DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Status of Claims
This is a final office action in response to the amendment filed 26 September 2025. Claims 1-15 have been amended, claims 1-15 remain pending and have been examined.
Response to Amendment
Applicant’s amendment to claims 1-15 has been entered.
Applicant’s amendment is insufficient to overcome the pending 35 U.S.C. 101 rejection. The rejection remains pending and is updated below, as necessitated by amendment.
Applicant’s amendment is sufficient to overcome the pending 35 U.S.C. 103 rejection for claims 1-7. Rejection for claims 8-15 is pending.
Response to Arguments
Applicant’s arguments regarding the 35 U.S.C. 102 and 35 U.S.C. 103 rejection have been fully considered regarding claims 1-7, and are persuasive. Particularly the arguments at page 14 of the Remarks that the prior art of record fails to teach or otherwise disclose the amended claim language requiring a multi-level matrix structure with overlapping employee groups and project groups. An updated prior art search was conducted, as necessitated by amendment. Examiner analyzed amended Claim 1 in view of the prior art of record and the updated prior art search and finds not all claim limitations are explicitly taught nor would one of ordinary skill in the art find it obvious to combine these references with a reasonable expectation of success. The prior art made of record and not relied upon is updated and detailed below. The prior art rejection is respectfully withdrawn.
Regarding Claims 8-15, per the 35 U.S.C. 112 rejection below, when reading the claim language in view of the Specification, all that is required is disclosure of one of the claimed feature extraction/data reduction technique. Primary reference Widanapathirana et al. discloses use of Principal Components of a set of metrics for data analysis in paragraph [0134-0136]. Therefore, the claim amendment is disclosed by the prior art and the 35 U.S.C. 103 rejection of claims 8-15 is proper, maintained, and updated below, as necessitated by amendment.
Applicant’s arguments regarding the 35 U.S.C. 101 rejection have been fully considered, but are not persuasive. Applicant asserts that the amended claims recite specific feature extraction techniques that provide concrete improvements to computer functionality, computer processing efficiency and accuracy by reducing dimensionality through principal component analysis, zero-phase component analysis, and autoencoder techniques, as well as application of multi-level matrix structure and machine learning techniques that solves a technical problem inherent in computer based data processing systems that are integrated into the practical application of project management by solving technological problems of data sparsity, computational inefficiency, and prediction accuracy in computer-based project management systems. Examiner respectfully disagrees.
The focus of Applicant’s invention is not an improvement to computer performance or any underlying technology, instead, the focus us to user generic computer components and data analysis tools to gather, manipulate, and analyze business data to administer, create, or modify project tasks/milestones or identify project risks such as monitoring key performance indicators at each milestone to help make strategic business decisions (see Spec. at [0077]) and predicting project performance for each milestone (see Spec. at [0094]), without significantly more. While the data analysis includes reducing dimensionality using specific data analysis tools such as principal component analysis, machine learning, autoencoder, and zero-phase filtering, these additional elements are simply used as data analysis tools, without improvement to the underlying technology. As a result, they do not integrate the recited abstract idea into a practical application. Therefore, the 35 U.S.C. 101 rejection is proper, maintained, and updated below, as necessitated by amendment.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
Claims 1-15 are rejected under 35 U.S.C. 112(b) as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor regards as the invention.
Claims 1 and 8 include the limitation: “executing feature extraction on the project data and the employee data using Principal Component Analysis (PCA), Zero-phase Component Analysis (ZCA), and autoencoder techniques to generate features that reduce feature space and noise while preserving data variation.” Paragraph [0095] of the Specification states: “One way to derive features based on domain knowledge and data analysis is to aggregate performance data for each individual based on the performance of the projects engaged by the individual. Similarly, the aggregation can be done for each team, organization, and client. Further, the automatic feature derivation techniques include PCA, ZCA and AutoEncoder which help reduce feature space, reduce noises and keep key signals in the data.” See also Spec. at [0050-0053] defining each technique separately, and stating: “If the goal is compression of the original data (because principal components are sorted according to their explained variance), then PCA is the optimal analysis. ZCA is the optimal analysis if the goal is to keep the transformed random vector as similar as possible to the original one. An autoencoder is another technique for dimensionality reduction and noise reduction. An autoencoder uses a type of artificial neural network to learn efficient data coding in an unsupervised manner.”
It is unclear from the claim language whether Applicant is claiming selection and use of PCA, ZCA, or AutoEncoder techniques based on a determination of which technique is fits the data reduction goal, or whether the recited limitations requires data analysis using each of the techniques. When reading the Specification in view of the claim limitation, the disclosure requires selection of one of the named techniques to generate features. As a result, the claim language does not clearly reflect the disclosure. For examining purposes, examiner construes the recited claim language in view of the Specification as requiring use of only one of the claimed techniques for feature generation. Claims 2-7 depend on claim 1, and claims 8-15 depend from claim 8 and inherit the same deficiencies.
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 1-15 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea of collecting data, analyzing it, and outputting results of the collection and analysis for predicting project performance, without significantly more. Independent claim 1 recites a process and independent claim 8 recites a process for performance based project management.
Under Step 1, independent claim 1 recites at least one step or act, including executing feature extraction on the project data and employee data. Independent claim 8 recites at least one step or act, including generating a matrix from project performance data. Thus, the claims fall within one of the statutory categories of invention.
Independent claim 1 recites at least the following limitations:
generating a multi-level matrix from project performance data, wherein: the multi-level matrix is structured with rows representing employee related data across different dimensions and columns representing project related data;
the employee related data comprises individual employee data, team data, and organization data that creates overlapping employee groups;
the project related data comprises individual project data and project group data; and
the overlapping employee groups and project groups create a denser matrix compared to a matrix with only the individual employee data and the individual project data;
for an input of a project comprising project data and employee data, executing feature extraction on the project data and the employee data using Principal Component Analysis (PCA), Zero-phase Component Analysis (ZCA), and autoencoder techniques to generate features that reduce feature space and noise while preserving data variation;
executing a self-profiling algorithm configured with unsupervised machine learning on the generated features to derive clusters and anomalies of the project;
executing a performance monitoring process on the generated features to determine a probability of a key performance indicator value at a milestone; and
executing a supervised machine learning model on the generated features, the derived clusters, the derived anomalies, and the probability of the key performance indicator value at the milestone to generate a predicted performance value of the project.
Independent claim 8 recites at least the following limitations:
generating a matrix from project performance data, the matrix structured in rows of employee related data across different dimensions and columns of project related data, the matrix comprising values indicative of performance scores derived from the project performance data;
for missing ones of the values in the matrix, generating the missing ones of the values in the matrix from a training process;
executing feature extraction on the project performance data using Principal Component Analysis (PCA), Zero-phase Component Analysis (ZCA), and autoencoder techniques to generate features that reduce feature space and noise while preserving data variation;
for an input of a new project: identifying ones of the rows and ones of the columns that match the employee related data in the new project and the project related data of the new project;
aggregating ones of the values corresponding to the identified ones of the rows and the ones of the columns;
ranking employee groups from the employee related data according to the aggregated ones of the values; and
selecting an employee group from the employee related data having a highest rank for execution of the new project.
Under Step 2A Prong One, the limitations of independent claim 1 for generating a multi-level matrix from project performance data, executing feature extraction on project data and employee data, executing a self-profiling algorithm, executing a performance monitoring process on generated features, and executing a supervised machine learning model on the generated features, derived clusters, derived anomalies, probability of the key performance indicator value at the milestone, as drafted, illustrates a process that, under its broadest reasonable interpretation covers performance of the limitation in the mind (data collection, analysis, and result determination). None of the additional elements preclude the steps from practically being performed in the human mind, or by a human using a pen and paper. Specifically, predicting a performance value for a project is a mental process because the claimed prediction is a process that is practically performed in the human mind by a human project data and using “evaluation, judgment, and opinion” to detect whether key performance indicators and project milestones have been met. Therefore, the limitations fall into the mental processes grouping and accordingly the claims recite an abstract idea.
Per the Specification at [0046-0049]: “Project specific data includes the performance of an employee based on the performance of completed projects or the performance at each milestone of completed or ongoing projects.” The limitations of claim 1 additionally fall within certain methods of organizing human activity because the project data includes employee data related to individual employee data, team data, and organization data that creates overlapping employee groups and project groups that are used for feature extraction to determine the probability of a key performance indicator value at a milestone based on monitoring employee performance.
Under Step 2A Prong One, the limitations of independent claim 8 for generating a matrix, generating missing ones of the values in the matrix, executing feature extraction on the project performance data, identifying ones of the rows and ones of the columns that match employee related data and project related data, aggregating ones of the values corresponding to the identified rows and columns, ranking employee groups according to the aggregated values, selecting an employee group form the employee related data having a highest rank, as drafted, illustrates a process that, under its broadest reasonable interpretation covers performance of the limitation in the mind (data collection, analysis, and result determination). None of the additional elements preclude the steps from practically being performed in the human mind, or by a human using a pen and paper. Specifically, selecting a group of employees for execution of a new project is a mental process because the claimed prediction is a process that is practically performed in the human mind by a human project data and using “evaluation, judgment, and opinion” to determine which employee group meets project needs. Therefore, the limitations fall into the mental processes grouping and accordingly the claims recite an abstract idea.
The limitations of claim 8 additionally fall within certain methods of organizing human activity because assigning employees to work on a project for meeting project goals is a form of managing human behavior through assignment of tasks. As a result, the limitations of claim 8 fall within the certain methods of organizing human activities grouping of abstract concepts.
Under Step 2A Prong Two, the judicial exception of claims 1 and 8 are not integrated into a practical application. In particular, the claims only recite a processor, storage device, self-profiling algorithm, principal component analysis, zero-phase component analysis, autoencoder techniques, and a machine learning models for performing the recited steps. These elements are recited at a high level of generality (i.e., as a generic processor performing a generic computer function) and amount to no more than mere instructions to apply the exception using generic computer components. See MPEP 2106.05(f). For example, Applicant’s specification at paragraph [0148] states: “… the operations described above can be performed by hardware, software, or some combination of software and hardware … When performed by software, the methods may be executed by a processor, such as a general purpose computer.” Adding generic computer components to perform generic functions, such as data gathering, performing calculations, and outputting a result would not transform the claim into eligible subject matter. See MPEP 2106.05(h). Accordingly, the additional elements do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea.
The focus of Applicant’s invention is not an improvement to computer performance or any underlying technology, instead, the focus us to user generic computer components and data analysis tools to gather, manipulate, and analyze business data to administer, create, or modify project tasks/milestones or identify project risks such as monitoring key performance indicators at each milestone to help make strategic business decisions (see Spec. at [0077]) and predicting project performance for each milestone (see Spec. at [0094]), without significantly more. While the data analysis includes reducing dimensionality using specific data analysis tools such as principal component analysis, machine learning, autoencoder, and zero-phase filtering, these additional elements are simply used as data analysis tools, without improvement to the underlying technology. As a result, they do not integrate the recited abstract idea into a practical application.
Regarding claim 1, the recitation of “executing a self-profiling algorithm and executing a supervised machine learning model in limitations (b) and (d) merely indicates a field of use or technological environment in which the judicial exception is performed. Although the additional element “executing a self-profiling algorithm configured with unsupervised machine learning” and “executing a supervised machine learning model” limits the identified judicial exception “determining a probability of a key performance value at a milestone” and “generating a predicted performance value of the project” using a self-profiling algorithm comprising unsupervised machine learning and a supervised machine learning model, merely confines the use of the abstract idea to a particular technological environment (machine learning) and thus fails to add an inventive concept to the claims. See MPEP 2106.05(h). Further, the claim does not include any details regarding how the machine learning algorithm and model operates. Even when viewed in combination, these additional elements do not integrate the recited judicial exception into a practical application.
Under Step 2B, the limitations of claims 1 and 8 are at best mere instructions to “apply” the abstract ideas, which cannot provide an inventive concept. See MPEP 2106.05(f). The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract idea into a practical application, the additional elements of a processor and storage device amount to no more than mere instructions to apply the exception using a generic computer component which cannot provide an inventive concept.
Dependent claims 2-7 and 9-15 include the abstract ideas of the independent claims. The limitations of the dependent claims merely narrow the mental process/certain methods of organizing human activity by describing how the data is analyzed to generate a prediction. The limitations of the dependent claims are not integrated into a practical application because none of the additional elements set forth any limitations that meaningfully limit the abstract idea implementation. There are no additional elements that transform the claim into a patent eligible idea by amounting to significantly more. The analysis above applies to all statutory categories of invention. Accordingly, claims 1 through 15 are ineligible under 35 U.S.C. 101.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or non-obviousness.
Claims 8-10 and 14-15 are rejected under 35 U.S.C. 103 as being unpatentable over Widanapathirana et al. (US 2019/0108471) in view of Jersin et al. (US 2019/0197180).
Regarding Amended Claim 8, Widanapathirana et al. discloses a computer-implemented method, comprising: generating a matrix from project performance data, ( … a computer implemented method is described that is performed by a computing system including at least one processor and memory for at least executing instructions. Widanapathirana et al. [para. 0003]… Data inputs 125 may also come from one or more other systems such as finance or scheduling systems, resource planning system, time sheeting, and specialised additional data streams. Widanapathirana et al. [para. 0021-0023, 0131-0139]. … Project metadata may be compared against the segmentation weighting 435 and a project weighting matrix 440 that is specific to the project may be derived. The project weighting matrix 440 helps to account for project-to-project differences that occur in real world projects. Widanapathirana et al. [para. 0078]);
executing feature extraction on the project performance data using Principal Component Analysis (PCA), Zero-phase Component Analysis (ZCA), and autoencoder techniques to generate features that reduce feature space and noise while preserving data variation; (Non-limiting examples of derived metric calculations may for example comprise: … Principal Components of a set of metrics; LDA components of a set of metrics; Normalized value of a metrics; Scaled value of a metric; Binary representations of a categorical variable (categorical to numerical conversion). … ] In one embodiment, all of the metrics raw and derived, and together with labels are used to train models of varying granularity. … The extracted features together with custom thresholds or ranges specific to the project are compared against the trained ensemble model. Widanapathirana et al. [para. 0134-0139);
Widanapathirana et al. fails to explicitly disclose the matrix structured in rows of employee related data across different dimensions and columns of project related data, the matrix comprising values indicative of performance scores derived from the project performance data; for missing ones of the values in the matrix, generating the missing ones of the values in the matrix from a training process; for an input of a new project: identifying ones of the rows and ones of the columns that match the employee related data in the new project and the project related data of the new project; aggregating ones of the values corresponding to the identified ones of the rows and the ones of the columns; ranking employee groups from the employee related data according to the aggregated ones of the values; and selecting an employee group from the employee related data having a highest rank for execution of the new project. Jersin et al. discloses these limitations. (Factorization machines (FMs) and their extension, field-aware factorization machines (FFMs), which have a broad range of applications for machine learning tasks including regression, classification, collaborative filtering, search ranking, and recommendation. Jersin et al. [para. 0030]. … The referenced document describes how the FFM model extends the FM model with information about groups (called fields) of features. In some embodiments, a latent space with more than two dimensions (e.g., three, four, or five dimensions) may be used. To generate this mapping, latent vectors learned by a factorization machine trained on examples of <Current Title, Skills>tuples are extracted from a data store. … embodiments create streams of suggested candidates (i.e., suggested streams) based on predicted positions for which a recruiter may be hiring. In certain embodiments, positions or job titles may be predicted based on criteria. Jersin et al. [para. 0032-0036]. … an intelligent matching system can return potential candidates to a user of the system … the intelligent matching system can revise a stream of suggested candidates that are returned to the user. Jersin et al. [para. 0039-0040]. … techniques for predicting numerical outcomes in a matrix-defined problem space where the numerical outcomes may include any outcomes that may be expressed numerically. …The referenced documents also describe a control server that stores (or is coupled with a data repository that stores) a matrix with multiple dimensions, where one of the dimensions represents features, such as employers, job titles, universities attended, and the like, and another one of the dimensions represents entities, such as individuals or employees. The control server separates the matrix into multiple submatrices along a first dimension (e.g. features). … the server provides, for display at the client device, an output based on the ordered set of job candidates. Jersin et al. [para. 0058-0065]. … machine-learning tools operate by building a model from example training data in order to make data-driven predictions or decisions expressed as outputs or assessments. Jersin et al. [para. 0073]. … the query builder 306 may be implemented as a system component or block that is configured to aggregate the raw attributes across the input candidates, expand them to similar attributes, and then select the top attributes that most closely represent the suggested candidates. Jersin et al. [para. 0100; Fig.3-5]. … probabilities of occurrences of clusters of skills may be determined for suggested candidates. The suggested candidates can be conceptualized as a training dataset used to determine probabilities of occurrences of skills amongst suggested candidates for a given organization. Jersin et al. [para. 0112, 0150-0152, 0171-0174]. …. a formed expertise matrix 1004 is very sparse since only a small percentage of the pairs can be predicted with any degree of certainty. The formed expertise matrix 1004 may be factorized into a member matrix 1006 and a skill matrix 1008 in K-dimensional latent space. Then; the dot product of the formed expertise matrix 1004 and the skill matrix 1008 is computed to fill in the ‘unknown’ cells. Jersin et al. [para. 0177-0179; Fig. 9-10]. … different machine-learning tools may be used. For example, Logistic Regression (LR), Naive-Bayes, Random Forest (RF), neural networks (NN), matrix factorization, and Support Vector Machines (SVM) tools may be used for classifying or scoring job candidate attributes. Jersin et al. [para. 0186-0188, 0192-0198; Fig. 11-13]). It would have been obvious to one or ordinary skill in the art of human resource and project management before the effective filing date of the claimed invention to modify the data analysis steps of Widanapathirana et al. to include matrix based data analysis to match employee attributes with project features and output top ranked matches as disclosed by Jersin et al. for building a model from example training data in order to make data-driven predictions or decisions expressed as outputs or assessments. Jersin et al. [para. 0073], in a manner that would have yielded predictable results to one of ordinary skill in the art at the time of invention.
Regarding Amended Claim 9, Widanapathirana et al. and Jersin et al. combined disclose the method further comprising, for execution of a supervised machine learning model on generated features, derived clusters, derived anomalies, and a probability of a key performance indicator value at a milestone to generate a predicted performance value of the project: (Data inputs 125 may also come from one or more other systems such as finance or scheduling systems, resource planning system, time sheeting, and specialised additional data streams. Widanapathirana et al. [para. 0023]. … input data is processed to extract direct features (metrics) that represent process characteristics such as execution efficiency, project segment, project profile, process volumes, process execution performance. Timelines may be used to derive the evolution of metrics over the duration of the project. Widanapathirana et al. [para. 0131-0132] … In one embodiment, the same metrics calculations (as for the initiation stage 905) are applied to the incoming data from a project to calculate raw and derived features (metrics). Widanapathirana et al. [para. 0136-0139; Fig. 9]);
executing a self-profiling algorithm configured with unsupervised machine learning on the generated features to derive clusters and anomalies of the project; (The data is used to construct a training set and apply regression analyses, classification analyses and novel anomaly detection analyses to construct a model that calculates a probability of an ongoing project performing inefficiently, poorly, or indeed failing, and/or construct a model predicting an ongoing project performing inefficiently, poorly, or indeed failing. Widanapathirana et al. [para. 0017-0018]. …input data 125 and 130 is processed using predictive analytics and simulation. A number of techniques may be used for such analysis…. the technique may comprise one of the following techniques or combinations of: Linear regression, Multivariate regression, Decision trees, Generalised neural network classification, Neural network regression, Support vector regression, Support vector classifications, K-means clusters, Gradient boosting, and/or Random forests. Widanapathirana et al. [para. 0033]. … the sub-system 110 monitors an ongoing project, or a collection of projects, for an organization in near-real time (e.g., daily and/or weekly) and detects unusual statistical patterns in activity data using a custom machine learning algorithm. Widanapathirana et al. [para. 0060-0066]. … project milestone outcomes and historical intra-project problem flags are used to label processes to indicate the historical outcomes or performance or process execution. Project outcomes are used to label overall project performance across various outcome metrics. … all of the metrics raw and derived, and together with labels are used to train models of varying granularity. Widanapathirana et al. [para. 0133-0136]);
executing a performance monitoring process on the generated features to determine the probability of the key performance indicator value at the milestone; ( … The sub-system 110 and machine learning algorithm calculates expected levels (counts) and confidence bands for each activity type over time-windows of fixed length (e.g. daily or weekly) and compares the actual observed activity levels of each type with the calculated confidence bands. A confidence band (or tolerance, e.g. 80%, 90%, 95% etc.) can be set with default values and/or by a user via the interface 135. Widanapathirana et al. [para. 0063]. … predict an ongoing project's likelihood of facing similar problems in the future. … Historical specific project problem flags and milestone outcomes. Widanapathirana et al. [para. 0120-0132]);
and executing the supervised machine learning model on the generated features, the derived clusters, the derived anomalies, and the probability of the key performance indicator value at the milestone to generate a predicted performance of the project. (Project outcomes are used to label overall project performance across various outcome metrics. Some models are process specific and trained to detect process problems. Some models are based on a specific overall project outcome and trained to identify when that outcome is at risk. The models together act as an ensemble where each model is trained to detect macro or micro process anomalies that might lead to sub-optimal performance. … The extracted features together with custom thresholds or ranges specific to the project are compared against the trained ensemble model. Widanapathirana et al. [para. 0133-0139]).
Regarding Amended Claim 10, Widanapathirana et al. and Jersin et al. combined disclose the method, wherein the executing the self- profiling algorithm comprises: executing the unsupervised machine learning to generate unsupervised machine learning models based on the generated features; (The data is used to construct a training set and apply regression analyses, classification analyses and novel anomaly detection analyses to construct a model that calculates a probability of an ongoing project performing inefficiently, poorly, or indeed failing, and/or construct a model predicting an ongoing project performing inefficiently, poorly, or indeed failing. Widanapathirana et al. [para. 0017]. … input data is processed to extract direct features (metrics) that represent process characteristics… Extracted features (metrics) are combined using various mathematical techniques to derive additional metrics that can represent complex process behavior. Widanapathirana et al. [para. 0131-0136]);
executing supervised machine learning on results from each of the unsupervised machine learning models to generate supervised ensembled machine learning models, each of the supervised ensemble machine learning models corresponding to each of the unsupervised machine learning models; (Project outcomes are used to label overall project performance across various outcome metrics. Some models are process specific and trained to detect process problems. Some models are based on a specific overall project outcome and trained to identify when that outcome is at risk. The models together act as an ensemble where each model is trained to detect macro or micro process anomalies that might lead to sub-optimal performance. … The extracted features together with custom thresholds or ranges specific to the project are compared against the trained ensemble model. Widanapathirana et al. [para. 0133-0139]);
and selecting ones of the unsupervised machine learning models as the models configured to derive the clusters and the anomalies based on an evaluation of the results of the unsupervised machine learning models against predictions generated by the supervised ensemble machine learning models. (Based on the project-specific thresholds 130 (see FIG. 1), anomalous metrics may be filtered to select only highly anomalous metrics. In some embodiments, a scoring is applied to each detected anomaly based on the severity of anomaly. Widanapathirana et al. [para. 0116] … In one embodiment, the same metrics calculations (as for the initiation stage 905) are applied to the incoming data from a project to calculate raw and derived features (metrics). The extracted features together with custom thresholds or ranges specific to the project are compared against the trained ensemble model. Widanapathirana et al. [para. 0139-0140]).
Regarding Amended Claim 14, Widanapathirana et al. and Jersin et al. combined disclose the method, wherein the executing the performance monitoring process comprises: generating, from historical projects, a transition network relating a transition between a plurality of first key performance indicators in a first milestone to a plurality of second key performance indicators in a second milestone; wherein the probability of the transition is determined based on a number of times the plurality of first key performance indicators of the first milestone transitioned to the second key performance indicators in the second milestone. (the observed levels of process activities are calculated and compared to expected levels of process activities for corresponding processes in corresponding time periods calculated by the machine learning model. … The machine learning model is trained, in one embodiment, on historical data from projects similar to the project being monitored. … The sub-system 110 and machine learning algorithm calculates expected levels (counts) and confidence bands for each activity type over time-windows of fixed length (e.g. daily or weekly) and compares the actual observed activity levels of each type with the calculated confidence bands. Widanapathirana et al. [para. 0061-0063]. … Each metric for the project is compared against an expected range for example to identify a level of deviation from the expected range and a deviation direction. … based on at least the historical project data from similar projects, the machine learning model generates an expected level of process activities that are expected to occur for each of the plurality of processes during each of the plurality of time periods. Widanapathirana et al. [para. 0078-0087] … the machine learning model compares, for each of the processes identified, the observed levels of process activities to the expected levels of process activities in a corresponding time period. … the expected levels of activities for process X may be defined as: if there are five counts of process activity type-A, then there should also be three counts of process activity type-B. A contextual threshold value(s) may be set for each process based on the learned data. … if the observed levels of process activities fail to match (within a threshold, contextual and/or individually) or fall within a defined range…the sub-system 110 generates an anomaly alert for the associated process. Widanapathirana et al. [para. 0090-0095]).
Regarding Amended Clam 15, Widanapathirana et al. and Jersin et al. combined disclose the method, wherein the executing the performance monitoring process on the generated features to determine the probability of a key performance indicator value at the milestone comprises: generating a multi-tasking supervised machine learning model to predict the key performance indicator values at the milestone based on the key performance indicator values at earlier milestones; wherein the probability of the key performance indicator value is one of a category for when a classification model is used as the multi-tasking supervised machine learning model or a numerical score for when a regression model is used as the multi-tasking supervised machine learning model; (The data is used to construct a training set and apply regression analyses, classification analyses and novel anomaly detection analyses to construct a model that calculates a probability of an ongoing project performing inefficiently, poorly, or indeed failing, and/or construct a model predicting an ongoing project performing inefficiently, poorly, or indeed failing. Widanapathirana et al. [para. 0017]. … In some embodiments, input data 125 and 130 is processed using predictive analytics and simulation. A number of techniques may be used for such analysis. … the technique may comprise one of the following techniques or combinations of: Linear regression, Multivariate regression, Decision trees, Generalised neural network classification, Neural network regression, Support vector regression, Support vector classifications, K-means clusters, Gradient boosting, and/or Random forests. Widanapathirana et al. [para. 0033]. … project milestone outcomes and historical intra-project problem flags are used to label processes to indicate the historical outcomes or performance or process execution. Project outcomes are used to label overall project performance across various outcome metrics. Some models are process specific and trained to detect process problems. Some models are based on a specific overall project outcome and trained to identify when that outcome is at risk. The models together act as an ensemble where each model is trained to detect macro or micro process anomalies that might lead to sub-optimal performance. … The extracted features together with custom thresholds or ranges specific to the project are compared against the trained ensemble model. … These detected anomalies are scored by considering the severity of the problem, the likelihood of having a negative performance impact and for example a pre-defined weighting across the ensemble.
In one embodiment, scores may be aggregated to derive an overall predictive warning score. Widanapathirana et al. [para. 0133- 0144]);
and wherein the key performance indicator values at the milestone are predicted concurrently. (Method 500 is shown as a series of blocks that represent functions/actions being performed. However, the functions/actions may be performed in other orders, and/or one or more functions/actions may occur concurrently with other functions/actions. Widanapathirana et al. [para. 0081, 0086-0087, 0158; Fig. 5]. … The anomaly detection algorithm identifies an activity that is too high or low given a surrounding context of other concurrent process activities that are also expected in the process at that point in time. Widanapathirana et al. [para. 0092]).
Claims 11-13 are rejected under 35 U.S.C. 103 as being unpatentable over Widanapathirana et al. (US 2019/0108471), in view of Jersin et al. (US 2019/0197180), and in further view of Phillipps et al. (US 2014/0372346).
Regarding Amended Claim 11, Widanapathirana et al. and Jersin et al. combined fail to explicitly disclose the method, wherein the executing the self- profiling algorithm comprises: executing each unsupervised machine learning model algorithm from a set of unsupervised learning model algorithms on the generated features; determining one of the unsupervised machine learning models with an associated parameter set for the each unsupervised machine learning model algorithm that meets a selection criterion; determining an unsupervised machine learning model for deployment across the set of the unsupervised machine learning model algorithms from the one of the unsupervised machine learning models of the each unsupervised machine learning model algorithm that meets the selection criterion. Phillipps et al. discloses these limitations. (… the function generator module 301 may generate a plurality of learned functions for various combinations of features, and the machine learning compiler module 302 may evaluate the learned functions and generate evaluation metadata. Based on the evaluation metadata, the feature selector module 304 may select a subset of features that are most accurate or effective, and the machine learning compiler module 302 may use learned functions that utilize the selected features to build the machine learning ensemble 222. The feature selector module 304 may select features for use in the machine learning ensemble 222 based on evaluation metadata for learned functions from the function generator module 301, combined learned functions from the combiner module 306, extended learned functions from the extender module 308, combined extended functions, synthesized learned functions from the synthesizer module 310, or the like. … At each iteration, the function evaluator module 312 may determine an overall effectiveness of the learned functions in aggregate for the current iteration's selected combination of features. Phillipps et al. [para. 0107-0117; Fig. 3]. … The combiner module 306 combines learned functions, forming sets, strings, groups, trees, or clusters of combined learned functions. … the combiner module 306, in one embodiment, may determine each possible combination of generated learned functions, as many combinations of generated learned functions as possible given one or more limitations or constraints, a selected subset of combinations of generated learned functions, or the like. Phillipps et al. [para. 0121-0124] ). It would have been obvious to one of ordinary skill in the art of project management and machine learning implementation before the effective filing date of the claimed invention to modify the combined machine learning steps of Widanapathirana et al. and Jersin et al. to include the machine learning steps of Phillipps et al. for performing data analytics using machine learning (Phillipps et al. [para. 0007), in a manner that would have yielded predictable results to one of ordinary skill in the art before the effective filing date of the claimed invention.
Regarding Amended Claim 12, Widanapathirana et al. and Jersin et al. combined fail to explicitly disclose the method, further comprising deploying the unsupervised machine learning model for deployment; and during deployment of the unsupervised machine learning model for deployment: applying the unsupervised machine learning model for deployment to the features to generate unsupervised output; attaching the unsupervised output to the features to obtain expanded features; randomly selecting another unsupervised learning model algorithm from the set of unsupervised machine learning model algorithms; training the randomly selected unsupervised learning model algorithm on the expanded features to find another unsupervised machine learning model with another associated parameter set that meets the selection criterion; for the another unsupervised learning model generated from the randomly selected unsupervised learning model algorithm having a better evaluation than the deployed unsupervised learning model for deployment, replacing the deployed unsupervised learning model for deployment with the another unsupervised learning model. Phillipps et al. discloses these limitations. (… the machine learning compiler module 302 includes an extender module 308. The extender module 308, in certain embodiments, is configured to add one or more layers to a learned function. For example, the extender module 308 may extend a learned function or combined learned function by adding a probabilistic model layer, such as a Bayesian belief network layer, a Bayes classifier layer, a Boltzmann layer, or the like. … While the extending of learned functions may be informed by evaluation metadata for the learned functions, in certain embodiments, the extender module 308 generates a large number of extended learned functions pseudo-randomly. Phillipps et al. [para. 0124-0127]. … The function evaluator module 312, in one embodiment, evaluates a learned function by inputting the test data into the learned function to produce a result, such as a classification, a confidence metric, an inferred function, a regression function, an answer, a prediction, a recognized pattern, a rule, a recommendation, an evaluation, or another result. … the evaluation metadata may include evaluation metrics, learning metrics, effectiveness metrics, convergence metrics, or the like for a learned function based on an evaluation of the learned function. Phillipps et al. [para. 0130-0135]. … The function selector module 316, in certain embodiments, may use evaluation metadata from the metadata library 314 to select learned functions for the combiner module 306 to combine, for the extender module 308 to extend, for the synthesizer module 310 to include in the machine learning ensemble 222, or the like. Phillipps et al. [para. 0137]). It would have been obvious to one of ordinary skill in the art of project management and machine learning implementation before the effective filing date of the claimed invention to modify the combined machine learning steps of Widanapathirana et al. and Jersin et al. to include the machine learning steps of Phillipps et al. for performing data analytics using machine learning (Phillipps et al. [para. 0007), in a manner that would have yielded predictable results to one of ordinary skill in the art before the effective filing date of the claimed invention.
Regarding Amended Claim 13, Widanapathirana et al. and Jersin et al. combined fail to explicitly disclose the method, wherein the executing the self- profiling algorithm comprises: a) applying a set of unsupervised machine learning model algorithms to the generated features to generate an unsupervised machine learning model for each of the unsupervised machine learning algorithms; b) attaching unsupervised output from the unsupervised machine learning model to the features; and c) reiterating steps a) and b) until an exit criterion is met. Phillipps et al. discloses these limitation. (the feature selector module 304 causes the machine learning compiler module 302 to repeat generating, combining, extending, and/or evaluating learned functions while iterating through permutations of feature sets. At each iteration, the function evaluator module 312 may determine an overall effectiveness of the learned functions in aggregate for the current iteration's selected combination of features. Phillipps et al. [para. 0117]. … The combiner module 306 may determine which learned functions to combine, how to combine learned functions, or the like based on evaluation metadata for the learned functions from the metadata library 314, generated based on an evaluation of the learned functions using test data. Phillipps et al. [para. 0122]. … the synthesizer module 310 organizes learned functions by preparing the learned functions and the associated evaluation metadata for processing workload data to reach a result. … The function evaluator module 312, in one embodiment, evaluates a learned function by inputting the test data into the learned function to produce a result. Phillipps et al. [para. 0126-0135]). It would have been obvious to one of ordinary skill in the art of project management and machine learning implementation before the effective filing date of the claimed invention to modify the combined machine learning steps of Widanapathirana et al. and Jersin et al. to include the machine learning steps of Phillipps et al. for performing data analytics using machine learning (Phillipps et al. [para. 0007), in a manner that would have yielded predictable results to one of ordinary skill in the art before the effective filing date of the claimed invention.
Allowable Subject Matter
Claims 1-7 are rejected under 35 U.S.C. 112(b) and 35 U.S.C. 101, but the claims would be allowable if the aforementioned rejections are overcome.
An updated prior art search was conducted, as necessitated by amendment. Examiner analyzed amended Claim 1 in view of the prior art of record and the updated prior art search and finds not all claim limitations are explicitly taught nor would one of ordinary skill in the art find it obvious to combine these references with a reasonable expectation of success. Moreover since the specific ordered combined sequence of claim elements recited in claim 1 can only be found as recited in Applicant’s specification, any combination of the cited references and/or additional references to teach all the claim elements would be the result of impermissible hindsight reconstruction. Accordingly, any combination of the prior art of record and any of the additional references would be improper to teach the claimed invention. As a result, claims 1-7 are eligible over the prior art.
Conclusion
The prior art made of record and not relied upon is considered pertinent to Applicant’s disclosure:
Principe et al. (US 2016/0242690) - classification and dimensionality reduction methods in accordance with various embodiments. In the top right quadrant are the most versatile and powerful methods, such as auto-encoders for non-linear dimensionality reduction and multilayer perceptron. These methods comprise a priori selection of the architecture, but the optimal parameters and hyper-parameters are inferred from the data. These models typically utilize extensive computation time. A number of unsupervised methods can be used for the exploratory analysis of neural data. Principal component analysis (PCA) can reveal patterns in neural data, but the results from PCA on neural data may be unsatisfactory, as the directions of largest variance may not contain any useful information. In the linear case, independent component analysis (ICA) can optimize a linear projection so the resulting vector has maximally independent elements.
Beauchesne et al. (US 11,853,853) - methods, systems, and processes for implementing a machine learning anomaly detection system capable of detecting anomalies (e.g. outlier processes or hosts) in hunt data gathered from a computer network. Embodiments of the anomaly detection system implement a detection pipeline comprising multiple stages, including pre-processing (e.g. feature engineering), dimensionality reduction, anomaly scoring using machine learning algorithms, and an interpretability layer. Depending on the embodiment, dimensionality reduction in the anomaly detection system may be performed using one or more of several methodologies. The methodologies include but are not limited to: Principal Component Analysis (PCA); Non-Negative Matrix Factorization (NMF), which may be adapted to positive data, with variants such as Binary Matrix Factorization (BMF); Logistic PCA, which is type of PCA that accounts for the binary nature of the data and is particularly relevant to hunt data. Auto Encoder, which is a type of feed-forward neural network that is trained using machine learning to approximate an identity function that maps a high-dimensional feature vector onto itself, but using an intermediate representation that has reduced dimensionality; and Randomized Projection, which uses random projection to reduce dimensionality while preserving pairwise distances, thus allowing for distance-based methods in downstream tasks. However, in some embodiments, the foregoing process can be very time consuming. Therefore, in some embodiments, a regression algorithm is trained on one or more metrics related to the data (such as matrix rank, sparsity, dimensions, etc.) to predict the associated k. In this manner, computing resource heavy calculations are avoided in production, while an effective predicted value of k is obtained fairly quickly (e.g., with a mean squared error (MSE) of 0.22 for normalized data).
Applicant’s amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, 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 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 LETORIA G KNIGHT whose telephone number is (571)270-0485. The examiner can normally be reached M-F 9am-5pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Rutao WU can be reached at 571-272-6045. 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.
/L.G.K/Examiner, Art Unit 3623 /RUTAO WU/Supervisory Patent Examiner, Art Unit 3623
Prosecution Insights