed elements cannot transform the judicial exception into a practical application at Step 2A or provide an inventive concept in Step 2B.
Thus, considering the additional elements individually and in combination and the claims as a whole, the additional elements do not provide significantly more than the abstract idea. This claim is not patent eligible.
Claim 8: Does claim fall within a statutory category? Yes.
Step 2A Prong 1: Evaluate whether the claim recites a judicial exception.
Abstract idea noted in claim 1.
Step 2A Prong 2: Evaluate whether the claim as a whole integrates the recited judicial exception into a practical application of the exception
The preamble is deemed insufficient to transform the judicial exception to a patentable invention because the preamble generally links the use of a judicial exception to a particular technological environment or field of use, see MPEP 2106.05(h).
further comprising providing a visualization of the fitted shape functions as 2-dimensional plots for classifier interpretation. (Deemed insufficient to transform the judicial exception to a patentable invention because the recitation merely include instructions to implement an abstract idea on a computer, or merely use a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f).)
The additional elements do not appear to be sufficient to transform the judicial exception into a practical application at Step 2A as analyzed above.
Step 2B: Evaluates whether the claim as a whole/in combination integrates the recited judicial exception into a practical application of the exception
The claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception and fail to integrate the abstract into practical application.
The additional limitations, as noted above, are directed to elements that generally link the use of a judicial exception to a particular technological environment or field of use and merely invoke the use of computer technology as a tool for applying the judicial exception.
These types of claimed elements cannot transform the judicial exception into a practical application at Step 2A or provide an inventive concept in Step 2B.
Thus, considering the additional elements individually and in combination and the claims as a whole, the additional elements do not provide significantly more than the abstract idea. This claim is not patent eligible.
Claim 9: Does claim fall within a statutory category? Yes.
Step 2A Prong 1: Evaluate whether the claim recites a judicial exception.
Abstract idea noted in claim 1.
Step 2A Prong 2: Evaluate whether the claim as a whole integrates the recited judicial exception into a practical application of the exception
The preamble is deemed insufficient to transform the judicial exception to a patentable invention because the preamble generally links the use of a judicial exception to a particular technological environment or field of use, see MPEP 2106.05(h).
wherein the aggregated statistic objects are generated by segmented training data, wherein the training data is segmented base at least in part on one or more heterogeneous behavior. (Deemed insufficient to transform the judicial exception to a patentable invention because the recitation is directed to generally linking the use of a judicial exception to a particular technological environment or field of use. See 2106.05(h).)
The additional elements do not appear to be sufficient to transform the judicial exception into a practical application at Step 2A as analyzed above.
Step 2B: Evaluates whether the claim as a whole/in combination integrates the recited judicial exception into a practical application of the exception
The claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception and fail to integrate the abstract into practical application.
The additional limitations, as noted above, are directed to elements that generally link the use of a judicial exception to a particular technological environment or field of use.
These types of claimed elements cannot transform the judicial exception into a practical application at Step 2A or provide an inventive concept in Step 2B.
Thus, considering the additional elements individually and in combination and the claims as a whole, the additional elements do not provide significantly more than the abstract idea. This claim is not patent eligible.
Regarding claims 10-13, the claim limitations are similar to those in claims 1-4 and are rejected under the same rationale.
Regarding claim 14, the claim limitations are similar to those in claim 7 and is rejected under the same rationale.
Regarding claims 15-18, the claim limitations are similar to those in claims 1-4 and are rejected under the same rationale.
Regarding claims 19-20, the claim limitations are similar to those in claims 7-8 and are rejected under the same rationale.
As shown above, claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed a judicial exception and does not recite, when claim elements are examined individually and as a whole, elements that the courts have identified as "significantly more” than the recited judicial exception. The claims are therefore directed to an abstract idea.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-2, 9, 11-12, and 15-16 are rejected under 35 U.S.C. 103 as being unpatentable over Sarferaz (US 20210192376, hereinafter ‘Sar’) in view of He et al. (US 20160247045, hereinafter ‘He’)
Regarding independent claim 1, Sar teaches a computer-implemented method for generating a classifier, comprising: (in [0027] The above-noted problems can be compounded by legal requirements that actions taken on the basis of machine learning models be explainable [a computer-implemented method for generating a classifier a learning model making decisions], such as to an individual who was affected by such action. For example, if a machine learning algorithm is used to automatically approve or deny credit, a company may be legally required to explain to the individual why they were denied credit. As described above, typical machine learning applications do not easily provide this functionality, either for end users within a company or in a way that may allow automated or self-service by the affected individual (e.g., an application can automatically display the underlying factors used by a machine learning model in reaching a decision). )
receiving aggregated statistics objects, wherein the aggregated statistics objects comprise: bin frequencies F0 and F1, wherein the bin frequencies F0 and F1 are calculated for each of a plurality of predictive features, conditioned on a target value being 0 or 1, respectively; feeding the aggregate statistics objects into the classifier, (in [0051] FIG. 3 illustrates a plot 300 (e.g., a matrix) of mutual information for ten features. Each square 310 [receiving aggregated statistics objects] represents the mutual information, or correlation or dependence, for a pair of different features [wherein the aggregated statistics objects comprise: bin frequencies F0 and F1, wherein the bin frequencies F0 and F1 are calculated for each of a plurality of predictive features,]. For example, square 310a reflects the dependence between feature 3 and feature 4. The squares 310 can be associated with discrete numerical values indicating any dependence between the variables, or the values can be binned, including to provide a heat map of dependencies... [0056] When used to evaluate a first feature with respect to a specified (target) second feature, supervised correlation can be used: scorr(X,Y)=corr(ψ.sub.X,ψ.sub.Y), where scorr is Pearson's correlation and ψ.sub.X=logit({circumflex over (P)}(Y|X))−logit({circumflex over (P)}(Y)) (binary classification) [conditioned on a target value being 0 or 1, respectively; feeding the aggregate statistics objects into the classifier,])
While Ser discloses the use of binary class bins where the features correlate to each class using statical measurements. Ser does not disclose how the margins, hyperplane or hyperspace is define for separating class features as claimed.
He teaches how the margins, hyperplane or hyperspace is define for separating class features as claimed:
wherein the classifier generates a score calculated as a sum of a plurality of flexible nonlinear shape functions applied to the plurality of predictive features, respectively; and training the classifier by fitting the plurality of shape functions to maximize a divergence for score separation between target value 0 and target value 1. (in [0021] Support vector machines (SVMs) are supervised machine learning tools useful in classification and regression analysis. In their original forms, SVMs are designed for binary classification. If non-binary classification is required, a common approach is to decompose the problem into a series of binary classification problem. Given a set of training samples, which are either labelled positive “+” or negative “−” (or whatever other binary notation is desired for identifying one class versus another class) [… separation between target value 0 and target value 1], a SVM training algorithm [training the classifier] intends to find a hyperplane that separates the two classes of training samples with the maximum margin [and training the classifier by fitting the plurality of shape functions to maximize a divergence for score separation between target value 0 and target value 1]. In one example, the maximum margin is found when the sum of the distance from the hyperplane [wherein the classifier generates a score calculated as a sum of a plurality of flexible nonlinear shape functions applied to the plurality of predictive features, respectively] to the nearest positive sample and the distance from the hyperplane to the nearest negative sample is maximized.)
He and Sar are analogous art because both involve developing information retrieval and processing techniques using machine learning systems and algorithms.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the prior art for retrieving and processing information based on support vector machines used in binary classification task, as disclosed by He with the method of developing information retrieval and processing techniques for complex decision systems used across a variety of applications in infrastructures, as disclosed by Sar.
One of ordinary skill in the arts would have been motivated to combine the methods disclosed by He and Sar, as noted above. Doing so allows for training a classifier based on optimization expression that attempts to balance the maximization of the distance between the hyperplanes with the minimization of the number of samples between the hyperplanes and for selecting an optimal pair of candidate hyperplanes, (He, Abstract & 0144).
Regarding claim 2, the rejection of claim 1 is incorporated and Sar in combination with He teaches the method of claim 1, wherein the aggregated statistics objects further comprise bin-level covariances C0 and C1, wherein the bin-level covariances C0 and C1 are calculated for each pair of bins, conditioned on the target value being 0 or 1, respectively. (in [0209] FIG. 27 illustrates a plot 2700 (e.g., a matrix) of mutual information for ten features. Each square 2710 represents the mutual information, or correlation [wherein the aggregated statistics objects further comprise bin-level covariances C0 and C1, wherein the bin-level covariances C0 and C1 are calculated for each pair of bins, conditioned on the target value being 0 or 1, respectively] or dependence, for a pair of different features. For example, square 2710a reflects the dependence between feature 3 and feature 4. The squares 2710 can be associated with discrete numerical values indicating any dependence between the variables, or the values can be binned [wherein the bin-level covariances C0 and C1 are calculated for each pair of bins, conditioned on the target value being 0 or 1, respectively into the same class], including to provide a heat map of dependencies.)
Regarding claim 9, the rejection of claim 1 is incorporated and Sar in combination with He teaches the method of claim 1, wherein the aggregated statistic objects are generated by segmented training data, wherein the training data is segmented base at least in part on one or more heterogeneous behavior.(in [0077] The machine learning framework 360 can include an inference manager 386. The interference manager 386 can allow a user to configure criteria for different machine learning model segments, which can represent segments of a data set [wherein the aggregated statistic objects are generated by segmented training data, wherein the training data is segmented base at least in part on one or more heterogeneous behavior] (or input criteria, such as properties or attributes that might be associated with a data set used with machine learning model). A configuration user interface 388 (also shown as the configuration user interface 319 of the client system 318) can allow a user (e.g., a key user associated with a client 316 or a client 318) to define segmentation criteria, such as using filters 390. The filters 390 can be used to define model segment criteria, where suitable model segments can be configured and trained by a model trainer component 392 [wherein the training data is segmented base at least in part on one or more heterogeneous behavior]. And in [0140] One or more filters 1350 can be defined for the machine learning scenario 1300. The filters 1350 can be used to define what machine learning model segments are created, what machine learning model segments are made available, and criteria that can be used to determine what machine learning model segment will be used to satisfy a particular inference request [wherein the training data is segmented base at least in part on one or more heterogeneous behavior]. [0141] FIG. 13 illustrates that filters 1350 can have particular types or categories, and particular values for a given type or category [wherein the training data is segmented base at least in part on one or more heterogeneous behavior]. In particular, the machine learning scenario 1300 is shown as providing filters for a region type 1354, where possible values 1356 for the region type include all regions, all of North America, all of Europe, values by country (e.g., Germany, United States), or values by state (e.g., Alaska, Nevada). Although a single filter type is shown, a given machine learning scenario 1300 can include multiple filter types. In the example of network traffic analysis, additional filters 1350 could include time (e.g., traffic during a particular time of a day), a time period (e.g., data within the last week), or traffic type (e.g., media streaming). When multiple filter categories are used, model segments can be created for individual values of individual filters (or particular values selected by a user) or for combinations of filter values (e.g., streaming traffic in North America), where the combinations can optionally be those explicitly specified by a user (particularly in the case where multiple filter types and/or multiple values for a given type exist, which can vastly increase the number of model segments).)
Regarding claims 10-11, the claim limitations are similar with those in claims 1-2 and are thus rejected under a similar rationale.
Regarding claims 15-16, the claim limitations are similar with those in claims 1-2 and are thus rejected under a similar rationale.
Claims 3-4, 6, 12-13, and 17-18 are rejected under 35 U.S.C. 103 as being unpatentable over Sarferaz (US 20210192376, hereinafter ‘Sar’) in view of He et al. (US 20160247045, hereinafter ‘He’) in further view of Zhang (US 20090204557).
Regarding claim 3, the rejection of claim 2 is incorporated and He further teaches the method of claim 2, wherein the shape function is a . (in [0024] Formally, a SVM training algorithm learns the following function from the training samples [wherein the shape function … using kernels learned by the Support vector machine algorithm], … [0029] In cases where non-linear classifiers are needed, Equation (2) can be further modified by using the so-called kernel trick, i.e., replacing the dot product in Equation (1) and Equation (2) by a kernel K(•,•). Since K(•,•) can be interpreted as a dot product in a higher dimensional space, the kernel trick is essentially trying to project the samples into a higher dimensional space where they might become linearly separable. The corresponding QP problem with a kernel K(•,•) is as follows…)
He does not expressly teach the shape function as kernel function using claim spline operations.
Zhang does expressly teach the shape function as kernel function using claim spline operations, in [0009] Kernels [wherein the shape function is a weighted linear combination of B-splines using B-spline coefficients] play a critical role in modern machine learning technologies such as support vector machines (SVM). A support vector machine for classification is defined as an optimal hyperplane in a feature space, which is often a high dimensional (even infinite dimensional) inner product space.… [0010] ... The important features of such data are usually in the distributions of the points in certain spaces, rather than the isolated values of individual points. The standard kernels (e.g., polynomial kernels and Gaussian kernels) are often ineffective on this type of data because the standard kernels treat all vector components equally, so that the large input volumes tend to make the kernels insensitive to the underlying structures and the distributional features of the specific problems. As a result, they are not well suited for distributional data. For example, SVM analysis of flow cytometry data has been reported using radial basis function (RBF) kernels, examples of which are Gaussian and B-spline kernels [wherein the shape function is a weighted linear combination of B-splines using B-spline coefficients] …
Zhang, He and Sar are analogous art because both involve developing information retrieval and processing techniques using machine learning systems and algorithms.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the prior art for retrieving and processing information using support vector machines, as disclosed by Zhang with the method of developing information retrieval and processing techniques for complex decision systems used across a variety of applications in infrastructures, as disclosed collectively by He and Sar.
One of ordinary skill in the arts would have been motivated to combine the methods disclosed by Zhang, He and Sar, as noted above. Doing so allows for training a classifier using kernels that the ability to harvest hidden patterns based solely on the natural similarity measure of the kernel, without using explicit feature extractions, (Zhang, 0009).
Regarding claim 4, the rejection of claim 3 is incorporated and He further teaches the method of claim 3, wherein a fitting objective for fitting the plurality of shape functions is to determine the. (in [0021] Support vector machines (SVMs) are supervised machine learning tools useful in classification and regression analysis. In their original forms, SVMs are designed for binary classification. If non-binary classification is required, a common approach is to decompose the problem into a series of binary classification problem. Given a set of training samples, which are either labelled positive “+” or negative “−” (or whatever other binary notation is desired for identifying one class versus another class), a SVM training algorithm [wherein a fitting objective for fitting the plurality of shape functions is to determine the] intends to find a hyperplane that separates the two classes of training samples with the maximum margin [wherein a fitting objective for fitting the plurality of shape functions is to determine the. In one example, the maximum margin is found when the sum of the distance from the hyperplane [wherein a fitting objective for fitting the plurality of shape functions is to determine the] to the nearest positive sample and the distance from the hyperplane to the nearest negative sample is maximized.)
He does not expressly teach the shape function as kernel function using claim spline operations.
Zhang does expressly teach the shape function as kernel function using claim spline operations, in [0009] Kernels [wherein a fitting objective for fitting the plurality of shape functions is to determine the B-splines coefficients that maximize the divergence] play a critical role in modern machine learning technologies such as support vector machines (SVM). A support vector machine for classification is defined as an optimal hyperplane [wherein a fitting objective for fitting the plurality of shape functions is to determine the B-splines coefficients that maximize the divergence] in a feature space, which is often a high dimensional (even infinite dimensional) inner product space.… [0010] ... The important features of such data are usually in the distributions of the points in certain spaces, rather than the isolated values of individual points. The standard kernels (e.g., polynomial kernels and Gaussian kernels) are often ineffective on this type of data because the standard kernels treat all vector components equally, so that the large input volumes tend to make the kernels insensitive to the underlying structures and the distributional features of the specific problems. As a result, they are not well suited for distributional data. For example, SVM analysis of flow cytometry data has been reported using radial basis function (RBF) kernels, examples of which are Gaussian and B-spline kernels [wherein a fitting objective for fitting the plurality of shape functions is to determine the B-splines coefficients that maximize the divergence] …
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to combine the teachings of Zhang, He and Sar for the same reasons disclosed above.
Regarding claim 6, the rejection of claim 3 is incorporated and Sar in combination with He further teaches the method of claim 3, further comprising intervening by rejecting one or more predictive features. (in [0034] Various actions can be taken in response to an explanation, including training/retraining a machine learning model. For example, when a local explanation provides an indication of features that were more or less relevant to a result, less relevant features can be removed [further comprising intervening by rejecting one or more predictive features] from use with the machine learning model.)
He teaches in [0031] In accordance with one aspect of the present application, instead of formulating SVM training as a QP problem, a desired hyperplane separating the training samples is found directly. Specifically, the processes described below have linear computational complexity O(n), and are also are highly parallelizable. In some embodiments, the processes described herein are capable of working with dynamic training sets, i.e. sets in which samples are added or removed [further comprising intervening by rejecting one or more predictive features] over time…
Regarding claims 12-13, the claim limitations are similar with those in claims 3-4 are thus rejected under a similar rationale.
Regarding claims 17-18, the claim limitations are similar with those in claims 3-4 are thus rejected under a similar rationale.
Claim 5 is rejected under 35 U.S.C. 103 as being unpatentable over Sarferaz (US 20210192376, hereinafter ‘Sar’) in view of He et al. (US 20160247045, hereinafter ‘He’) in further view of Zhang (US 20090204557) and Deng et al. (US 12333397, hereinafter ‘Deng’).
Regarding claim 5, the rejection of claim 3 is incorporated and He further teaches the method of claim 3, further comprising intervening by applying constraints on the (in [0021] Support vector machines (SVMs) are supervised machine learning tools useful in classification and regression analysis. In their original forms, SVMs are designed for binary classification. If non-binary classification is required, a common approach is to decompose the problem into a series of binary classification problem. Given a set of training samples, which are either labelled positive “+” or negative “−” (or whatever other binary notation is desired for identifying one class versus another class), a SVM training algorithm intends to find a hyperplane [further comprising intervening by applying constraints on the ] that separates the two classes of training samples with the maximum margin. In one example, the maximum margin is found when the sum of the distance from the hyperplane to the nearest positive sample and the distance from the hyperplane to the nearest negative sample is maximized... [0058] In operation 104, for each candidate angle θ.sub.j, and for each sample {right arrow over (x)}.sub.i, the length of the radius to a candidate hyperplane passing through the sample is determined From among all the determined radii at angle θ.sub.j, the minimum and maximum are found for both positive and negative samples, i.e. ρ.sub.j,min.sup.−, ρ.sub.j,max.sup.−, ρ.sub.j,min.sup.+, and ρj,max.sup.+, as indicated by operation 106. These maximums and minimums help identify those candidate pairs of hyperplanes [further comprising intervening by applying constraints on the ] that will result in no samples between the pair.)
Zhang teaches, in [0009] Kernels play a critical role in modern machine learning technologies such as support vector machines (SVM) [further comprising intervening by applying constraints on the B-spline coefficients of the plurality of shape functions]. A support vector machine for classification is defined as an optimal hyperplane in a feature space, which is often a high dimensional (even infinite dimensional) inner product space. The construction of the optimal hyperplane requires the inner products, in the feature space, of mapped input vectors. A kernel function defined on the input space provides an effective way to compute the inner products without actually mapping the input to the feature space. The kernel defines a similarity measure between two vectors … [0010] In many applications such as image recognition and flow cytometry data analysis, the input data are usually of high dimensions and in large quantities. The important features of such data are usually in the distributions of the points in certain spaces, rather than the isolated values of individual points. The standard kernels (e.g., polynomial kernels and Gaussian kernels) are often ineffective on this type of data because the standard kernels treat all vector components equally, so that the large input volumes tend to make the kernels insensitive to the underlying structures and the distributional features of the specific problems. As a result, they are not well suited for distributional data. For example, SVM analysis of flow cytometry data has been reported using radial basis function (RBF) kernels, examples of which are Gaussian and B-spline kernels [further comprising intervening by applying constraints on the B-spline coefficients of the plurality of shape functions] …
Zhang teaches the use of spline kernels applying claimed constraints as the B parameter in the B-spline kernel shape functions for determining the classification decision boundary, as noted above.
Additionally, Deng teaches the B parameter as known parameter, in 11:41-49 The key to tackling these exotic shape constraints is to apply expansion over some basis functions at a predetermined knot set. The nonlinear function ƒ.sub.i(x.sub.i) is then approximated by stacking all the basis functions with weights, resulting in an approximation spline function to be either a piecewise linear or polynomial function. As a result, the conceptually complicated nonlinear SVM with shape constraints problem is converted to a standard linear SVM with some bound constraints (coefficients β>0) [further comprising intervening by applying constraints on the B-spline coefficients of the plurality of shape functions]… And in 23:61-67: As shown by operation 1304, the apparatus 200 includes means, such as training engine 210 or the like, for combining the approximation spline functions generated in operation 502 to produce a shape-restricted hyperplane. As set forth in Equation 8 previously, combining the approximation spline functions may comprise a sum of the various spline functions ƒ.sub.i(x.sub.i) plus a coefficient β.sub.0 [further comprising intervening by applying constraints on the B-spline coefficients of the plurality of shape functions].
Deng, Zhang, He and Sar are analogous art because both involve developing information retrieval and processing techniques using machine learning systems and algorithms.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the prior art for retrieving and processing information using shape-restricted support vector machines, as disclosed by Deng with the method of developing information retrieval and processing techniques for complex decision systems used across a variety of applications in infrastructures, as disclosed collectively by Zhang ,He and Sar.
One of ordinary skill in the arts would have been motivated to combine the methods disclosed by Deng, Zhang, He and Sar, as noted above. Doing so allows for training a classifier using a shape-restricted support vector machine that incorporates component-wise shape information to enhance classification accuracy, (Deng, 1:8-13).
Claims 7, 14, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Sar in view of He in further view of Deng.
Regarding claim 7, the rejection of claim 1 is incorporated He further teaches the method of claim 1, further comprising iteratively adding additional predictive features and their associated shape functions to the classifier subject to a stopping criterion based on a predetermined threshold for score separation improvement. in [0031] In accordance with one aspect of the present application, instead of formulating SVM training as a QP problem, a desired hyperplane separating the training samples is found directly. Specifically, the processes described below have linear computational complexity O(n), and are also are highly parallelizable. In some embodiments, the processes described herein are capable of working with dynamic training sets, i.e. sets in which samples are added [further comprising iteratively adding additional predictive features ] or removed over time… [0056] The possible sibling lines or hyperplanes may be referred to as “candidate lines” or “candidate hyperplanes” herein. A pair of “candidate hyperplanes” at the same angle are necessarily parallel to each other and, thus, are sibling lines. The pair of candidate hyperplanes at the same angle that result in an optimal distance d and/or number of samples between them (zero or more), may be selected [and their associated shape functions to the classifier subject to a stopping criterion based on a predetermined threshold for score separation improvement] as the optimal pair of candidate hyperplanes that then serve as sibling hyperplanes for defining the classification hyperplane to be used in the SVM [the classifier subject to a stopping criterion based on a predetermined threshold for score separation improvement as the optimal classifier used with the selected claimed elements ]. And in [0071] In operation 208, a count of the number of samples on the “wrong” side of the respective hyperplanes is determined. Because it is not necessarily known which hyperplane corresponds to which class, the count may be determined for both options. In one example, for k=0, the count C is assessed as:... [0074] In operation 212, the method 200 evaluates whether the search space is finished [the classifier subject to a stopping criterion based on a predetermined threshold for score separation improvement as the optimal classifier used with the selected claimed elements]. If not, then index k is incremented and radius length l is increased by the step size δ in operation 214. Subsequent iterations of operation 208 may evaluate the expressions:..
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to combine the teachings of He and Sar for the same reasons disclosed above.
Additionally, Deng teaches in 21:53-22:26: …The training engine 210 may select the set of shape restrictions in any number of ways. For instance, the user may provide, via input-output circuitry 208 of the apparatus 200 (or by a separate client device, and then relayed to the apparatus 200 via communications circuitry 206) input comprising a shape restriction selection for one or more of the features in the training dataset. Following receipt of any shape restriction selections, the training engine 210 may then select the set of shape restrictions to include the shape restriction selections provided by the user. However, the user may not provide shape restriction selection for any of the features in the training dataset. In such situations, the apparatus 200 may utilize a trial-and-error approach to identify shape-restriction information for one or more of the features in the training dataset. To this end, the training engine 210 may initially identify a linear shape restriction for every feature in the training dataset. Subsequently, the training engine 210 may generate an approximation spline function for the various features in the training dataset using a monotone increasing or decreasing shape restriction selection [further comprising iteratively adding additional predictive features and their associated shape functions to the classifier subject to a stopping criterion based on a predetermined threshold for score separation improvement]. Where the approximation spline function for a given feature is a flat line (e.g., having a slope of zero), that indicates that the assigned shape restriction is not the correct shape-restriction for the given feature, and the training engine 210 then selects the other monotone shape restriction for that feature and generates a new approximation spline function for the feature. If the new approximation spline function for the feature does not comprise a flat line, the training engine 210 selects additional convex or concave shape restriction to the monotone shape restriction of the feature and generates another new approximation spline function for the feature. This iterative process may occur until a shape restriction is selected for the feature such that the approximation spline function does not produce a flat line [… subject to a stopping criterion based on a predetermined threshold for score separation improvemen], at which point the then-current shape restriction for the feature is selected and used for training of the SR-SVM. This process may be performed by the training engine 210 for each feature in the training dataset to select the set of shape restrictions for the training dataset even in situations where there is no a priori knowledge of the shape restrictions for the training dataset…
Deng, He and Sar are analogous art because both involve developing information retrieval and processing techniques using machine learning systems and algorithms.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the prior art for retrieving and processing information using shape-restricted support vector machines, as disclosed by Deng with the method of developing information retrieval and processing techniques for complex decision systems used across a variety of applications in infrastructures, as disclosed collectively by He and Sar.
One of ordinary skill in the arts would have been motivated to combine the methods disclosed by Deng, He and Sar, as noted above. Doing so allows for training a classifier using a shape-restricted support vector machine that incorporates component-wise shape information to enhance classification accuracy, (Deng, 1:8-13).
Regarding claims 14 and 19, the claim limitations are similar with those in claim 7 are thus rejected under a similar rationale.
Claims 8 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Sar in view of He and in further view of Lee et al. (US 20240118178, hereinafter ‘Lee’).
Regarding claim 8 , the rejection of claim 1 is incorporated.
Lee teaches the method of claim 1, further comprising providing a visualization of the fitted shape functions as 2-dimensional plots for classifier interpretation. (in [0039] The term “decision function” as used herein refers to a function that calculates the distance of the subject to the separating hyperplane of SVM classifier… [0091] …. The training set with the selected features was applied to train the models using three different types of machine learning (ML) algorithms. Random Forest (RF), Logistic Regression (LR), and Support Vector Machines (SVM) were used to classify flow cytometry data from 2 classes of HC and NPC subjects. To minimize the impact of the feature with a relatively higher magnitude on the distance calculation, the Min-Max scaling was applied for SVM to ensure that every feature had a similar effect when the classifier constructed the hyperplane [further comprising providing a visualization of the fitted shape functions as 2-dimensional plots for classifier interpretation]. The Min-Max scaling is a normalization technique that transforms the minimal feature value to 0 and the maximal feature value to 1. The discriminative ability of the models was then evaluated by the area under curve (AUC) of the receiver operating characteristic (ROC) curve…. The better the discriminative ability of the model, the closer the ROC curve is to the upper left corner of the plot. Finally, the discriminative ability of the models was quantified by computing the area under the ROC curve using the trapezoidal rule to obtain the AUC results. The ROC curves and the AUC results of the training set were visualized to compare the model performance. The plots are shown in FIGS. 3A-3C. The Shapley Additive exPlanations (SHAP) was applied to explain the model by computing the contribution of each selected feature to prediction [further comprising providing a visualization of the fitted shape functions as 2-dimensional plots for classifier interpretation]. The SHAP summary plot was depicted to visualize the ranking of feature importance and the value of the feature per subject with the SHAP values [further comprising providing a visualization of the fitted shape functions as 2-dimensional plots for classifier interpretation]. The color of the data point in each feature represents a high or a low feature value, in which each data point represents one subject. Red indicates high feature value, and blue indicates low feature value. The y-axis of the plot is the feature importance ranking of the selected features, and the x-axis is the SHAP value range. According to the trained model, the higher the SHAP value, the higher the risk of NPC. The summary plots of the training set are shown in FIGS. 4A-4C. )
Lee, He and Sar are analogous art because both involve developing information retrieval and processing techniques using machine learning systems and algorithms.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the prior art for retrieving and processing information using support vector machines, as disclosed by Lee with the method of developing information retrieval and processing techniques for complex decision systems used across a variety of applications in infrastructures, as disclosed collectively by He and Sar.
One of ordinary skill in the arts would have been motivated to combine the methods disclosed by Lee, He and Sar, as noted above. Doing so allows for selection of important features with an acceptable range using visualized model performance plots, (Lee, 0091).
Regarding claim 20 the claim limitations are similar with those in claim 8 are thus rejected under a similar rationale.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Mathewson et al. (US 20040034612): [0056] A regression SVM (see, e.g., FIG. 3) determines a best fit curve to selectively fit input data. Regression modeling makes use of splines or Bezier curves as kernel functions. Again, key data points may be identified, and the degree of exactness of fit tailored to appropriate accuracy.
Golub et al. (US 20030225526): teaches in [0088] SVMs assume the target values are binary and that the classification problem is intrinsically binary.
Vilkamo et al. (US 20230402050): teaches in [0138] ... The SVM can comprise a supervised binary classifier that receives training data with known class labels as an input. In this case the training data would comprise the extracted audio feature vectors that are comprised within the classifier input 407. The SVM can be configured to find a line or possibly a multi-dimensional (hyper)plane that separates two classes in the feature space. The SVM algorithm is configured to maximize, or substantially maximize, the distance to the nearest samples of both classes from the hyperplane. These nearest samples are called “support vectors”. The obtained hyperplane shape is dependent on the underlying kernel function. Any suitable functions such as which linear, polynomial, sigmoid, or radial basis functions can be used for the kernel function. The most suitable function to use for the kernel function depends on the case. For example, a linear kernel is not suitable for a non-linear classification problem. For a non-linear classification problem a transformation to a higher feature space dimension (a so-called kernel trick) should be applied instead. For cases that comprise classification into a plurality of classes a plurality of SVM binary outputs are combined at the end of each classification task to find the most probable output class. In some examples a library such as a libSVM, or any other suitable type of library, could be used by the classifier block 409. An example type of SVM that could be used would be a C-Support Vector Classification (C-SVC) and radial basis function could be used as a kernel function.
Szepannek (NPL: An overview on the landscape of packages for credit scoring): teaches receiving aggregated statistics objects, wherein the aggregated statistics objects comprise: bin frequencies F0 and F1, wherein the bin frequencies F0 and F1 are calculated for each of a plurality of predictive features, conditioned on a target value being 0 or 1, respectively; (pgs. 3-4: Binning of numeric variables is often considered to be the most relevant part of a scorecard development... It is important to note that binning not just corresponds to exploratory data analysis but its results have to be considered as an integral part of the model, i.e. the resulting preprocessing has to be applied to new data in order to be able to make use of the final scorecard model. For this reason important requirements on an implementation of the binning step are the possibility to (i) store the binning results for all variables as well as to (ii) apply the binning to new data with some kind of predict() function… Often binning is followed by subsequent assignment of numeric weights of evidence to the factor levels x of the binned variable which are given by:
PNG
media_image1.png
92
314
media_image1.png
Greyscale
(1) …Note that just like the bins, the WoEs as computed on the training data of course also belong to the model. Further, an implementation of WoE computation has to take into account for potentially occurring bins that are empty w.r.t. the target level y = 0 (typically by adding a small constant when computing the relative frequencies f())…)
Wang et al. (US 20240177071): teaches in [0003] Classification (e.g., predicting a likelihood of given data instances to be different categories, etc.) is a fundamental problem in machine learning (ML). Numerous classification models have been proposed for this problem, including traditional models (e.g., support vector machines (SVMs), naïve Bayes classifiers, etc.), ensemble learning models (e.g., random forest models, tree boosting models, etc.), and deep learning models (e.g., convolution neural networks (CNNs), recurrent neural networks (RNNs), etc.)… [0015] In some non-limiting embodiments or aspects, the model interpretation technique is a model interpretation technique that involves Shapley additive explanations (SHAP) values.
Hu et al. (US 20240029154): teaches in [0003] As discussed above, the interpretability of machine learning (ML) algorithms has been the subject of considerable discussion in recent years. Early approaches relied on post hoc techniques, including variable importance, partial dependence plots or PDPs, and H-statistics. These are low-dimensional summaries of high-dimensional models with complex structure, and hence can be inadequate in capturing the full picture. A second approach for model interpretability is the use of surrogate models (or distillation techniques) that fit simpler models to extract information and explanations from the original complex models. Examples include: i) local interpretable model-agnostic (LIME) models which are based on linear models for local explanations; and ii) locally additive trees for local and global explanation.
Shakerin et al. (NPL: White-box Induction From SVM Models: Explainable AI with Logic Programming): teaches the categorical features should be binarized before an SVM model can be trained. Binarization (aka one-hot encoding) is the process of transforming each categorical feature with domain of cardinality n, into n new binary predicates (features)… SHAP is able to quantitatively explain the features that would push the model towards predicting a specific outcome. In particular, for each support vector, SHAP determines a subset of feature value pairs that would make the model arrive at a certain decision. It turns out that just by having the Shapley values of support vectors, our algorithm can learn the global underlying behavior of SVM model.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to OLUWATOSIN ALABI whose telephone number is (571)272-0516. The examiner can normally be reached Monday-Friday, 8:00am-5:00pm EST..
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Michael Huntley can be reached at (303) 297-4307. 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.
/OLUWATOSIN ALABI/ Primary Examiner, Art Unit 2129
Prosecution Insights