DETAILED ACTION
Remarks
Claims 1-22 have been examined and rejected. This Office Action is responsive to the amendment filed on 01/06/2026, which has been entered in the above identified application.
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 .
Claims 1-22 are presented for examination.
Respond to Amendment
The amendment filed 01/06/2026 has been entered. Claims 1, 3-5, 10-17, 19 and 20 have been amended. Claims 1-22 are pending in the application.
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 1-22 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Independent claims
Step 1
Claim 1 is drawn to a method for modifying a trained classification model, claim 10 is drawn to a non-transitory computer readable storage medium storing program code to perform the method of claim 1, and claim 18 is drawn to an electronic device. Each of these claim groups falls under one of four categories of statutory subject matter (process/method, machines/product/apparatus, manufactures, and composition of matter).
Step 2A – Prong 1
Claims 1, 10 and 18 are directed to a judicially recognized exception of an abstract idea without significantly more.
Claims 1, 10 and 18 recite a method of receiving feature data extracted from sensor data that under its broadest reasonable interpretation enumerates a mental concept. A human can mentally, with or without a pen and paper, to analyze and to observe the feature data. Therefore, the step of receiving feature data is nothing more than a mental concept (MPEP 2106.04(a)(2)(III)).
Claims 1, 10 and 18 recite further a method of classifying the feature data according to the trained classification model to identify a label corresponding to the feature data, wherein the trained classification model comprises a decision tree comprising a plurality of decision nodes for feature identification for a plurality of features, wherein the trained classification model is deployed on an electronic device that under its broadest reasonable interpretation enumerates a mathematical concept. Classifying data using decision tree is a conventional mathematical step using conventional mathematical tool to analyze data. Therefore, the step of classifying the feature data according to the trained classification model is nothing more than a mathematical concept (MPEP 2106.04(a)(2)(I)).
Step 2A – Prong 2
Claims 1, 10 and 18 recite further tracking identified features of the plurality of features over a predetermined amount of time that fails to integrate the abstract idea into a practical application. The step of tracking features is a form of insignificant input and output solution activities, where tracking identified features over a predetermined amount of time is necessary for all uses of the judicial exception. This additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (MPEP 2106.05(g)).
Claims 1, 10 and 18 recite further responsive to determining that a feature of the trained classification model does not satisfy a frequency of usage threshold over the predetermined amount of time, deactivating a decision node of the decision tree of the trained classification model corresponding to the feature such that a subsequent instance of the classifying does not consider a deactivated decision node for subsequently received feature data that fails to integrate the abstract idea into a practical application. The step of deactivating a decision node is a form of insignificant input and output solution activities, where deactivating a decision node of the decision tree is necessary for all uses of the judicial exception. This additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (MPEP 2106.05(g)).
Step 2B
The additional elements in step 2A-Prong 2 those are forms of insignificant extra-solution activities, do not amount to significantly more than an abstract idea because the court decision have determined that these additional elements of tracking identified features over a predetermined amount of time; and deactivating a decision node of the decision tree to be well-understood, routine, and conventional when claimed in a merely generic manner (MPEP 2106.05(d)(II)).
As such, claims 1, 10 and 18 are not patent eligible.
Dependent claims
Claims 2-9, 11-17 and 19-22 merely narrow the previously recited abstract idea limitations. For the reasons described above with respect to claims 1, 10 and 18, this judicial exception is not meaningfully integrated into a practical application, or significantly more than the abstract idea. The claims disclose similar limitations described for the independent claims above and do not provide anything more than the mathematical process. Therefore, claims 2-9, 11-17 and 19-22 also recite abstract ideas that do not integrate into a practical application or amount to significantly more than the judicial exception, and are rejected under U.S.C. 101.
Step 1
Claims 2-9 are drawn to a method for modifying a trained classification model, claims 11-17 are drawn to a non-transitory computer readable storage medium storing program code to perform the method of claims 2-9, and claims 19-22 are drawn to an electronic device. Each of these claim groups falls under one of four categories of statutory subject matter (process/method, machines/product/apparatus, manufactures, and composition of matter).
Step 2A – Prong 1
Dependent claims 2 and 11 recite further the mental concept by tracking a usage rate of each feature of the plurality of features that is based on one or more features of the ML project (MPEP 2106.04(a)(2)(III)).
Dependent claims 3, 12 and 19 recite further the mathematical concept by comparing the usage rate of each feature of the plurality of features to the frequency of usage threshold over the predetermined amount of time that is based on one or more features of the ML project (MPEP 2106.04(a)(2)(I)).
Dependent claims 4 and 13 recite further the mathematical concept by for each instance of an identified feature, applying a weight to the identified feature of the plurality of features; and aggregating the weights for the plurality of features over the predetermined amount of time those are based on one or more features of the ML project (MPEP 2106.04(a)(2)(I)).
Dependent claims 5, 14 and 20 recite further the mathematical concept by comparing aggregated weights for each feature of the plurality of features to the frequency of usage threshold over the predetermined amount of time, wherein the frequency of usage threshold corresponds to a minimum aggregated weight that is based on one or more features of the ML project (MPEP 2106.04(a)(2)(I)).
Dependent claims 6, 15 and 21 recite further the mental concept by reactivating the decision node of the decision tree responsive to a reset command that is based on one or more features of the ML project (MPEP 2106.04(a)(2)(III)).
Step 2A – Prong 2
Dependent claims 7 and 16 recite further the insignificant extra solution activities by wherein the trained classification model is implemented within a feature sensing device for use in feature identification for a user. This additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (MPEP 2106.05(g)).
Dependent claims 8, 17 and 22 recite further the insignificant extra solution activities by wherein the feature sensing device is configured for use by a plurality of users, such that each user of the plurality of users has an associated trained classification model and wherein the method is performed separately for each user of the plurality of users. This additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (MPEP 2106.05(g)).
Dependent claim 9 recites further the insignificant extra solution activities by wherein the subsequent instance of the classifying that does not consider the deactivated decision node requires less computational load than a prior instance of the classifying prior to the deactivating the decision node. This additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (MPEP 2106.05(g)).
As such, dependent claims 2-9, 11-17 and 19-22 are not patent eligible.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-22 are rejected under 35 U.S.C. 103 as being unpatentable over Hasanian et al (US 20210365016 A1) hereafter Hasanian, and further in view of Kumari et al (US 10475125 B1) hereafter Kumari.
With respect to claim 1, Hasanian teaches a method for modifying a trained classification model (a neural network or a type of machine learning technique may be used to identify features or defects or normal AM processing and operations, and may be utilized for decision-making such as modifying or adjusting an AM process. Waveforms and extracted signal features may be used for pattern recognition and data classification [par. 0029-0031]), the method comprising:
receiving feature data extracted from sensor data (The plurality of sensors may provide sensor data to the control system. The sensor data may be used by the control system to detect the occurrence of anomalies or defects in the additive manufacturing process. Patterns may be identified within sensor data or AE sensor data [par. 0029-0031]);
classifying the feature data according to the trained classification model to identify a label corresponding to the feature data (a classifier may be utilized to classify patterns as being indicative. Patterns that are recognized but are different from defect-related or normal printing operations may be classified as anomalies. The database of sensor data may include acoustic signatures or features indicative of normal operations of the manufacturing devices. [par. 0031, 0032]);
tracking identified features of the plurality of features over a predetermined amount of time (data fusion operations may involve analysis of multiple types of sensor data. Sensor data received from a plurality of sensors may be timestamped to enable the sensor data to be aligned in time. Features extracted from emission signals may be amplitude, rise time, energy, number of counts, duration, entropy, frequency content, etc. Samples of acoustic waves or other sensor data may be timestamped and associated with AM process to facilitate synchronization [par. 0020, 0038, 0050, 0051]).
However, Hasanian does not disclose wherein the trained classification model comprises a decision tree comprising a plurality of decision nodes for feature identification for a plurality of features, wherein the trained classification model is deployed on an electronic device; and responsive to determining that a feature of the trained classification model does not satisfy a frequency of usage threshold over the predetermined amount of time, deactivating a decision node of the decision tree of the trained classification model corresponding to the feature such that a subsequent instance of the classifying does not consider a deactivated decision node for subsequently received feature data.
In the same field of endeavor, Kumari teaches wherein the trained classification model comprises a decision tree comprising a plurality of decision nodes for feature identification for a plurality of features, wherein the trained classification model is deployed on an electronic device (Pruning the historical data to reduce a number of features available for inclusion in a classification model, wherein the classification model includes a decision tree-based classifier. Pruning the data may ultimately reduce a number of features available for use as nodes or decisions within decision trees. Each tree includes a plurality of decision nodes used to determine whether a particular life event is affecting a user [col. 6, lines 1-30; col. 6, line 63 – col. 7, line 15]); and
responsive to determining that a feature of the trained classification model does not satisfy a frequency of usage threshold over the predetermined amount of time (each decision node results in a decision based on the comparison of a value from the financial data of the user with a threshold value associated with a feature in the decision tree. For example, a feature engineering may include users in the historical data summing the values associated with the user’s credit card purchases to determine a monthly credit card purchase amount, and counting a number of purchase transactions occurring over a span of time to determine an average transaction frequency. A pre-determined threshold may be selected for use as an attribute in the classification model [col. 6, line 63 – col. 7, line 15; col. 10, line 60 – col. 11, line 55]), deactivating a decision node of the decision tree of the trained classification model corresponding to the feature such that a subsequent instance of the classifying does not consider a deactivated decision node for subsequently received feature data (In an example, decision tree includes decisions based on comparison of the user’s income to an income threshold (or user’s age, user’s pay period). Deactivating a decision node corresponding to a feature such that the instance of the classifying does not consider a deactivated decision nodes for subsequently received feature data is actually considered as the pruning technique in decision tree. Pruning to reduce a number of features available for inclusion in a classification model, while excluding some unnecessary features. Pruning the historical data may include any operation that reduces a number of features available for inclusion in a classification model. Reducing/pruning features results in corresponding nodes no longer being used in classification [col. 6, lines 1-15; col. 15, lines 25-45]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have incorporated the concept of identifying a life event affecting a user by employing a classification model as suggested by Kumari into the concept of utilizing acoustic sensors to detect defects as suggested by Hasanian because both of these systems addressing the process of extracting feature data and using a type of machine learning for the purposes of data classification. Doing so would be desirable because the system of Hasanian would be more efficient by obtaining a classification model along with decision tree technique to model how financial data of a user relates to life events, and pruning the historical data to reduce a number of features to be included in the classification model (Kumari, [col. 1, line 25 – col. 2, line 15; col. 6, lines 1-15]).
With respect to claim 2, the combination of Hasanian and Kumari teaches wherein the tracking identified features of the plurality of features over a predetermined amount of time comprises: tracking a usage rate of each feature of the plurality of features (Hasanian, the decision-making logic may be used to detect anomalies of defects based on the flow rate. The control operations may include halting the additive manufacturing process or modifying the flow rate of the material. Flow rate along with sensor data can be fused and analyzed together to assess the process [par. 0020, 0028, 0035, 0039]).
With respect to claim 3, the combination of Hasanian and Kumari teaches further comprising: comparing the usage rate of each feature of the plurality of features to the frequency of usage threshold over the predetermined amount of time (Hasanian, data fusion operations may involve analysis of multiple types of sensor data. Sensor data received from a plurality of sensors may be timestamped to enable the sensor data to be aligned in time. Features extracted from emission signals may be amplitude, rise time, energy, number of counts, duration, entropy, frequency content, etc. Samples of acoustic waves or other sensor data may be timestamped and associated with AM process to facilitate synchronization. For example, when accumulated defects exceed a threshold number of defects, the process may be stopped. If the threshold number of defects is not reached, the process may proceed to completion [par. 0020, 0038, 0040, 0050, 0051]).
With respect to claim 4, the combination of Hasanian and Kumari teaches wherein the tracking identified features over the predetermined amount of time comprise:
for each instance of an identified feature, applying a weight to the identified feature of the plurality of features (Hasanian, a machine learning process may utilize an artificial neural network with connected nodes that transmit the input data (feature data) throughout the specific functions and coefficients. A chain or series of matrices and functions may be configured with coefficients and weights to transform the input data to identify certain patterns or determine decisions based on the set of input data [par. 0030, 0049]); and
aggregating the weights for the plurality of features over the predetermined amount of time (Hasanian, the classification method includes determining whether the extracted features or signatures are important. Upon determining the desired features, coefficients and optimized weights may be calculated by exposing and training the models to generate data. The defects may be monitored with timestamped information during the manufacturing process [par. 0051-0053]).
With respect to claim 5, the combination of Hasanian and Kumari teaches further comprising: comparing aggregated weights for each feature of the plurality of features to the frequency of usage threshold over the predetermined amount of time, wherein the frequency of usage threshold corresponds to a minimum aggregated weight (Hasanian, upon determining the desired features, coefficients and optimized weights may be calculated by exposing and training the models to generate data. The defects may be monitored with timestamped information during the manufacturing process, wherein the detected defects may be compared to one or more defined thresholds [par. 0046, 0051-0053]).
With respect to claim 7, the combination of Hasanian and Kumari teaches wherein the trained classification model is implemented within a feature sensing device for use in feature identification for a user (Kumari, data may be collected to determine whether the identification of a life event of a user is correct. If it is correct, then the method ends with respect to the user [col. 9, line 35 – col. 10, line 7]).
With respect to claim 8, the combination of Hasanian and Kumari teaches wherein the feature sensing device is configured for use by a plurality of users, such that each user of the plurality of users has an associated trained classification model and wherein the method is performed separately for each user of the plurality of users (Kumari, a business may better operate if it knows the professions of the users. Once a life event is identified as affecting a user, other information about the user may be deduced. A platform with a large user base can perform analysis of the data associated with multiple users. The corrected identification is fed into the classification model. Feeding the corrected identification into the classification model may include adding a user and their financial data within a training dataset [col. 1, lines 5-25; col. 3, lines 15-30; col. 9, line 35 – col. 10, line 7]).
With respect to claim 9, the combination of Hasanian and Kumari teaches wherein the subsequent instance of the classifying that does not consider the deactivated decision node requires less computational load than a prior instance of the classifying prior to the deactivating the decision node (Kumari, pruning the historical data may ultimately reduce a number of features available for use as nodes or decisions within decision trees, such process may reduce the computational resources needed to classify a new data point, and it makes the classification process faster and more efficient. The dataset generator 110 may perform other operations on the historical data that optimize the historical data for queries run against the historical data [col. 1, lines 1-15]).
With respect to claim 10, it is a non-transitory computer readable storage medium that is corresponding to the method of claim 1. Therefore, it is rejected for the same reason as claimed in claim 1 above.
With respect to claim 11, it is a non-transitory computer readable storage medium that is corresponding to the method of claim 2. Therefore, it is rejected for the same reason as claimed in claim 2 above.
With respect to claim 12, it is a non-transitory computer readable storage medium that is corresponding to the method of claim 3. Therefore, it is rejected for the same reason as claimed in claim 3 above.
With respect to claim 13, it is a non-transitory computer readable storage medium that is corresponding to the method of claim 4. Therefore, it is rejected for the same reason as claimed in claim 4 above.
With respect to claim 14, it is a non-transitory computer readable storage medium that is corresponding to the method of claim 5. Therefore, it is rejected for the same reason as claimed in claim 5 above.
With respect to claim 16, it is a non-transitory computer readable storage medium that is corresponding to the method of claim 7. Therefore, it is rejected for the same reason as claimed in claim 7 above.
With respect to claim 17, it is a non-transitory computer readable storage medium that is corresponding to the method of claim 8. Therefore, it is rejected for the same reason as claimed in claim 8 above.
With respect to claim 18, it is an electronic device that is corresponding to the method of claim 1. Therefore, it is rejected for the same reason as claimed in claim 1 above.
With respect to claim 19, it is an electronic device that is corresponding to the method of claim 3. Therefore, it is rejected for the same reason as claimed in claim 3 above.
With respect to claim 20, it is an electronic device that is corresponding to the method of claim 5. Therefore, it is rejected for the same reason as claimed in claim 5 above.
With respect to claim 22, it is an electronic device that is corresponding to the method of claim 8. Therefore, it is rejected for the same reason as claimed in claim 8 above.
Claims 6, 15 and 21 are rejected under 35 U.S.C. 103 as being unpatentable over Hasanian et al (US 20210365016 A1) hereafter Hasanian, further in view of Kumari et al (US 10475125 B1) hereafter Kumari, as claimed in claim 1 above, and further in view of Palekar et al (US 12008583 B2) hereafter Palekar.
With respect to claim 6, the combination of Hasanian and Kumari does not disclose further comprising: reactivating the decision node of the decision tree responsive to a reset command.
In the same field of endeavor, Palekar teaches further comprising: reactivating the decision node of the decision tree responsive to a reset command (Tree pruning may prune trees to avoid overfitting. Tree nodes may choose a different number of nodes for the decision tree, and tree depth may use different tree depths for the decision tree. Multiple decision tree models may be generated [col. 4, lines 50-60]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have incorporated the concept of developing Autonomous Fraud Risk Management to identify emerging fraud trends as suggested by Palekar into the combination of Hasanian and Kumari because all of these systems addressing the process of extracting feature data and using a type of machine learning for the purposes of data classification. Doing so would be desirable because the combination of Hasanian and Kumari would be more efficient by developing fraud rule recommendations by applying one or more supervised machine learning techniques to the set of features to identify a decision tree model, wherein the tree nodes of pruning technique may choose different number of nodes for the decision tree (Palekar, [col. 1, line 35 – col. 2, line 30; col. 4, lines 50-60]).
With respect to claim 15, it is a non-transitory computer readable storage medium that is corresponding to the method of claim 6. Therefore, it is rejected for the same reason as claimed in claim 6 above.
With respect to claim 21, it is an electronic device that is corresponding to the method of claim 6. Therefore, it is rejected for the same reason as claimed in claim 6 above.
Response to Arguments
The examiner respectfully acknowledges the applicant’s amendments to claims 1, 3-5, 10-17, 19 and 20.
Applicant’s amendment filed on 01/06/2026 regarding the claim objections to claim 1 have been considered and are consequently withdrawn.
Applicant’s amendment filed on 01/06/2026 regarding the rejections to claims 3-5, 11-17, 19 and 20 under 35 USC 112(b) have been considered and are consequently withdrawn.
Applicant’s arguments filed on 01/06/2026 regarding the rejections to claims 1-22 under 35 USC 101 have been fully considered but are not persuasive.
Applicant argued that “A. The pending claims are not directed to an abstract idea … Applicant respectfully submits that at least the claimed "receivinq feature data extracted from sensor data; classifying the feature data according to the trained classification model to identify a label corresponding to the feature data, wherein the trained classification model comprises a decision tree comprisinq a plurality of decision nodes for feature identification for a plurality of features, wherein the trained classification model is deployed on an electronic device" (emphasis added) is not a "concept[] performed in the human mind (including an observation, evaluation, judgment, opinion" and is not "recited on [its] own or per se" as required by the 2019 Revised Patent Subject Matter Eligibility Guidance. Applicant submits that the claimed embodiments cannot be practically performed in the mind and absent a computer system and are not simply enumerations of a mathematical concept.”
Examiner respectfully disagrees.
Based on what is recited in claims 1, 10 and 18, the limitations receiving feature data extracted from sensor data encompass a mentally performable process of receiving and observing the feature data; and classifying the feature data according to the trained classification model to identify a label corresponding to the feature data, wherein the trained classification model comprises a decision tree comprising a plurality of decision nodes for feature identification for a plurality of features, wherein the trained classification model is deployed on an electronic device encompass a step of classifying data using decision tree that uses conventional mathematical tool to analyze data. Human can mentally analyze and observe a piece of data; and human can, with an aid of pen and paper, classify a piece of data using a decision tree model using mathematical tool to analyze such data. In fact, the claim as a whole does recite an abstract idea and falls within one of the groupings. The step of receiving feature data extracted from sensor data may not recite a mental process, but the step of classifying the feature data according to the trained classification model to identify a label corresponding to the feature data, wherein the trained classification model comprises a decision tree comprising a plurality of decision nodes for feature identification for a plurality of features, wherein the trained classification model is deployed on an electronic device clearly recites a mathematical concept, where decision tree classification and/or threshold-based feature pruning are fundamentally algorithmic mathematical operations.
The rejection does not just rely on the claim reciting a mental process. Rather, claim 1 recites mathematical concepts in the form of algorithmic classification and statistical evaluation operations. Specifically, the claim recites classifying data using decision tree classification model, tracking features over time, determining whether a feature satisfies a usage frequency threshold, and modifying the decision tree operation based on the determination.
Applicant argued that “B. The pending claims contain an inventive concept … Applicant respectfully submits that the pending claims do contain an inventive concept. In other words, the pending claims include recitations that amount to significantly more than the basic idea itself. Indeed, the claimed invention includes meaningful recitations beyond generally linking the alleged abstract idea to a particular technological environment, such as a computer. Rather, the pending claims recite a particular way of designing a processing chain of a sensor system. Specifically, the pending claims recite "tracking identified features of the plurality of features over a predetermined amount of time; and responsive to determininq that a feature of the trained classification model does not satisfy a frequency of usaqe threshold over the predetermined amount of time, deactivatinq a decision node of the decision tree of the trained classification model corresponding to the feature such that a subsequent instance of the classifying does not consider a deactivated decision node for subsequently received feature data" (emphasis added).”
Examiner respectfully disagrees.
Merely stating that the invention operates within a sensor system or modifies a processing chain does not automatically establish a practical application or technological environment. The actual claim recites receiving sensor data, classifying data using a decision tree, tracking feature usage frequency, and deactivating infrequently used nodes. The actual claim does not recite sensor hardware improvement, signal acquisition improvement, improved sampling, improved sensor fusion, etc. The claim looks like it recites improving the abstract model itself, rather than recites improving computer technology. The alleged processing chain is itself defined by the abstract data processing operations. The processing chain is not a separate technological structure, rather it is the abstract algorithm.
In fact, claim 1 as a whole recites abstract ideas including mathematical concepts in the form of algorithmic classification, statistical feature tracking, threshold evaluation, and decision tree classification operations. These specifically recites mathematical relationships, evaluations and algorithms. The additional elements do not integrate the abstract idea into a practical application. Claim 1 as a whole does not recite a specific technological improvement but rather recite a generic hardware only.
Therefore, claim 1 does not recite a technological improvement to sensor operation, sensor hardware, or computer functionality. Rather, the claim recites receiving sensor data, classifying the data using decision tree model, tracking feature data, and deactivating decision tree nodes based on thresholds. These limitations describe mathematical and algorithmic data processing operations applied in the context of a generic sensor environment and generic electronic device. The claim, as a whole, merely uses generic computing and sensor components as tools to perform abstract analytical operations.
Independent claim 1 and its corresponding claims 10 and 18 do not integrate the judicial exception into a practical application and are not patentable for at least the reasons above. Dependent claims 2-9, 11-17 and 19-22, those either directly or indirectly depended on the independent claims, are not patentable for the same reasons.
Applicant’s arguments filed on 01/06/2026 regarding the rejections to claims 1-22 under 35 USC 103 have been fully considered but are not persuasive.
Applicant argued that “Applicant understands Kumari to teach the generation of a classification model using a training dataset generated from historical data. In particular, prior to training the classification model, Kumari teaches that the historical data can be pruned "prior to generating the training dataset" (emphasis added; col. 6, lines 2-3). While this pruning impacts the number of features included in the classification model, the pruning is performed prior to the generation of the trained classification model. Therefore, Applicant submits that Kumari does not teach or suggest "deactivatinq a decision node of the decision tree of the trained classification model corresponding to the feature such that a subsequent instance of the classifying does not consider a deactivated decision node for subsequently received feature data" (emphasis added) as recited in independent Claim 1, and the similar recitations of independent Claims 10 and 18.”
Examiner respectfully disagrees.
Applicant argued that Kumari only discloses pruning historical data “prior to generating the training dataset,” and therefore fails to disclose the limitation “deactivating a node of the decision tree of the classification model.” However, the claim does not recite:
Any retraining restrictions,
Any requirement that node deactivation occur only after deployment,
Any preservation of the original tree structure,
Or that the node remains physically present but logically bypassed.
Under the broadest reasonable interpretation (BRI), “deactivating a node” reasonably encompasses removing, pruning, disabling, or excluding a node/feature from consideration during classification. Kumari teaches reducing the number of features used in the decision trees by pruning historical data/features. Because decision tree nodes correspond to features, pruning features necessarily result in associated decision nodes that no longer being considered during classification. Thus, Kumari teaches or at least suggests the “deactivating a node” limitation.
Further, the amended claim recites tracking features over time, and responding to a determination that a feature fails a frequency of usage threshold by preventing consideration of the corresponding node in subsequent classification. Applicant’s argument that Kumari performs pruning “prior to generating the training dataset” is not persuasive because the claim does not exclude the preprocessing implementations, the claim does not require runtime-only modification, and the claim does not require a particular mechanism for deactivation.
For at least the reasons listed above, the amended claim 1 and its corresponding claims 10 and 18 are not patentable over the combination of the cited references. Dependent claims 2-9, 11-17 and 19-22, those either directly or indirectly depended on the independent claims, are not patentable for the same reasons.
Conclusion
THIS ACTION IS MADE FINAL. See MPEP 706.07(a). Applicant is remined 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 filled within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any extension fee pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Quoc Phung whose telephone number is (703) 756 1330. The examiner can normally be reached on Monday through Friday from 9am to 5pm PT.
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) athttp://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Jennifer Welch can be reached on 571-272-7212. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/Q.L.P./Examiner, Art Unit 2143
/JENNIFER N WELCH/Supervisory Patent Examiner, Art Unit 2143
Prosecution Insights