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
Status of Claims
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
The following is a FINAL OFFICE ACTION in response to applicant’s amendments to and response for Application #18/106,146, filed on 03/19/2026.
Claims 1-20 are pending and have been examined.
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 non-statutory subject matter. The rationale for this finding is explained below.
Per Step 1 of the analysis, the claims are analyzed to determine if they are directed to statutory subject matter. Claim 1 claims a method, or process. A process is a statutory category for patentability. Claim 14 claims an apparatus. The apparatus comprises a processor coupled to a memory. An apparatus is a statutory category for patentability. Claim 18 claims an article of manufacture comprising a non-transitory processor-readable storage medium. An article of manufacture, which is a statutory category for patentability. Further, the claim is in conformity with the Kappos Memorandum of 2010 regarding medium claims, as the claim includes the phrase “non-transitory.”
Per Step 2A, Prong 1 of the analysis, the examiner must now determine if the claims recite an abstract idea or eligible subject matter. In the instant case, the independent claims are directed towards an abstract idea. Specifically, independent claims 1, 14, and 18 recite “receiving product purchase data, product service data, or product return data corresponding to a plurality of users and product operation data corresponding to operations of products of the plurality of users, receiving an input dataset corresponding to at least one user, wherein the input dataset comprise product purchase data, product service data, and product return data and product operation data corresponding to operation of at least one product associated with the at least one user, analyzing the input dataset for predicting a risk level of the at least one user, determining a prediction result which indicates a likelihood of whether the at least one product corresponding to the at least one user will fail to be returned to a product providing entity when a return of the at least one product has been requested based at least in part on a predicted risk level of the at least one user and a predicted operational state of the at least one product and reporting the prediction result to the product providing entity. Therefore, the claims recite an abstract idea, namely a mental process. A human operator with access to the historical purchase or return data of a plurality of users and the current user data could analyze the data associated with a current user who has requested a return and calculate or determine the probability that the user will actually return an item for which they have requested a return. These steps could all be done mentally with aid of pen and paper or other data files. The claims simply automate these practices using a computer. The claims secondarily recite “certain methods of organizing human activity.” Specifically, the claims recite “sales activities, business relations.” A business owner managing product sales can identify users who have initiated returns and identify based on stored data of a plurality of users and data regarding the current user whether they are likely to return the product. These types of analyses are commonly done by business owners, especially with larger ticket items, and are important for revenue expectations and inventory management. The claims simply automate these practices using a computer.
Per Step 2A, Prong 2 of the analysis, the examiner must now determine if the claims integrate the abstract idea into a practical application. The additional elements of the independent claims include “implementing a risk prediction engine which executes on at least one processing platform in a computing network to perform risk prediction and remediation operations,” “by the risk prediction engine,” “a processing device operatively coupled to a memory,” and a “processor-readable storage medium.” However, these additional elements are considered generic recitations of a technical element and are recited at a high level of generality. These additional elements are being used as “tools to automate the abstract idea” (see MPEP 2106.05 (f)), and do not integrate the abstract idea into a practical application. They are not recitations of a special purpose computer or transformation (see MPEP 2106.05 (b) and (c)). The additional elements also include “executing by the risk prediction engine machine learning training processes to train machine learning models using a training dataset…,” and analyzing “using the trained machine learning models for predicting….” These additional elements, absent further detail on how the algorithm is trained or the steps for using the algorithm, are considered generic recitations of technical elements, the equivalent of “apply it,” or using a computer as a tool to automate the abstract idea. The computer is simply automating the abstract idea using an algorithm. The examiner points the applicant to the recent 2024 USPTO Memorandum regarding machine learning claims in which it is clearly laid out that the claims should show detailed steps outlining the training or use of an algorithm or mode, and the tie to a subsequent technical step. Therefore, this additional element is not considered to integrate the abstract idea into a practical application.
Per Step 2B of the analysis, the examiner must now determine if the claims include limitations that are “significantly more” than the abstract idea by demonstrating an improvement to another technology or technical field, an improvement to the functioning of the computer itself, or meaningful limitations beyond generally linking the use of an abstract idea to a particular technological environment. The additional elements of the independent claims include “implementing a risk prediction engine which executes on at least one processing platform in a computing network to perform risk prediction and remediation operations,” “by the risk prediction engine,” “a processing device operatively coupled to a memory,” and a “processor-readable storage medium.” However, these additional elements are considered generic recitations of a technical element and are recited at a high level of generality. These additional elements are being used as “tools to automate the abstract idea” (see MPEP 2106.05 (f)), and are not considered significantly more than the abstract idea itself. They are not recitations of a special purpose computer or transformation (see MPEP 2106.05 (b) and (c)). The additional elements also include “executing by the risk prediction engine machine learning training processes to train machine learning models using a training dataset…,” and analyzing “using the trained machine learning models for predicting….” These additional elements, absent further detail on how the algorithm is trained or the steps for using the algorithm, are considered generic recitations of technical elements, the equivalent of “apply it,” or using a computer as a tool to automate the abstract idea. The computer is simply automating the abstract idea using an algorithm. The examiner points the applicant to the recent 2024 USPTO Memorandum regarding machine learning claims in which it is clearly laid out that the claims should show detailed steps outlining the training or use of an algorithm or mode, and the tie to a subsequent technical step. Therefore, this additional element is not considered significantly more.
When considered as an ordered combination, the claim is still considered to be directed to an abstract idea as the claim steps in the ordered combination simply recite the logical steps for receiving and analyzing historical purchase and return data for a plurality of users and current data for a user and predicting the likelihood of actual return of the product following a request for a return. Therefore, the ordered combination does not lead to a determination of significantly more.
When considering the dependent claims, claims 2-5 are considered part of the abstract idea as they simply expand on the steps already determined to be abstract above also using the same trained algorithm. The use of specifically a stochastic model is considered insignificant extra-solution activity and does not integrate the abstract idea into a practical application. Claim 6 expands the training to include an additional training set. This additional element is still considered a generic recitation of a technical element, the equivalent of “apply it,” or using a computer as a tool to automate the abstract idea. Claim 7 is considered part of the mental process. Claim 8 is considered part of the abstract idea as they simply expand on the steps already determined to be abstract above. The use of specifically multiple linear regression is considered insignificant extra-solution activity and does not integrate the abstract idea into a practical application. Claims 9 and 11 are considered part of the abstract idea. Claim 10 is considered part of the abstract idea as an analyzing step. The use of binary logistic regression is considered the equivalent of “apply it,” or using a computer as a tool to automate the abstract idea. Claims 12-13 are considered part of the abstract idea as an analyzing step. The use of a supervised learning algorithm and/or specifically a k-nearest neighbor algorithm is considered the equivalent of “apply it,” or using a computer as a tool to automate the abstract idea. The other dependent claims mirror those already discussed above.
Therefore, claims 1-20 are rejected under 35 U.S.C. 101 as being directed to non-statutory subject matter. See Alice Corporation Pty. Ltd. Vs. CLS Bank International et al., 2014 (please reference link to updated publicly available Alice memo at http://www.uspto.gov/patents/announce/alice_pec_25jun2014.pdf as well as the USPTO January 2019 Updated Patent Eligibility Guidance.)
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries set forth in Graham v. John Deere Co., 383 U.S. 1, 148 USPQ 459 (1966), that are applied 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 1-4, 6, 14, and 18 are rejected under 35 USC 103 as being unpatentable over Burris, Pre-Grant Publication No. 2022/0292519 A1 in view of Murali, et al., Pre-Grant Publication No. 2024/0103959 A1 and in further view of Singh, et al., Pre-Grant Publication No. 2021/0264438 A1.
Regarding Claims 1, 14, and 18, Burris teaches:
A method comprising:
implementing a risk prediction engine which executes on at least one processing platform in a computing network to perform risk prediction and remediation operations which comprise: (see Figure 1 and [0011]-[0026])
executing, by the risk prediction engine, machine learning training processes to train machine learning models using a training dataset corresponding to a plurality of users, wherein the training dataset comprises product purchase data, and product return data corresponding to the plurality of users (see [0057] in which a trained machine learning algorithm is trained on purchase and return data of a plurality of users)
receiving… an input dataset corresponding to at least one user, wherein the input dataset comprises product purchase data, product return data corresponding to the at least one user (see [0055]-[0056] in which the input data set is for a current user and their information regarding initiating of a return, especially [0056] in which the user has initiated the return request but has not yet visited the retailer or other entity in person to complete the return of the item)
analyzing… the input dataset using the trained machine learning models (see [0056]-[0059])
for predicting a risk level of the at least one user and determining a prediction result which indicates a likelihood of whether the at least one product corresponding to the at least one user will fail to be returned to a product providing entity when a return of the at least one product has been requested based at least in part on a predicted risk level of the at least one user (see [0057] in which the trained ML algorithm predicts the likelihood of the user returning or failing to go forward with the return as well as predicts other aspects of the potential return)
reporting…the prediction result to the product providing entity (see Abstract, [0055], [0057], and [0075]-[0076])
Burris, however, does not appear to specify:
(wherein the training dataset comprises) product operation data corresponding to operations of products of the plurality of users
predicting an operational state of the at least one product
Murali teaches:
(wherein the training dataset comprises) product operation data corresponding to operations of products of the plurality of users (see Abstract, [0005]-[0007], and [0016])
predicting an operational state of the at least one product (see Abstract, [0005]-[0007], and [0016] in which a probability of a future operational state of a product, such as reduction in optimal performance or failure, is predicted based on the receiving and analysis of as current operational state of an asset such as a device or computer)
It would have been obvious to one of ordinary skill in the art at the time of the filing of the application to combine Murali with Burris because Burris already teaches prediction using ML of other states of a product and identifies and predicts user returns, and users often return products due to defects or failures, and knowing operational states of products and predictions of future operational states could allow for addressing of any potential issues before they occur, which would lead to less product returns.
Burris and Murali, however, does not appear to specify:
a prediction result…based at least in part on a predicted risk level of the at least one user and a predicted operational state of the at least one product
Burris does, however, teach based at least in part on a predicted risk level of the at least one user in [0057] in which the trained ML algorithm predicts the likelihood of the user returning or failing to go forward with the return as well as predicts other aspects of the potential return. Murali does teach predicting an operational state of the at least one product in such as Abstract, [0005]-[0007], and [0016] in which a probability of a future operational state of a product, such as reduction in optimal performance or failure, is predicted based on the receiving and analysis of as current operational state of an asset such as a device or computer.
Therefore, it would have been obvious to one of ordinary skill in the art at the time of the filing of the application to combine a prediction result…based at least in part on a predicted risk level of the at least one user and a predicted operational state of the at least one product with Burris and Murali because Burris already teaches prediction using ML of other states of a product and identifies and predicts user returns, and Murali teaches predicting operational states of products and predictions based on future operational states AND risk predictions for the user could allow for a better prediction of return likelihood and allow addressing of any potential issues before they occur, which would lead to less product returns.
Burris and Murali, however, does not appear to specify:
(wherein the training dataset comprises) product service data
Singh teaches:
(wherein the training dataset comprises) product service data (see [0059]-[0061] in which the historical product data used includes servicing and maintenance data, and in which the models are trained on the data prior to using the historical data)
Therefore, it would have been obvious to one of ordinary skill in the art at the time of the filing of the application to combine Singh with Burris and Murali because Burris already teaches prediction using ML of other states of a product and identifies and predicts user returns, and Murali teaches predicting operational states of products and using product service data would give another reliable and relevant data point that can be used in the prediction calculations.
Regarding Claim 2, the combination of Burris, Murali, and Singh teaches:
the method of claim 1
Murali teaches:
wherein the at least one product comprises at least one of a device and a device component (see Abstract, [0005]-[0007], and [0016] in which a probability of a future operational state of a product is predicted based on the receiving and analysis of as current operational state of an asset such as a device or computer)
wherein the operation data corresponding to the operation of the at least one product comprises an operational state the at least one product (see Abstract, [0005]-[0007], and [0016] in which a probability of a future operational state of a product is predicted based on the receiving and analysis of as current operational state of an asset such as a device or computer)
predicting the operational state of the at least one product comprises predicting one or more future operational states of the at least one product based, at least in part, on the data corresponding to the operation of the at least one product (see Abstract, [0005]-[0007], and [0016] in which a probability of a future operational state of a product, such as reduction in optimal performance or failure, is predicted based on the receiving and analysis of as current operational state of an asset such as a device or computer)
It would have been obvious to one of ordinary skill in the art at the time of the filing of the application to combine Murali with Burris because Burris already teaches prediction using ML of other states of a product and identifies and predicts user returns, and users often return products due to defects or failures, and knowing operational states of products and predictions of future operational states could allow for addressing of any potential issues before they occur, which would lead to less product returns.
Regarding Claim 3, the combination of Burris, Murali, and Singh teaches:
the method of claim 2
Murali further teaches:
…predicting one or more probabilities of respective ones of the one or more future operational states of the at least one product (see Abstract, [0005]-[0006], and [0016])
It would have been obvious to one of ordinary skill in the art at the time of the filing of the application to combine Murali with Burris because Burris already teaches prediction using ML of other states of a product and identifies and predicts user returns, and users often return products due to defects or failures, and knowing operational states of products and predictions of future operational states could allow for addressing of any potential issues before they occur, which would lead to less product returns.
Regarding Claim 4, the combination of Burris, Murali, and Sethi teaches:
the method of claim 3
Murali further teaches:
wherein the predicting of the one or more probabilities of the respective ones of the one or more future operational states is performed using a stochastic model (see at least Abstract, [0005]-[0007], [0016], [0046], [0050], and [0052])
It would have been obvious to one of ordinary skill in the art at the time of the filing of the application to combine Murali with Burris because Burris already teaches prediction using ML of other states of a product and identifies and predicts user returns, and users often return products due to defects or failures, and knowing operational states of products and predictions of future operational states could allow for addressing of any potential issues before they occur, which would lead to less product returns.
Regarding Claim 6, the combination of Burris, Murali, and Singh teaches:
the method of claim 2
Murali further teaches:
wherein the product operation data of the training dataset corresponding to operations of the plurality of products comprises information regarding one or more changes in operational states for respective ones of the plurality of products (see at least Abstract, [0005]-[0007], [0016], [0046], [0050], and [0052])
Singh further teaches:
wherein the product operation data of the training dataset corresponding to operations of the plurality of products comprises information regarding one or more changes in operational states for respective ones of the plurality of products (see [0058]-[0061])
Claim 5 is rejected under 35 U.S.C. 103 as being unpatentable over Burris, Pre-Grant Publication No. 2022/0292519 A1 in view of Murali, et al., Pre-Grant Publication No. 2024/0103959 A1 and in further view of Singh, et al., Pre-Grant Publication No. 2021/0264438 A1 and in further view of Prasad, et al., Pre-Grant Publication No. 2021/0374013 A1.
Regarding Claim 5, the combination of Burris, Murali, and Singh teaches:
the method of claim 3
Burris, Murali, and Sethi, however, does not appear to specify:
wherein the predicting of the one or more future operational states of the at least one product comprises using a conformal prediction model to predict the one or more future operational states with a confidence level based on the one or more probabilities
Prasad further teaches:
wherein the predicting of the one or more future operational states of the at least one product comprises using a conformal prediction model to predict the one or more future operational states with a confidence level based on the one or more probabilities (see at least [0025]-[0028] and [0076]-[0077] in which a conformal prediction model is used to predict a future operational state such as normal or failed with a confidence level based on the probabilities)
It would have been obvious to one of ordinary skill in the art at the time of the filing of the application to combine Prasad with Burris, Murali, and Sethi because Burris already teaches prediction using ML of other states of a product and Murali teaches operational states of products and predictions of future operational states, and using a conformal prediction model and using confidence levels would give a better gauge to the receiver of the likelihood accuracy of the prediction based on a confidence level and not just a statement of possibility.
Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Burris, Pre-Grant Publication No. 2022/0292519 A1 in view of Murali, et al., Pre-Grant Publication No. 2024/0103959 A1 and in further view of Singh, et al., Pre-Grant Publication No. 2021/0264438 A1 and in further view of Auvenshine, et al., Pre-Grant Publication No. 2018/0270128 A1.
Regarding Claim 7, the combination of Burris, Murali, and Singh teaches:
the method of claim 2
Burris, Murali, and Singh, however, does not appear to specify:
assigning a priority to the at least one product and to the at least one user based at least in part on the one or more future operational states of the at least one product
Auvenshine teaches:
assigning a priority to the at least one product and to the at least one user based at least in part on the one or more future operational states of the at least one product (see [0031] in which computational components/products and users are prioritized for repair/replacement based on a prediction of future change in operational state, namely failure or reduction in performance of the component or product)
It would have been obvious to one of ordinary skill in the art at the time of the filing of the application to combine Auvenshine with Burris, Murali, and Singh because Murali teaches operational states of products and predictions of future operational states, and prioritizing products and users based on future operational state predictions would ensure that those products and users that are most likely to imminently experience an operational state change have any concern or potential disruption addressed before it occurs.
Claims 8, 9, 15, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Burris, Pre-Grant Publication No. 2022/0292519 A1 in view of Murali, et al., Pre-Grant Publication No. 2024/0103959 A1 and in further view of Singh, et al., Pre-Grant Publication No. 2021/0264438 A1 and in further view of Doreswamy, et al., Pre-Grant Publication No. 2021/0241288 A1.
Regarding Claims 8, 15, and 19, the combination of Burris, Murali, and Singh teaches:
the method of claim 1…
Burris, Murali, and Singh, however, does not appear to specify:
wherein analyzing the input dataset using the trained machine learning models comprises using a trained machine learning model to analyze one or more independent variables to determine a trust factor of the at least one user
Doreswamy teaches:
wherein analyzing the input dataset using the trained machine learning models comprises using a trained machine learning model to analyze one or more independent variables to determine a trust factor of the at least one user (see [0061]-[0066] in which a trust factor for a user is determined using a trained ML algorithm that analyzes independent variables of the user)
It would have been obvious to one of ordinary skill in the art at the time of the filing of the application to combine Doreswamy with Burris, Murali, and Singh because Burris teaches using a trained ML model to predict likelihood of a user following through on a return request, and determining a trust factor the user will actually return a request would give another measure of determination, allowing for better forecasting and decision making by a business to address repairs and returns.
Burris, Murali, Singh, and Doreswamy, however, does not appear to specify:
using multiple linear regression…
The examiner, however, takes Official Notice that it is old and well known in the computer arts to use a multiple linear regression to analyze variables. Companies such as IBM, Microsoft, and Google have done so for at least a decade prior to the filing date of the application.
Therefore, it would be obvious to one of ordinary skill in the art to combine using multiple linear regression… with Burris, Murali, Singh, and Doreswamy because Burris and Murali already teach training and use of ML models and using a multiple linear regression maximizes the ability to predict how one outcome would be influenced by the user variables.
Regarding Claim 9, the combination of Burris, Murali, Singh, and Doreswamy teaches:
the method of claim 8
Doreswamy further teaches:
wherein the one or more independent variables comprise at least one of a number of service requests created for at least one of the at least one product and the at least one user in a designated time period, a number of service requests for at least one of the at least one product and the at least one user that have been reopened, and a ratio of a number of products for which issues have been reported by the at least one user to a number of products purchased by the at least one user (see [0050]-[0057] in which the independent user variables include the number of service requests in a time period and the ratio of return requests to purchases)
Therefore, it would be obvious to one of ordinary skill in the art to combine Doreswamy with Burris, Murali, and Singh because Burris already teaches prediction using ML of other states of a product and identifies and predicts user returns, and Murali teaches predicting operational states of products and using product service data would give another reliable and relevant data point that can be used in the prediction calculations.
**The examiner notes that Singh has already been shown to teach service requests for MLM training and use, but in order to preserve the finality of this Office Action the rejection of claim 9 has not been changed.**
Claims 10-13, 16, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Burris, Pre-Grant Publication No. 2022/0292519 A1 in view of Murali, et al., Pre-Grant Publication No. 2024/0103959 A1 and in further view of Singh, et al., Pre-Grant Publication No. 2021/0264438 A1 and in further view of Doreswamy, et al., Pre-Grant Publication No. 2021/0241288 A1 and in further view of Dinh, et al., Pre-Grant Publication No. 2021/0126845 A1.
Regarding Claims 10, 16, and 20, the combination of Burris, Murali, and Singh teaches:
the method of claim 1…
Burris, Murali, and Singh, however, does not appear to specify:
wherein analyzing the input dataset using the trained models comprises analyze one or more independent variables to determine whether the at least one user corresponds to a risk that the at least one product will fail to be returned to the product providing entity
Doreswamy teaches:
wherein analyzing the input dataset using the trained models comprises analyze one or more independent variables to determine whether the at least one user corresponds to a risk that the at least one product will fail to be returned to the product providing entity (see [0050]-[0057] and [0061]-[0066])
It would have been obvious to one of ordinary skill in the art at the time of the filing of the application to combine Doreswamy with Burris, Murali, and Singh because Burris teaches using a trained ML model to predict likelihood of a user following through on a return request, and determining a risk factor the user will actually return a request would give another measure of determination, allowing for better forecasting and decision making by a business to address repairs and returns.
Burris, Murali, Singh, and Doreswamy, however, does not appear to specify:
using binary logistic regression…
Dinh teaches:
using binary logistic regression… (see [0040]-[0042], [0058], and claim 1 in which binary logistic regression is used as a type of supervised learning in order to analyze the independent variables to classify the user/product into a risk level such as “likely to fail” or “not likely to fail” or other similar risk levels)
Therefore, it would be obvious to one of ordinary skill in the art to combine Dinh with Burris, Murali, Singh, and Doreswamy because Burris and Doreswamy already teach training and use of ML models and using a binary logistic regression would maximizes the ability to predict how one of two possible outcomes would be influenced by the user variables.
Regarding Claim 11, the combination of Burris, Murali, Singh, Doreswamy, and Dinh teaches:
the method of claim 10
Doreswamy further teaches:
wherein the one or more independent variables comprise at least one of a trust factor of the at least one user (see [0061]-[0067], a geographic location associated with the at least one user (see [0065] and [0077] , a type of the at least one user (see [0048]-[0049], one or more product types associated with the at least one user (see [0049]-[0050])
It would have been obvious to one of ordinary skill in the art at the time of the filing of the application to combine Doreswamy with Burris, Murali, and SIngh because Burris teaches using a trained ML model to predict likelihood of a user following through on a return request, and using these independent variables would use variables specific to returns to make a better estimate of customer behavior.
Regarding Claims 12 and 17, the combination of Burris, Murali, Singh, Doreswamy, and Dinh teaches:
the method of claim 10
Dinh further teaches:
classifying a risk level to which the at least one user corresponds in response to an affirmative determination and wherein the one or more machine learning algorithms perform the classifying and comprise a supervised learning algorithm (see [0040]-[0042], [0058], and claim 1 in which binary logistic regression is used as a type of supervised learning in order to analyze the independent variables to classify the user/product into a risk level such as “likely to fail” or “not likely to fail” or other similar risk levels)
Therefore, it would be obvious to one of ordinary skill in the art to combine Dinh with Burris and Doreswamy because Burris, Murali, Singh, and Doreswamy already teach training and use of ML models and classifying a risk level to which the at least one user corresponds in response to an affirmative determination and wherein the one or more machine learning algorithms perform the classifying and comprise a supervised learning algorithm would maximizes the ability to predict how the outcomes would be influenced by the risk level of the particular user.
Regarding Claim 13, the combination of Burris, Murali, Singh, Doreswamy, and Dinh teaches:
the method of claim 12
Burris, Murali, Singh, Doreswamy, and Dinh, however, does not appear to specify:
wherein the supervised learning algorithm comprises k-nearest neighbor algorithm
The examiner, however, takes Official Notice that it is old and well known in the computer arts when using a supervised learning algorithm to use a k-nearest neighbor algorithm. Companies such as IBM, Microsoft, and Google have done so for at least a decade prior to the filing date of the application.
Therefore, it would be obvious to one of ordinary skill in the art to combine k-nearest neighbor algorithm with Burris, Murali, Singh, Doreswamy, and Dinh because Dinh already teaches using of a supervised learning model and using a k-nearest neighbor algorithm allows for classification using similarity between data points such as the user variables.
Response to Arguments
Regarding the rejections based on 35 USC 101
Regarding the applicant’s argument on page 9 of the response that “the Memo essentially instructs…that claim limitations which encompass artificial intelligence and machine learning in a way that cannot be practically performed in the human mind do not fall within the mental process grouping:”
The applicant is interpreting the Memorandum erroneously. First of all, in the MPEP 2106.04, Step 2A, Prong 1 is described as the examiner determining if the claims recite an abstract idea. That does not mean that if there is any AI or machine learning present that automatically negates the existence of a recited abstract idea. There is no AI or machine learning that can in and of itself be practically performed in the human mind. But, the analysis can address the AI/machine learning in the following steps after identifying an abstract idea.
A good example of the USPTO’s intention with these Memorandums is looking at Example 47 of the Updated Eligibility Guidance examples, which is specifically used to address claims with machine learning. Claim 2 is considered ineligible because even thought it clearly contains the training and use of an AI model, it is recited at a high level of generality and considered the equivalent of “apply it,” unlike Claim 3 of the example which adds steps at the end that show an improved technical process as a “practical application.”:
CLAIM 2:
A method of using an artificial neural network (ANN) comprising:
(a) receiving, at a computer, continuous training data;
(b) discretizing, by the computer, the continuous training data to generate input data;
(c) training, by the computer, the ANN based on the input data and a selected training algorithm to generate a trained ANN, wherein the selected training algorithm includes a backpropagation algorithm and a gradient descent algorithm;
(d) detecting one or more anomalies in a data set using the trained ANN;
(e) analyzing the one or more detected anomalies using the trained ANN to generate anomaly data; and
(f) outputting the anomaly data from the trained ANN.
Regarding the applicant’s argument on pages 9-10 of the response that “Step 2A determines whether the claim as a WHOLE is not directed to a judicial exception” and in Step 2B that the “claim as a WHOLE amounts to significantly more…”:
The examiner responds that the applicant is taking the process out of context. Under this reasoning, if there was a single element that was not part of the judicial exception, such as electronic transmission of data, using a MLM, or even the use of a computer, then the claim would be eligible. The whole point of Step 2A, Prong 1 is to determine whether the claims RECITE AND ABSTRACT IDEA. This does not mean that each and every element of the claims need be fully encompassed by the abstract idea in order for the claims to be considered ineligible in this step and determined to recite an abstract idea. The whole point of Steps 2A, Prong 2 and Step 2B is to examine the ADDITIONAL ELEMENTS that are not recited as part of the abstract idea and do a separate analysis.
The examiner points the applicant to Court decisions such as buySAFE v Google and OIP Techs v Amazon.com which recite e-commerce systems with multiple interacting components like servers, processors, interfaces, networks, databases, etc. performing multiple processes and yet the Court still found the claims were directed to abstract ideas.
Therefore the applicant’s arguments in light of the amendments to the claims are not persuasive and the rejection is sustained.
Regarding the rejections based on 35 USC 102
The applicant’s amendments to the claims have overcome the rejection and the rejection has been withdrawn.
Regarding the rejections based on 35 USC 103
The applicant’s arguments in light of the amendments to the claims have been considered but are moot in light of the new grounds of rejection necessitated by the applicant’s amendments.
The examiner does point out that in Burris the customer has initiated a return, but has not actually returned the product as of yet, so the prediction is indeed as to whether the customer will actually return the product. For example, in [0057] the trained ML algorithm predicts the likelihood of the user returning or failing to go forward with the return as well as predicts other aspects of the potential return.
Conclusion
Applicant amendment(s) necessitated any new grounds of rejection set forth in this Office Action. Therefore, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any extension fee pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the date of this final action.
Any inquiry of a general nature or relating to the status of this application or concerning this communication or earlier communications from the Examiner should be directed to Luis A. Brown whose telephone number is 571.270.1394. The Examiner can normally be reached on Monday-Friday 8:30am-5:00pm EST. If attempts to reach the examiner by telephone are unsuccessful, the Examiner’s supervisor, JESSICA LEMIEUX can be reached at 571.270.3445.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://portal.uspto.gov/external/portal/pair . Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866.217.9197 (toll-free).
Any response to this action should be mailed to:
Commissioner of Patents and Trademarks
Washington, D.C. 20231
or faxed to 571-273-8300.
Hand delivered responses should be brought to the United States Patent and Trademark Office Customer Service Window:
Randolph Building
401 Dulany Street
Alexandria, VA 22314.
/LUIS A BROWN/Primary Examiner, Art Unit 3626
Prosecution Insights