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 .
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 21-40 is/are rejected under 35 U.S.C. 101 because they are directed to an abstract idea without significantly more.
Step 1:
According to the first part of the analysis, in the instant case, claims 21-27 are directed to a method, claims 28-34 are directed to a system, and claims 35-40 are directed to a one or more non-transitory storage media. Thus, each of the claims falls within one of the four statutory categories (i.e., process, machine, manufacture, or composition of matter).
Regarding claim 21,
Step 2A – Prong One: Claim 21 recites:
A method comprising:
performing the following for one or more cases with missing data in a computer-based reasoning model:
determining imputation order information for the one or more cases with missing data based at least in part on:
determining numbers of features that need imputed data for each of the one or more cases with missing data; and
determining the imputation order information based on the number of features that need imputed data for each of the one or more cases with missing data; (Ranking and ordering data items can be performed mentally. This is understood to be a recitation of a mental process.)
for each particular case of the one or more cases with missing data in order based on the determined imputation order information:
determining imputed data for a missing field of the particular case based on the particular case, and an imputation model, and the missing fields in the particular case; (Statistical/algorithmic prediction is a mathematical calculation. This is understood to be a recitation of a mathematical concept.)
modifying the particular case with the imputed data, wherein the modified particular case becomes part of the computer-based reasoning model in place of the original particular case to create an updated computer-based reasoning model; (Updating stored information is a mental act. This is understood to be a recitation of a mental process.)
Step 2A – Prong Two:
causing, with a control system, control of a system with the updated computer-based reasoning model, (Links the abstract idea to use of a generic "control system" in its technological environment. See MPEP § 2106.05(h).)
wherein the method is performed by one or more computing devices. (Generic instruction to implement the abstract idea on conventional computers. See MPEP § 2106.05(f).)
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because, when considered individually and in combination, they do not add an inventive concept to the exception. The claim recites generic computing devices and a control system executing routine data-imputation and model-update steps, which links the abstract idea to a particular technological environment without any technological improvement or inventive concept.
Regarding claims 28 and 35,
These claims are similar in scope to claim 21 and are rejected under similar rationale. The systems, processors, and non-transitory storage media are also generic computing components.
Claims 28 and 35 are ineligible.
Regarding Claim 22
Step 2A Prong 1:
The method of claim 21, wherein causing control of the system comprises:
receiving a request for an action to take in the system, including a context for the system; (Interpreting information and identifying context can be done mentally. This is understood to be a recitation of a mental process.)
determining the action to take based at least in part on the context for the system and the updated computer-based reasoning model; (Selecting an action via rules or algorithms is a judgment. This is understood to be a recitation of a mental process.)
Step 2A Prong 2:
causing the control system to perform the determined action in the system. (Links the abstract idea to execution by a generic "control system" in its technological environment. See MPEP § 2106.05(h))
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because, when considered individually and in combination, they do not add an inventive concept to the exception. The claim recites a generic control system executing the chosen action, which links the abstract idea to a technological environment without any technological improvement or inventive concept.
Regarding claims 29 and 36,
These claims are similar in scope to claim 22 and are rejected under similar rationale. The systems, processors, and non-transitory storage media are also generic computing components.
Claims 29 and 36 are ineligible.
Regarding claim 23
Step 2A Prong 1
The method of claim 21, wherein determining the imputation order information based on the number of features that need imputed data for each of the one or more cases with missing data comprises:
determining one or more particular cases with a lowest number of missing fields to impute; (Selecting records according to a count can be done by mental comparison. This is understood to be a recitation of a mental process.)
determining the imputation order information to comprise first imputing data for the one or more particular cases with the lowest number of missing fields to impute. (Deciding an execution order is a judgment that can be performed mentally. This is understood to be a recitation of a mental process.)
Step 2A Prong 2:
The claim does not include additional elements, when considered separately and in combination, that integrate the judicial exception into a practical application.
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because, when considered individually and in combination, it does not add significantly more (also known as an inventive concept) to the exception. The claim is directed to a mental process of selecting and prioritizing cases with the fewest missing fields, without any technological improvement or inventive concept.
Regarding claims 30 and 37,
These claims are similar in scope to claim 23 and are rejected under similar rationale. The systems, processors, and non-transitory storage media are also generic computing components.
Claims 30 and 37 are ineligible.
Regarding claim 24
Step 2A Prong 1:
The method of claim 21, wherein determining the imputation order information based on the number of features that need imputed data for each of the one or more cases with missing data comprises:
determining two or more particular cases with a lowest number of missing fields to impute; (Selecting records by comparing counts can be done mentally. This is understood to be a recitation of a mental process.)
determining for which one or more cases of the two or more particular cases with the lowest number of missing fields to impute data for the missing fields based at least in part on: (Deciding which case(s) to process first is a judgment. This is understood to be a recitation of a mental process.)
determining which of the two or more particular cases has a highest certainty score among the two or more particular cases, (Picking the maximum value constitutes mental evaluation. This is understood to be a recitation of a mental process.)
wherein the certainty score is determined by a certainty function associated with: removing the case from the computer-based reasoning model;
adding the case back into the computer-based reasoning model, (The certainty score is generated via a mathematical function/ calculation. This is understood to be a recitation of a mathematical concept.)
wherein the certainty function is associated with a certainty that a particular set of data fits a model. (Describes a mathematical relationship. This is understood to be a recitation of a mathematical concept.)
Step 2A Prong 2:
The claim does not include additional elements, when considered separately and in combination, that integrate the judicial exception into a practical application.
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because, when considered individually and in combination, it does not add significantly more (also known as an inventive concept) to the exception. The claim is directed to a mental process of ranking cases using a mathematically derived information-distortion score, without any technological improvement or inventive concept.
Regarding claims 31 and 38,
These claims are similar in scope to claim 24 and are rejected under similar rationale. The systems, processors, and non-transitory storage media are also generic computing components.
Claims 31 and 38 are ineligible.
Regarding claim 25
Step 2A Prong 1:
The method of claim 21, wherein determining the imputation order information based on the number of features that need imputed data for each of the one or more cases with missing data comprises:
determining two or more particular cases with a lowest number of missing fields to impute; (Comparing counts and selecting cases can be done mentally. This is understood to be a recitation of a mental process.)
determining for which one or more cases of the two or more particular cases with the lowest number of missing fields to impute data for the missing fields based at least in part on: (Deciding which case(s) to process first is a judgment. This is understood to be a recitation of a mental process.)
determining for which one or more cases of the two or more particular cases with the lowest number of missing fields to impute data for the missing fields based at least in part on: (Deciding which case(s) to process first is a judgment. This is understood to be a recitation of a mental process.)
determining which of the two or more particular cases has a highest certainty score among the two or more particular cases, (Picking the maximum value constitutes mental evaluation. This is understood to be a recitation of a mental process.)
wherein the certainty score is determined by a certainty function associated with: removing the case from the computer-based reasoning model;
adding the case back into the computer-based reasoning model, (The certainty score is generated via a mathematical function/ calculation. This is understood to be a recitation of a mathematical concept.)
wherein the certainty function is associated with how much information a point distorts the model. (Expresses a mathematical relationship. This is understood to be a recitation of a mathematical concept.)
Step 2A Prong 2:
The claim does not include additional elements, when considered separately and in combination, that integrate the judicial exception into a practical application.
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because, when considered individually and in combination, it does not add significantly more (also known as an inventive concept) to the exception. The claim is directed to a mental process of ranking cases using a mathematically derived information-distortion score, without any technological improvement or inventive concept.
Regarding claims 32 and 39,
These claims are similar in scope to claim 25 and are rejected under similar rationale. The systems, processors, and non-transitory storage media are also generic computing components.
Claims 32 and 39 are ineligible.
Regarding claim 26
Step 2A Prong 1:
The method of claim 21, wherein determining the imputation order information based on the number of features that need imputed data for each of the one or more cases with missing data comprises:
determining two or more particular cases with a lowest number of missing fields to impute; (Comparing counts and selecting cases can be done mentally. This is understood to be a recitation of a mental process.)
determining for which one or more cases of the two or more particular cases with the lowest number of missing fields to impute data for the missing fields based at least in part on: (Deciding which case(s) to process first is a judgment. This is understood to be a recitation of a mental process.)
determining for which one or more cases of the two or more particular cases with the lowest number of missing fields to impute data for the missing fields based at least in part on: (Deciding which case(s) to process first is a judgment. This is understood to be a recitation of a mental process.)
determining which of the two or more particular cases has a highest certainty score among the two or more particular cases, (Picking the maximum value constitutes mental evaluation. This is understood to be a recitation of a mental process.)
wherein the certainty score is determined by a certainty function associated with: removing the case from the computer-based reasoning model;
adding the case back into the computer-based reasoning model, (The certainty score is generated via a mathematical function/ calculation. This is understood to be a recitation of a mathematical concept.)
wherein the certainty function is associated with information required to describe the position of the point in question relative to existing points. (Defines an information measure / mathematical relationship. This is understood to be a recitation of a mathematical concept.)
Step 2A Prong 2:
The claim does not include additional elements, when considered separately and in combination, that integrate the judicial exception into a practical application.
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because, when considered individually and in combination, it does not add significantly more (also known as an inventive concept) to the exception. The claim is directed to a mental process of prioritizing cases using a mathematically derived positional-information score, without any technological improvement or inventive concept.
Regarding claims 33 and 40,
These claims are similar in scope to claim 26 and are rejected under similar rationale. The systems, processors, and non-transitory storage media are also generic computing components.
Claims 33 and 40 are ineligible.
Regarding claim 27
Step 2A Prong 1:
The method of claim 21, wherein determining imputed data for the missing field comprises:
determining the imputed data based on a machine learning model for the computer-based reasoning model's data, wherein the machine learning model for the computer-based reasoning model's data has been trained using the data in the computer-based reasoning model; (Generation of predicted values by statistical/algorithmic calculation. This is understood to be a recitation of a mathematical concept.)
and the method further comprises:
determining an update to the machine learning model based on the updated computer based reasoning model. (Recalculating model parameters via algorithmic computation. This is understood to be a recitation of a mathematical concept.)
Step 2A Prong 2:
The claim does not include additional elements, when considered separately and in combination, that integrate the judicial exception into a practical application.
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because, when considered individually and in combination, it does not add significantly more (also known as an inventive concept) to the exception. The claim recites generic use of a machine-learning model to predict missing data and retrain the model, without any technological improvement or inventive concept.
Regarding claim 34,
This claim is similar in scope to claim 27 and is rejected under similar rationale. The systems, processors, and non-transitory storage media are also generic computing components.
Claim 34 is ineligible.
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.
Claim(s) 21-23, 27, 28-30, and 34-37 is/are rejected under 35 U.S.C. 103 as being unpatentable over Lee et al. (US Pub. No. 2019/0303471, published Oct. 2019, hereinafter “Lee”) in view of Li et al. (US Patent No. 10,387,419, published Aug. 2019, hereinafter “Li”), further in view of Chu et al. (US Patent No. 9,443,194, published Sept. 2016, hereinafter “Chu”) and Zoll et al. (US Pub. No. 2018/0081914, published March 2018, hereinafter “Zoll”).
Regarding claim 21, Lee teaches a method comprising:
performing the following for one or more cases with missing data in a computer-based reasoning model (Lee: "The process 600 represents a computer-implemented method of performing the missing data value imputation described herein. At block 602, the process 600 receives, in user input, data values of an expected input set of data values, where at least one data value of the expected input set of data values is missing from the user input." Paragraph [0109], p. 20) (Note: Lee establishes a computer method and the case with missing data.):
for each particular case of the one or more cases with missing data in order based on the determined imputation order information:
determining imputed data for a missing field of the particular case based on the particular case, and an imputation model, and the missing fields in the particular case (Lee: "At block 724, the process 700 imputes the missing data value from the initial estimate and centroids of the boundary and support clusters."- Paragraph [0122], p. 21) (Note: Lee uses the selected record, the clustering model, and its own missing field to compute the value.);
wherein the method is performed by one or more computing devices (Lee: “A computing device_1 102 through a computing device_N 104 communicate via a network 106 with several other devices. The other devices include a server_1 108 through a server_M 110. A database 112 provides shared storage within the system 100.” – Paragraph [0066], p. 16) (Note: Discloses multiple computing devices and networked servers executing the method.).
Lee fails to explicitly teach wherein determining imputation order information for the one or more cases with missing data based at least in part on: determining numbers of features that need imputed data for each of the one or more cases with missing data; and determining the imputation order information based on the number of features that need imputed data for each of the one or more cases with missing data.
However, analogous to the field of the claimed invention, Li teaches:
determining imputation order information for the one or more cases with missing data based at least in part on:
determining numbers of features that need imputed data for each of the one or more cases with missing data (Li: "In some example implementations, the processor may select a target record as a record with the fewest number of missing values."- p. 12, col. 4, lines 1-2) (Note: Li counts missing fields per record and identifies the record with the smallest count.); and
determining the imputation order information based on the number of features that need imputed data for each of the one or more cases with missing data (Li: "If the resultant filtered dataset includes zero (0) records, the processor may filter the dataset to include two (2) missing value. This process may continue until the resultant filtered dataset includes at least one (1) record."- p. 12, col. 4, lines 4-7) (Note: By incrementally increasing the missing value threshold step by step, Li creates the ordered sequence in which records are processed.);
Lee and Li are analogous art to the present invention because they both address imputation of missing values in data records, focusing on iteratively completing cases so the finished data can train or update predictive models for improved analytic performance. 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 Lee (for record-by-record, clustering-based imputation that updates the model each cycle) and Li (for ordering records by the fewest-missing-fields heuristic and expanding the threshold when no records remain) in order to achieve faster convergence and greater early-stage accuracy. One of ordinary skill in the art would have been motivated to make such a combination because it would allow the iterative imputation engine to process the most information-rich cases first, thereby yielding predictable gains in both speed and accuracy, as suggested by Li: "the processor may select a target record as a record with the fewest number of missing values" (Li, p. 12, col. 4, lines 1-2).
The combination of Lee and Li fails to explicitly teach modifying the particular case with the imputed data, wherein the modified particular case becomes part of the computer-based reasoning model in place of the original particular case to create an updated computer-based reasoning model.
However, analogous to the field of the claimed invention, Chu teaches:
modifying the particular case with the imputed data, wherein the modified particular case becomes part of the computer-based reasoning model in place of the original particular case to create an updated computer-based reasoning model (Chu, “An Expectation Maximization (EM) algorithm is an iterative technique that alternates between steps (1) and (2) until the process converges on stable estimates, where step (1) estimates the model parameters based on the current data set, and step (2) imputes the missing values based on those estimated parameters to update the data set. Then the fill-in data set is used to re-estimate the parameters. Typically, the EM algorithm is used under a multivariate normal model and missing values are imputed based on a regression model. Moreover, the EM algorithm is an iterative process that requires many data passes.” - p. 18, col. 2, lines 4-10; “Finally, the complete data sets for all possible predictor variables are used to build any models for prediction, discovery, and interpretation of relationships between the target variable and a set of the predictor variables.” - p. 19, col. 4, lines 30-34) (Note: Chu's EM loop writes each newly imputed value back into the working data set before the next iteration, so the imputed record replaces the original. After all records are completed, Chu uses the "complete data sets" to train fresh predictive models, demonstrating that the modified cases now create an updated computer-based reasoning model.);
Chu is analogous art to the present invention because it teaches an expectation maximization (EM) loop that rewrites each record with its newly imputed value and immediately re-estimates model parameters on the evolving data set, thus producing an updated reasoning model. 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 Lee (for record-by-record, clustering based imputation that updates the model each cycle) and Li (for ordering records by the fewest missing-fields heuristic and expanding the threshold when no records remain) with Chu's EM methodology (for continuously refining the predictive model as soon as each record is completed) in order to achieve faster convergence and improved final accuracy from a single, unified process. One of ordinary skill in the art would have been motivated to make such a combination because Chu teaches that its EM algorithm " imputes the missing values based on those estimated parameters to update the data set. Then the fill-in data set is used to re-estimate the parameters." (Chu, p. 18, col. 2, lines 4-10).
The combination of Lee, Li, and Chu fails to explicitly teach causing, with a control system, control of a system with the updated computer-based reasoning model.
However, analogous to the field of the claimed invention, Zoll teaches:
causing, with a control system, control of a system with the updated computer-based reasoning model (Zoll: "For example, the predictive models 122 can be applied to perform health monitoring 124 for the database cluster 114."- Paragraph [0027], p.18; "Once faults are detected, automatic probabilistic-based problem diagnosis can be invoked, where the system infers the type of the underlying problems and their root causes from the set of faulty readings. Users will be advised with a set of recommended correction actions to alleviate the reported problem(s)."- Paragraph [0068], p. 22) (Note: Zoll shows the model deployed in an automated monitoring/diagnosis loop, which is "control” of a system.),
Zoll is analogous art to the present invention because it addresses the post-imputation use of predictive models to monitor and control a live database cluster, thereby completing the same overall workflow that Lee, Li, and Chu initiate (data completion, model updating, and subsequent operational application). 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 Lee, Li, and Chu (which collectively yield a rapidly converging, continuously retrained imputation engine) with Zoll's deployment step (which feeds the completed-data model into an automated health monitor loop) in order to obtain an end-to-end system that both produces higher-quality models and immediately applies them to real-time system management. One of ordinary skill in the art would have been motivated to make such a combination because Zoll teaches that "the predictive models 122 can be applied to perform health monitoring 124 for the database cluster 114 automatic probabilistic-based problem diagnosis can be invoked Users will be advised with a set of recommended correction actions to alleviate the reported problem(s)" (Zoll, Paragraphs [0027] and [0068]).
Claims 28 and 35 incorporate substantively all the limitations of claim 21 in a system and a non-transitory storage media, and are rejected on similar grounds as above.
Regarding claim 22, the combination of Lee, Li, Chu, and Zoll teaches the method of claim 21, wherein causing control of the system comprises:
receiving a request for an action to take in the system, including a context for the system (Zoll: "the components periodically sample wide variety of key measurements from the monitored system, which will then analyze the stream of observed data in reference to established base models."- Paragraph [0068], p. 22) (Note: The sampled measurement stream is the context, and its arrival at the health-advisor engine constitutes the system's internal request that the engine take diagnostic action.);
determining the action to take based at least in part on the context for the system and the updated computer-based reasoning model (Zoll: "Once faults are detected, automatic probabilistic-based problem diagnosis can be invoked, where the system infers the type of the underlying problems and their root causes from the set of faulty readings."- Paragraph [0068], p. 22) (Note: The engine uses the updated predictive model together with the incoming context (faulty readings) to decide which corrective action is appropriate.);
causing the control system to perform the determined action in the system (Zoll: "Users will be advised with a set of recommended correction actions to alleviate the reported problem(s)."- Paragraph [0068], p. 22) (Note: Emitting health alerts / recommended corrections is the control system performing the determined action.).
Claims 29 and 36 are similar to claim 22, hence similarly rejected.
Regarding claim 23, the combination of Lee, Li, Chu, and Zoll teaches the method of claim 21, wherein determining the imputation order information based on the number of features that need imputed data for each of the one or more cases with missing data comprises:
determining one or more particular cases with a lowest number of missing fields to impute (Li: "The processor may select a target record as a record with the fewest number of missing values."- p. 12, col. 4, lines 1-2) (Note: Li identifies the record(s) having the lowest count of missing fields.);
determining the imputation order information to comprise first imputing data for the one or more particular cases with the lowest number of missing fields to impute (Li: "If the resultant filtered dataset includes zero (0) records, the processor may filter the dataset to include two (2) missing value. This process may continue until the resultant filtered dataset includes at least one (1) record."- p. 12, col. 4, lines 4-7) (Note: By progressively widening the threshold of missing fields. Li establishes an ordered queue in which the records with the fewest missing fields are imputed first.).
Claims 30 and 37 are similar to claim 23, hence similarly rejected.
Regarding claim 27, the combination of Lee, Li, Chu, and Zoll teaches the method of claim 21, wherein determining imputed data for the missing field comprises:
determining the imputed data based on a machine learning model for the computer-based reasoning model's data, wherein the machine learning model for the computer-based reasoning model's data has been trained using the data in the computer-based reasoning model (Chu: "First, for each predictor variable that has missing values, imputation models based only on the target variable are built independently on different data sources and on different machines using the Map functions."- p. 19, col. 4, lines 15-18; "In Selection 440, each Mapper422, 424, 426 imputes the missing value for each of the one or more predictor variables based on a data source, one or more ensemble models and a selected imputation strategy to output the completed data."- p. 20, col. 6, lines 58-61; "Finally, the complete data sets for all possible predictor variables are used to build any models for prediction, discovery, and interpretation of relationships between the target variable and a set of the predictor variables."- p. 19, col. 4, lines 30-34) (Note: Chu trains imputation/ensemble models on the existing data, then uses those models to compute each missing value.);
and the method further comprises: determining an update to the machine learning model based on the updated computer-based reasoning model (Chu: "An Expectation Maximization (EM) algorithm is an iterative technique that alternates between steps (1) and (2) until the process converges on stable estimates, where step (1) estimates the model parameters based on the current dataset, and step (2) imputes the missing values based on those estimated parameters to update the data set. Then the fill-in data set is used to re-estimate the parameters."- p. 18, col. 2, II. 4-10) (Note: Chu's EM loop re-estimates (re-trains) the model parameters after each round of imputation, thereby updating the machine-learning model with the newly modified reasoning-model data.).
Claim 34 is similar to claim 27, hence similarly rejected.
Claim(s) 24-26, 31-33, and 38-40 is/are rejected under 35 U.S.C. 103 as being unpatentable over Lee, Li, Chu, and Zoll as applied to claims 21, 28, and 35 above, and further in view of Bak et al. (NPL: “Data Driven Estimation of Imputation Error – A Strategy for Imputation with a Reject Option”, published Oct. 2017, hereinafter “Bak”).
Regarding claim 24, the combination of Lee, Li, Chu, and Zoll teaches the method of claim 21, wherein determining the imputation order information based on the number of features that need imputed data for each of the one or more cases with missing data comprises:
determining two or more particular cases with a lowest number of missing fields to impute (Li: "The processor may select a target record as a record with the fewest number of missing values... If the resultant filtered dataset includes zero (0) records, the processor may filter the dataset to include two (2) missing value. This process may continue until the resultant filtered dataset includes at least one (1) record."- p. 12, col. 4, lines 1-7) (Note: Li first isolates the records that have the lowest missing-field count and processes that group before any record with a higher count.);
The combination of Lee, Li, Chu, and Zoll fails to explicitly teach determining for which one or more cases of the two or more particular cases with the lowest number of missing fields to impute data for the missing fields based at least in part on: determining which of the two or more particular cases has a highest certainty score among the two or more particular cases, wherein the certainty score is determined by a certainty function associated with: removing the case from the computer-based reasoning model; adding the case back into the computer-based reasoning model, wherein the certainty function is associated with a certainty that a particular set of data fits a model.
However, analogous to the field of the claimed invention, Bak teaches:
determining for which one or more cases of the two or more particular cases with the lowest number of missing fields to impute data for the missing fields based at least in part on:
determining which of the two or more particular cases has a highest certainty score among the two or more particular cases (Bak: "To help make decisions about which records should be imputed, we propose to use a machine learning approach to estimate the imputation error for each case with missing data."- p. 1, Abstract) (Note: Bak states that the per-record certainty (low estimated error is high certainty) is used to decide which of the candidate records is selected for imputation.), wherein the certainty score is determined by a certainty function associated with:
removing the case from the computer-based reasoning model;
adding the case back into the computer-based reasoning model,
wherein the certainty function is associated with a certainty that a particular set of data fits a model (Bak: "Each complete case data point is used to simulate all missingness patterns present in the data. Each of these Ncom . M "new" data points are then imputed, in a leave-one-out (LOO) [18] approach based on the other (Ncom- 1) complete cases. At this point we then have Ncom . M data points consisting of simulated missing data in the complete cases. These are then compared to the true value of the Ncom complete cases to determine the errors for all missingness patterns for each complete case."- p. 3, Methods) (Note: The LOO step removes the candidate record from the model, imputes it with the remaining data, then adds it back and measures the fit. This computed error (or its inverse) is the certainty score.).
Bak is analogous art to the present invention because it addresses record-level decision making during imputation, focusing on estimating a certainty score for each case so that users can choose which records to impute first, precisely the tie-breaking step that follows Li's fewest missing-fields grouping and precedes Lee/Chu's iterative model update, and Zoll's deployment context. 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 Li (for isolating the lowest-missing subset), Lee and Chu (for looping through records and re-training the model after each imputation), and Bak (for ranking those tied records by per-case certainty or estimated error) in order to obtain a more reliable, information-guided imputation schedule that predictably reduces overall error. One of ordinary skill in the art would have been motivated to make such a combination because Bak teaches that "to help make decisions about which records should be imputed, we propose to use a machine learning approach to estimate the imputation error for each case with missing data" (Bak, p. 1, Abstract).
Claims 31 and 38 are similar to claim 24, hence similarly rejected.
Regarding claim 25, the combination of Lee, Li, Chu, and Zoll teaches the method of claim 21, wherein determining the imputation order information based on the number of features that need imputed data for each of the one or more cases with missing data comprises:
determining two or more particular cases with a lowest number of missing fields to impute (Li: "The processor may select a target record as a record with the fewest number of missing values... If the resultant filtered dataset includes zero (0) records, the processor may filter the dataset to include two (2) missing value. This process may continue until the resultant filtered dataset includes at least one (1) record."- p. 12, col. 4, lines 1-7) (Note: Li first isolates the records that have the lowest missing-field count and processes that group before any record with a higher count.);
The combination of Lee, Li, Chu, and Zoll fails to explicitly teach determining for which one or more cases of the two or more particular cases with the lowest number of missing fields to impute data for the missing fields based at least in part on: determining which of the two or more particular cases has a highest certainty score among the two or more particular cases, wherein the certainty score is determined by a certainty function associated with: removing the case from the computer-based reasoning model; adding the case back into the computer-based reasoning model, wherein the certainty function is associated with how much information a point distorts the model.
However, analogous to the field of the claimed invention, Bak teaches:
determining for which one or more cases of the two or more particular cases with the lowest number of missing fields to impute data for the missing fields based at least in part on:
determining which of the two or more particular cases has a highest certainty score among the two or more particular cases (Bak: "To help make decisions about which records should be imputed, we propose to use a machine learning approach to estimate the imputation error for each case with missing data."- p. 1, Abstract) (Note: Bak states that the per-record certainty (low estimated error is high certainty) is used to decide which of the candidate records is selected for imputation.), wherein the certainty score is determined by a certainty function associated with:
removing the case from the computer-based reasoning model;
adding the case back into the computer-based reasoning model,
wherein the certainty function is associated with how much information a point distorts the model (Bak: "Each complete case data point is used to simulate all missingness patterns present in the data. Each of these Ncom . M "new" data points are then imputed, in a leave one-out (LOO) [18] approach based on the other (Ncom- 1) complete cases. At this point we then have Ncom . Mdata points consisting of simulated missing data in the complete cases. These are then compared to the true value of the Ncom complete cases to determine the errors for all missingness patterns for each complete case."- page 3, methods; "Our method, however focuses on the error and variance of the actual imputation algorithm rather than estimating the variance of the variables with missing values."- p. 2, Introduction) (Note: Bak's leave-one-out simulation removes a complete case, injects the missing-value pattern, adds it back through imputation, and measures the resulting error/variance. That measured error quantifies how much the point distorts the learned model (higher error is greater distortion)).
Claims 32 and 39 are similar to claim 25, hence similarly rejected.
Regarding claim 26, the combination of Lee, Li, Chu, and Zoll teaches the method of claim 21, wherein determining the imputation order information based on the number of features that need imputed data for each of the one or more cases with missing data comprises:
determining two or more particular cases with a lowest number of missing fields to impute (Li: "The processor may select a target record as a record with the fewest number of missing values... If the resultant filtered dataset includes zero (0) records, the processor may filter the dataset to include two (2) missing value. This process may continue until the resultant filtered dataset includes at least one (1) record."- p. 12, col. 4, lines 1-7) (Note: Li first isolates the records that have the lowest missing-field count and processes that group before any record with a higher count.);
The combination of Lee, Li, Chu, and Zoll fails to explicitly teach determining for which one or more cases of the two or more particular cases with the lowest number of missing fields to impute data for the missing fields based at least in part on: determining which of the two or more particular cases has a highest certainty score among the two or more particular cases, wherein the certainty score is determined by a certainty function associated with: removing the case from the computer-based reasoning model; adding the case back into the computer-based reasoning model, wherein the certainty function is associated with information required to describe the position of the point in question relative to existing points.
However, analogous to the field of the claimed invention, Bak teaches:
determining for which one or more cases of the two or more particular cases with the lowest number of missing fields to impute data for the missing fields based at least in part on:
determining which of the two or more particular cases has a highest certainty score among the two or more particular cases (Bak: "To help make decisions about which records should be imputed, we propose to use a machine learning approach to estimate the imputation error for each case with missing data."- p. 1, Abstract) (Note: Bak states that the per-record certainty (low estimated error is high certainty) is used to decide which of the candidate records is selected for imputation.), wherein the certainty score is determined by a certainty function associated with:
removing the case from the computer-based reasoning model;
adding the case back into the computer-based reasoning model,
wherein the certainty function is associated with information required to describe the position of the point in question relative to existing points (Bak: "As error measure we used the Euclidean distance between imputed and true data points in the 2D PCA space... where II µn O (x- xn ) || calculates the Euclidean distance only from present features, and g and d are scale parameters."- p. 4, Methods) (Note: Bak's leave-one-out loop first removes the target record, imputes its missing values with a model trained on the remaining data, then adds the record back and computes the Euclidean distance between the imputed point and its reference position. That distance is the information required to describe how far (i.e., how much the point distorts) the record's position relative to the existing points (smaller distance ” higher certainty score)).
Claims 33 and 40 are similar to claim 26, hence similarly rejected.
Response to Arguments
Applicant's arguments, see pp. 1-5 of Remarks, filed 8 January 2026 have been fully considered but they are not persuasive. Applicant argues that claim 21 is directed to patent eligible subject matter, and states that the Office Action analyzes claim language at a high level of generality without considering the interactions between elements of the claim and that the Office Action does not give sufficient weight to the technical solutions to technical problems described in the Specification (See pp. 2 of Remarks). Examiner respectfully disagrees. In the rejection of claim 21, the limitation regarding “determining imputed data for a missing field of the particular case based on the particular case, and an imputation model, and the missing fields in the particular case” is, under broadest reasonable interpretation in light of the Specification, a mathematical calculation which falls within the mathematical concepts grouping of abstract ideas (See MPEP 2106.04(a)(2)(C)). The statistical/algorithmic determination of imputed data for a missing field based on the particular case, an imputation model, and the missing fields requires statistical models (See Specification at [0028] – “After determining 130 which cases to impute data and / or the order in which to impute data, the imputed data is determined 140 based on the case with the missing data and the imputation model. The imputation model may be any appropriate statistical or other machine learning model.”) and calculation of a conviction score (See Specification at [0046] – “The noisiness of a feature can be determined 140 as a conviction score, in some embodiments, as a local noisiness and / or relative noisiness… local noisiness can be determined by looking for the minimum of Y… Relative noisiness may be determined based on the ratio of Z to W.” The Specification at [0047] describes further calculations for conviction score, and in [0048] describes that the calculations are provided to a human operator for review and subsequent decisions).
Even when viewed as a whole, the claim recites limitations regarding mental processes (ranking and ordering data and updating stored information) and mathematical concepts (use of statistical/algorithmic models for determining imputed data of a missing field of a case in the determined imputation order) which fall within the abstract ideas groupings (See MPEP 2106.04(a)(2)). The claim recites an additional element regarding “causing, with a control system, control of a system…” which is a mere attempt to generally link the judicial exception to a technological environment (See MPEP 2106.05(h)). The claim recites a further additional element regarding “wherein the method is performed by one or more computing devices” which are mere instructions to apply the judicial exception on a generic computer (See MPEP 2106.05(f)). The additional elements of the claim, even when viewed in combination, fails to integrate the abstract idea into a practical application (See MPEP 2106.04(d)(I) which describes identified limitations that did not integrate a judicial exception into a practical application, such as “Merely reciting the words "apply it" (or an equivalent) with the judicial exception, or merely including instructions to implement an abstract idea on a computer, or merely using a computer as a tool to perform an abstract idea…”). The control systems and computing devices recited in claim 21 are recited at a high level of generality and impose no meaningful limitations on the claim. Thus, the additional elements of the claim, even when viewed in combination, fail to amount to significantly more than the judicial exception.
MPEP 2106.04(d)(1) states “if the specification sets forth an improvement in technology, the claim must be evaluated to ensure that the claim itself reflects the disclosed improvement. That is, the claim includes the components or steps of the invention that provide the improvement described in the specification.” Claim 21 fails to reflect the purported improvements (see pp. 2-3 of Remarks) set forth in the Specification at [0003], [0015], and [0029-0031]. The claims do not recite any sensors (such as faulty, malfunctioning, or later-added sensors of oil pumps), batching of cases (such that larger batches improve performance and reduce computational spend, or smaller batches increasing accuracy of imputation of data), retraining the model with updated data, or a supervised machine learning model. Rather, the claim recites “determining imputation order information for the one or more cases with missing data based at least in part on: determining the number of features that need imputed data… and determining the imputation order information based on the number of features…” which is a mental process of ranking and ordering data that can be performed in the human mind. The claim further recites “for each particular case of the one or more cases with missing data in order… determining imputed data for a missing field of the particular case…” which is a mathematical concept of statistical/algorithmic determination requiring mathematical calculations (as in [0028] and [0046-0047]). The claim also recites “modifying the particular case with the imputed data…” which is a mental process of updating stored information that can be performed in the human mind. The additional elements of the claim are generic computing components and systems, and thus fail to integrate the judicial exception into a practical application and do not amount to significantly more than the judicial exception.
Applicant's arguments, see pp. 5-11 of Remarks, filed 8 January 2026 have been fully considered but they are not persuasive. Applicant argues on pp. 5-7 of Remarks that none of the cited references teaches the limitation of claim 21 regarding “modifying the particular case with the imputed data…”. Examiner respectfully disagrees and points to Chu at Pg. 18, Col. 2, Lines 4-10 – “An Expectation Maximization (EM) algorithm is an iterative technique that alternates between steps (1) and (2) until the process converges on stable estimates, where step (1) estimates the model parameters based on the current data set, and step (2) imputes the missing values based on those estimated parameters to update the data set. Then the fill-in data set is used to re-estimate the parameters.” and in Pg. 19, Col. 4, Lines 30-34 – “Finally, the complete data sets for all possible predictor variables are used to build any models for prediction, discovery, and interpretation of relationships between the target variable and a set of particular variables” – which teaches modifying the particular case with the imputed data (imputes missing values based on estimated parameters to update the data set, thus modifying, or updating, the particular case with the imputed data), wherein the modified particular case becomes part of the computer-based reasoning model in place of the original particular case to create an updated computer-based reasoning model (the complete, updated data sets are used to build any models, thus the modified particular case becomes part of the computer-based reasoning model to create an updated computer-based reasoning model).
Applicant argues on pp. 8-9 that Chu fails to teach sequential model updates being performed in order. Examiner respectfully disagrees, as Chu is not relied upon for determining the imputation order information. Li teaches "In some example implementations, the processor may select a target record as a record with the fewest number of missing values."- p. 12, col. 4, lines 1-2 and that "If the resultant filtered dataset includes zero (0) records, the processor may filter the dataset to include two (2) missing value. This process may continue until the resultant filtered dataset includes at least one (1) record."- p. 12, col. 4, lines 4-7 – which describes selecting records, or cases, in a certain order to be processed based on the number of features that need imputed data for each of the one or more records, or cases, with missing data. Applicant further argues on pp. 8-9 of Remarks that Chu’s model is static while multiple missing values are imputed in a single iteration. Examiner respectfully disagrees and points to Chu at Pg. 18, Col. 2, Lines 21-32 – “Another technique imputes missing values while building a predictive model. A population of solutions is created using the data set with missing values, where each solution includes parameters of the model and the missing values. Each of the solutions in a population is checked for fitness. After the fitness is checked, the solutions in a population are genetically evolved to establish a successive population of solutions. The process of evolving and checking fitness is continued until a stopping criterion is reached. This technique may need many runs of populations of solutions to reach the stopping criterion.” – which describes imputing values while building a predictive model, where each solution is checked for fitness, and evolving and checking fitness until a stopping criterion is reached, which teaches model evolution.
In response to applicant’s argument that there is no teaching, suggestion, or motivation to combine the references, the examiner recognizes that obviousness may be established by combining or modifying the teachings of the prior art to produce the claimed invention where there is some teaching, suggestion, or motivation to do so found either in the references themselves or in the knowledge generally available to one of ordinary skill in the art. See In re Fine, 837 F.2d 1071, 5 USPQ2d 1596 (Fed. Cir. 1988), In re Jones, 958 F.2d 347, 21 USPQ2d 1941 (Fed. Cir. 1992), and KSR International Co. v. Teleflex, Inc., 550 U.S. 398, 82 USPQ2d 1385 (2007). In this case, it would have been obvious to a person of ordinary skill in the art, before the effective filing date of the claimed invention, to incorporate the modification of particular cases with the imputed data of Chu to the imputation order information and determination of imputed data of Lee and Li in order to achieve faster convergence and improved final accuracy. Doing so would impute missing values based on estimations and utilize the imputed data to update a model (Chu, Pg. 18, Col. 2, Lines 4-10).
Applicant's arguments, see pp. 11-12 of Remarks, filed 8 January 2026 have been fully considered but they are not persuasive. Applicant argues that Bak fails to teach “determining which of the two or more particular cases has a highest certainty score… determined by a certainty function associated with: removing the case from the computer-based reasoning model; [and] adding the case back into the computer-based reasoning model”. Examiner respectfully disagrees. Bak teaches "Each complete case data point is used to simulate all missingness patterns present in the data. Each of these Ncom . M "new" data points are then imputed, in a leave-one-out (LOO) [18] approach based on the other (Ncom- 1) complete cases. At this point we then have Ncom . M data points consisting of simulated missing data in the complete cases. These are then compared to the true value of the Ncom complete cases to determine the errors for all missingness patterns for each complete case."- p. 3, Methods, which teaches a LOO step that removes the candidate record from the model, imputes it with the remaining data, then adds it back and measures the fit. This computed error (or its inverse) is the certainty score.
In response to applicant’s argument that there is no teaching, suggestion, or motivation to combine the references, the examiner recognizes that obviousness may be established by combining or modifying the teachings of the prior art to produce the claimed invention where there is some teaching, suggestion, or motivation to do so found either in the references themselves or in the knowledge generally available to one of ordinary skill in the art. See In re Fine, 837 F.2d 1071, 5 USPQ2d 1596 (Fed. Cir. 1988), In re Jones, 958 F.2d 347, 21 USPQ2d 1941 (Fed. Cir. 1992), and KSR International Co. v. Teleflex, Inc., 550 U.S. 398, 82 USPQ2d 1385 (2007). In this case, it would have been obvious to a person of ordinary skill in the art, before the effective filing date of the claimed invention, to incorporate the certainty score of Bak to the imputation order information of Lee, Li, Chu, and Zoll in order to order imputations based on a certainty. Doing so would help make decisions about which records should be imputed by using a machine learning approach to estimate the imputation error for each case with missing data (Bak, p. 1, Abstract).
Applicant further argues on pp. 10 of Remarks that Bak fails to teach the certainty function is “associated with how much information a point distorts a model”. Examiner respectfully disagrees, and points to Bak at pg. 3, Methods - "Each complete case data point is used to simulate all missingness patterns present in the data. Each of these Ncom . M "new" data points are then imputed, in a leave-one-out (LOO) [18] approach based on the other (Ncom- 1) complete cases. At this point we then have Ncom . M data points consisting of simulated missing data in the complete cases. These are then compared to the true value of the Ncom complete cases to determine the errors for all missingness patterns for each complete case."- which teaches a LOO step that removes the candidate record from the model, imputes it with the remaining data, and determines an error, wherein the measured error quantifies how much the point distorts the learned model. Bak teaches wherein the certainty is associated with how much information a point distorts the model, where greater error, or uncertainty, indicates greater distortion of the point on the model. The claim does not recite that the distortion is determined by entropy change or parameter shift (See pp. 10-11 of Remarks).
Applicant argues on pp. 11 of Remarks that Bak fails to teach wherein the certainty function is “associated with information required to describe the position of the point in question relative to existing points”. Examiner respectfully disagrees and points to Bak at Pg. 4, Methods - "As error measure we used the Euclidean distance between imputed and true data points in the 2D PCA space... where II µn O (x- xn ) || calculates the Euclidean distance only from present features, and g and d are scale parameters."- which teaches computing the Euclidean distance between the imputed point and its reference position. The distance is the information required to describe how far the record's position relative to the existing points (smaller distance provides higher certainty score). Thus, Bak teaches wherein the certainty score is associated with information that describes the position of the point in question relative to existing points (distance between imputed data and true data points in a PCA space). The claim does not recite that the information required to describe the position of the point in question is derived from information-theoretic measures (See pp. 11 of Remarks).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Kaiser et al. (NPL: Algorithm for Missing Values Imputation in Categorical Data with Use of Association Rules, published Nov. 2012) teaches a method for imputation of missing value using association rules. Teaches selection of missing values imputation based on the structure of the data set. Teaches imputation based on confidence and sorting the imputation order based on confidence.
Zhang et al. (NPL: Efficient Missing Data Imputation for Supervised Learning, published July 2010) teaches an expectation maximization-style method for iterative imputation in which each missing attribute-value is iteratively filled using a predictor constructed from known values and predicted values of the missing values in the particular case. Teaches considering imputation ordering for patching up multiple missing attribute values.
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to LOUIS C NYE whose telephone number is 571-272-0636. The examiner can normally be reached Monday - Friday 9:00AM - 5:00PM.
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, MATT ELL can be reached at 571-270-3264. 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.
/LOUIS CHRISTOPHER NYE/Examiner, Art Unit 2141
/MATTHEW ELL/Supervisory Patent Examiner, Art Unit 2141
Prosecution Insights