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 .
Response to Arguments
Applicant’s arguments with respect to claim(s) 1, 6-8, 13-15, and 20 on page 8 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
Applicant’s arguments with respect to claim(s) 2, 9, and 16 on page 9 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
Applicant’s arguments with respect to claim(s) 3-4, 10-11, and 17-18 on page 9 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
Applicant’s arguments with respect to claim(s) 5, 12, and 19 on page 10 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 1, 6-8, 13-15, and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Owen et al. (US 20200279025 A1) in view of Chowdhary et al. (US 10332032 B2) and further in view of Lei et al. (US 20190130425 A1).
As to claim 1, Owen teaches a method comprising:
receiving, at a computer system, a plurality of data sets, wherein each data set within the
plurality of data sets is associated with at least one data type within a plurality of data types; (see Owen paragraph [0017] "the system 100 includes a device 101 arranged to receive data 102 from one or more data sources.", and see Owen paragraph [0018] "The data 102 includes training data for generating models as described in more detail below. The data 102 may be any type of data. For example, the data 102 may be temporal or sequential data such as video data, audio data, image data, sensor data, an electronic document, financial data such as pricing and sales data, or any other type of suitable data recorded at various frequencies (a static frequency or a varying frequency). In some embodiments, the data 102 includes (i) the sales data for one or more locations, which fulfils the role of the dependent variable (i.e., the variable which is to be predicted) and (ii) non-sales data, which describes the attributes of the location associated with the sales data (site).", and see Owen paragraph [0025] "the method 300 includes receiving data including a plurality of data points representing training data (at block 302). As noted above, the data 102 includes training data for generating models and can include any type of data.")
training, via at least one processor of the computer system using the plurality of data sets, a plurality of machine learning models, (see Owen paragraph [0015] "some embodiments described herein may include one or more electronic processors configured to perform the described functionality by executing instructions stored in non-transitory, computer-readable medium.", and see Owen paragraph [0026] "the method 300 also includes generating a plurality of models based on the plurality of data points (at block 304). Each of the plurality of models uses at least one a different modelling methodology or different training data. For example, in some embodiments, each of the plurality of models is generated using at least one selected from a group consisting of the plurality of data points, a subset of the plurality, a subspace of the plurality of data points, and a transformed version of at least a portion of the plurality of data points.")
wherein each machine learning model in the plurality of machine learning models is configured to, upon execution, generate a transportation prediction; and (see Owen paragraph [0029] "the method 300 also includes receiving an input representing a requested prediction (at block 306) and generating an output based on the input using each of the plurality of models (at block 308). The input represents the requested prediction and, in some embodiments, is in the same format as used for training the created plurality of models.", and see Owen paragraph [0030] "the output generated by each of the plurality of models includes a predicted sales volume based on the hypothesised site attributes.”)
wherein each machine learning model in the plurality of machine learning models is trained using a data set within the plurality of data sets; (see Owen paragraph [0026] "Each of the plurality of models uses at least one a different modelling methodology or different training data.", and see Owen paragraph [0027] "In some embodiments, the training data is portioned (e.g., subsetted) as part of creating the plurality of models using random sampling.")
executing, via the at least one processor, the plurality of machine learning models using the real-time data type values as inputs, resulting in a plurality of predictions; and (see Owen paragraph [0029] "the method 300 also includes receiving an input representing a requested prediction (at block 306) and generating an output based on the input using each of the plurality of models (at block 308). The input represents the requested prediction and, in some embodiments, is in the same format as used for training the created plurality of models.", and see Owen paragraph [0030] "the output generated by each of the plurality of models includes a predicted sales volume based on the hypothesised site attributes.")
generating, via the at least one processor aggregating the plurality of predictions, a final prediction, (see Owen paragraph [0031] "The output generated by each of the plurality of models and the training data is then provided to a machine learning algorithm (at block 310). This machine learning algorithm is referred to herein as the second-level model as compared to the first-level models created based on the original training data (at block 304). Accordingly, the second-level model is trained using the outputs from the first-level models.", and see Owen paragraph [0033] "The machine learning algorithm generates the requested prediction using the output generated by each of the plurality of models and the training data (at block 312)")
Owen does not explicitly teaches “a transportation;”, and “receiving, at the computer system, real-time data type values associated with the plurality of data types;”
However, Chowdhary teaches,
wherein each machine learning model in the plurality of machine learning models is configured to, upon execution, generate a transportation prediction; and (see Chowdhary paragraph [Col 7 L 43] "At 612, based on the price elasticity curve and current network traffic, cargo transportation price associated with each of the subgroups may be determined. For example, as described above, an optimization function may be executed that maximizes revenue and win probability of a price quote with a penalty for heavy traffic.")
receiving, at the computer system, real-time data type values associated with the plurality of data types; (see Chowdhary paragraph [Col 7 L 65] "In some cases, the technology of the present disclosure may be used as part of a system of devices that tracks the current state of the logistic network. Tracking data may include the location of vehicles and/or specific loads. The method and system in one embodiment improves on existing tracking technology by continually observing load in the network and determining optimal prices for requested routes in real time based on network state.")
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the invention of Owen to include transportation-specific, real-time network-state inputs to so that the plurality of models would generate a transportation prediction based on current transportation conditions. Chowdhary [Col 7 L 59].
Owen as modified by Chowdhary does not explicitly teach "wherein the aggregating weights each prediction by a weight based on accuracy of each model in the plurality of machine learning models during previous iterations"
However; Lei teaches
wherein the aggregating weights each prediction by a weight based on accuracy of each model in the plurality of machine learning models during previous iterations. (see Lei paragraph [0044] "At 212, the root-mean-square error (“RMSE”) R(i) is determined for each of the forecasts F(i) for the periods from k-n to k−1 (i.e., the weeks prior to the forecast period), or weeks 90-99 in the example of FIGS. 3 and 4). In one example, the RMSE is determined by starting the forecast “n” weeks earlier than the present time period (i.e., the last period for which there are sales). Therefore, for these n weeks there is both sales data and a forecast that can be generated. With sales data and the forecast, the RMSE can be calculated using known RMSE calculations.", and see Lei paragraph [0045] "At 214, a weight is determined/assigned for each method/trained model “i” based on the RMSE determined at 212", and see Lei paragraph [0046] "At 216, the final forecast is determined by combining each weighted value for each trained model of the previously determined forecasts.").
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the invention of Owen-Chowdhary to include accuracy-based weights for each model's prediction so that the aggregation would generate a final prediction that gives greater influence to the historically more accurate models. Lei [0016].
As to claim 6, Owen as modified by Chowdhary teaches,
wherein the final prediction is a freight transportation price. (See Chowdhary [4] “based on the price elasticity curve and current network traffic, determine cargo transportation price associated with each of the subgroups.”, and see Chowdhary [8] “A system, method and computer program product that implements an automated pricing engine for spot pricing ... in a global logistics network may be provided.”, and see Chowdhary [Col 2 L 55] “A carrier may view optimal price by route”)
As to claim 7, Owen as modified by Chowdhary teaches,
wherein the plurality of data types comprise: holiday information, seasonal data, and fuel price data. (See Chowdhary [Col 3 L 18] “Quotes may be differentiated by ... a variety of special handling requirements, along with seasonal, market, and customer characteristics.”, and see Chowdhary [claim 1] “... influencing factors comprise at least ... seasonality indices indicating CWT demand relative to average CWT ...”
As to claim 8, Owen as modified by Chowdhary teaches,
at least one processor; and (see Owen paragraph [0015] "described herein may include one or more electronic processors configured to perform the described functionality by executing instructions stored in non-transitory")
a non-transitory computer-readable storage medium having instructions which, when executed by the at least one processor, cause the at least one processor to perform operations comprising: (see Owen paragraph [0015] "“non-transitory computer-readable medium” comprises all computer-readable media but does not consist of a transitory, propagating signal. Accordingly, non-transitory computer-readable medium may include, for example, a hard disk, a CD-ROM, an optical storage device, a magnetic storage device, a ROM (Read Only Memory), a RAM (Random Access Memory), register memory, a processor cache, or any combination thereof.")
receiving a plurality of data sets, wherein each data set within the plurality of data sets is associated with a different data type within a plurality of data types; (see Owen paragraph [0017] “the system 100 includes a device 101 arranged to receive data 102 from one or more data sources.”, and see Owen paragraph [0018] “The data 102 includes training data for generating models as described in more detail below. The data 102 may be any type of data. For example, the data 102 may be temporal or sequential data such as video data, audio data, image data, sensor data, an electronic document, financial data such as pricing and sales data, or any other type of suitable data recorded at various frequencies (a static frequency or a varying frequency). In some embodiments, the data 102 includes (i) the sales data for one or more locations, which fulfils the role of the dependent variable (i.e., the variable which is to be predicted) and (ii) non-sales data, which describes the attributes of the location associated with the sales data (site).”, and see Owen paragraph [0025] “the method 300 includes receiving data including a plurality of data points representing training data (at block 302). As noted above, the data 102 includes training data for generating models and can include any type of data. However, in some embodiments, the training data includes a plurality of data points, wherein each data point includes (i) sales data for a location, which fulfils the role of the dependent variable (i.e., the variable which is to be predicted) and (ii) non-sales data, which describes the attributes of the location (site).”, and see Chowdhary [18] “Examples of price influencers may include, but are not limited to, CWT, leadtime, number of pieces, number of stops, historical win rates, type of consumers, type of products, type of freights, competition, seasonality, and week-end effects. Price influencers may be correlated, e.g., CWT and number of pieces, freighter and week-end. L-1 penalized method (LASSO) may be executed to reduce correlation, and a p-value threshold may be executed to filter out statistically insignificant price influencers.”)
training, using the plurality of data sets, a plurality of machine learning models, (see Owen paragraph [0015] “executable by one or more electronic processors to perform the described functionality”, and see Owen paragraph [0025] “the method 300 includes receiving data including a plurality of data points representing training data (at block 302). As noted above, the data 102 includes training data for generating models and can include any type of data. However, in some embodiments, the training data includes a plurality of data points, wherein each data point includes (i) sales data for a location, which fulfils the role of the dependent variable (i.e., the variable which is to be predicted) and (ii) non-sales data, which describes the attributes of the location (site).”, and see Owen paragraph [0026] “the method 300 also includes generating a plurality of models based on the plurality of data points (at block 304). Each of the plurality of models uses at least one a different modelling methodology or different training data.”)
wherein each machine learning model in the plurality of machine learning models is configured to, upon execution, generate a transportation prediction; and (see Owen paragraph [0029] “the method 300 also includes receiving an input representing a requested prediction (at block 306) and generating an output based on the input using each of the plurality of models (at block 308). The input represents the requested prediction and, in some embodiments, is in the same format as used for training the created plurality of models.”, and see Owen paragraph [0030] “the output generated by each of the plurality of models includes a predicted sales volume based on the hypothesised site attributes.”, and see Chowdhary [32] “cargo transportation price associated with each of the subgroups may be determined. For example, as described above, an optimization function may be executed that maximizes revenue and win probability of a price quote with a penalty for heavy traffic.”)
wherein each machine learning model in the plurality of machine learning models is trained using a data set within the plurality of data sets;(see Owen paragraph [0026] “Each of the plurality of models uses at least one a different modelling methodology or different training data.”)
receiving real-time data type values associated with the plurality of data types; (see Chowdhary [34] “The method and system … improve on existing computer technology for cargo pricing by adding automation that enables the computer to recommend optimal prices for all routes in a network. This improves on existing systems which can capture pricing data but does not generate and provide optimal prices. In some cases, the technology of the present disclosure may be used as part of a system of devices that tracks the current state of the logistic network. Tracking data may include the location of vehicles and/or specific loads. The method and system in one embodiment improves on existing tracking technology by continually observing load in the network and determining optimal prices for requested routes in real time based on network state.”, and see Owen paragraph [0035] “The stacked machine learning algorithms can also adapted for online learning. For example, when the learning algorithms used for the first-level and second-level models are capable of online-learning, then as new observations are made”)
executing the plurality of machine learning models using the real-time data type values as inputs, resulting in a plurality of predictions; and (see Owen paragraph [0029] “the method 300 also includes receiving an input representing a requested prediction (at block 306) and generating an output based on the input using each of the plurality of models (at block 308). The input represents the requested prediction and, in some embodiments, is in the same format as used for training the created plurality of models.”, and see Owen paragraph [0030] “the output generated by each of the plurality of models includes a predicted sales volume”)
generating, by aggregating the plurality of predictions, a final prediction. (See Owen paragraph [0031] “The output generated by each of the plurality of models and the training data is then provided to a machine learning algorithm (at block 310). This machine learning algorithm is referred to herein as the second-level model as compared to the first-level models created based on the original training data (at block 304). Accordingly, the second-level model is trained using the outputs from the first-level models.”, and see [0032] “the machine learning used is non-linear in the inputs, as otherwise the second stage reduces to a simple weighted average of the inputs.”, and see Owen paragraph [0032] “the machine learning used is non-linear in the inputs, as otherwise the second stage reduces to a simple weighted average of the inputs.”, and see Owen paragraph [0033] “The machine learning algorithm generates the requested prediction using the output generated by each of the plurality of models and the training data (at block 312)”)
As to claim 13, this is directed to a system that corresponds to the method of claim 6, See the rejection for claim 6 above, which also applies to claim 13.
As to claim 14, this is directed to a system that corresponds to the method of claim 7, See the rejection for claim 7 above, which also applies to claim 14.
As to claim 15, this is directed to a computer program product that corresponds to the method of claim 1, See the rejection for claim 1 above, which also applies to claim 15. also see Owen paragraph [0015] " non-transitory, computer-readable medium storing instructions executable by one or more electronic processors to perform the described functionality."
As to claim 20, this is directed to a computer program product that corresponds to the method of claim 6, See the rejection for claim 6 above, which also applies to claim 20.
Claim(s) 2, 9, and 16 is/are rejected under 35 U.S.C. 103 as being unpatentable over Owen et al. (US 20200279025 A1) in view of Chowdhary et al. (US 10332032 B2) and further in view of Lei et al. (US 20190130425 A1) and further in view Morshed-Bozorgdel et al. (A Novel Framework Based on the Stacking Ensemble Machine Learning (SEML) Method: Application in Wind Speed Modeling, 8 May 2022).
As to claim 2, Owen as modified by Chowdhary teaches the method of claim 1,
Chowdhary teaches,
receiving, at the computer system, past industry-specific indicators; (see Chowdhary [Col 3 L 18] “Quotes may be differentiated by weight, volume, urgency, number of pieces, and a variety of special handling requirements, along with seasonal, market, and customer characteristics.”, and see [claim 15] “determine cargo transportation price … wherein the influencing factors comprise at least … customer industry … seasonality indices … average price quoted to customer in market, average price quoted to customer overall, customer percent (%) revenue through spot market …”)
receiving, at the computer system, current industry-specific indicators; and (see Chowdhary [Col 6 L 36] “In one embodiment, the system may price each route based on the network traffic to achieve global pricing optimization balancing demand and capacity through the network; enable real-time differentiated pricing strategy for a variety of cargo services for each local market, incorporating unique inputs such as regional competition mapping, customer's wallet shares, seasonality, routes, and/or other criteria, to provide better global market splits.”)
Chowdhary does not explicitly teaches “training, via the at least one processor using the industry-specific indicators, a machine learning multiplier model;”, “executing, via the at least one processor, the machine learning multiplier model using the current industry-specific indicators as input, resulting in a multiplier,”, “wherein the final prediction is further generated using the multiplier. ”
However, Morshed-Bozorgdel teaches
training, via the at least one processor using the industry-specific indicators, a machine learning multiplier model; (see Morshed-Bozorgdel [abstract] “The SEML method consists of two levels. In particular, eleven machine learning (ML) algorithms in six categories neuron based (artificial neural network (ANN), general regression neural network (GRNN), and radial basis function neural network (RBFNN)), kernel based (least squares support vector machine-grid search (LSSVM-GS)), tree based (M5 model tree (M5), gradient boosted regression (GBR), and least squares boost (LSBoost)), curve based (multivariate adaptive regression splines (MARS)), regression based (multiple linear regression (MLR) and multiple nonlinear regression (MNLR)), and hybrid algorithm based (LSSVM-Harris hawks optimization (LSSVM-HHO)) were selected as the base algorithms in level 1 of the SEML method. In addition, LSBoost was used as a meta-algorithm in level 2 of the SEML method. For this purpose, the output of the base algorithms was used as the input for the LSBoost.”, and see Morshed-Bozorgdel [2.1. Present Work Steps] “Level 2: Due to the simple structure and high speed of the LSBoost, this algorithm was used as a meta-algorithm to combine the results of the base algorithms.”
executing, via the at least one processor, the machine learning multiplier model using the current industry-specific indicators as input, resulting in a multiplier, (see Morshed-Bozorgdel [abstract] “For this purpose, the output of the base algorithms was used as the input for the LSBoost.”)
wherein the final prediction is further generated using the multiplier. (see Morshed-Bozorgdel [2. Materials and Methods] “Level 2: Due to the simple structure and high speed of the LSBoost, this algorithm was used as a meta-algorithm to combine the results of the base algorithms.”
Note: second-stage model trained to combine outputs from multiple first-stage models.
it would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the invention of Owen-Chowdhary to include a second-stage machine-learning combiner model that is trained on the outputs of multiple first-stage models and used to generate the final prediction from those combined outputs because Morshed-Bozorgdel expressly teaches that a two-level ensemble architecture improves predictive performance by using a meta-algorithm to combine the outputs of multiple base algorithms, and further explains that “due to the simple structure and high speed of the LSBoost, this algorithm was used as a meta-algorithm to combine the results of the base algorithms.”
As to claim 9, this is directed to a system that corresponds to the method of claim 2, See the rejection for claim 2 above, which also applies to claim 9.
As to claim 16, this is directed to a computer program product that corresponds to the method of claim 2, See the rejection for claim 2 above, which also applies to claim 16.
Claim(s) 3-4, 10-11, and 17-18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Owen et al. (US 20200279025 A1) in view of Chowdhary et al. (US 10332032 B2) and further in view of Lei et al. (US 20190130425 A1) and further in view Morshed-Bozorgdel et al. (A Novel Framework Based on the Stacking Ensemble Machine Learning (SEML) Method: Application in Wind Speed Modeling, 8 May 2022) and further in view of Lu et al. (Learning under Concept Drift: A Review, 18 October 2018).
As to claim 3, Owen-Chowdhary as modified by Morshed-Bozorgdel teaches the method of claim 2,
Owen-Chowdhary, and Morshed-Bozorgdel do not explicitly teach “receiving, at the computer system after generating the final prediction, an actual result;”, “determining, via the at least one processor, that the actual result is distinct from the final prediction, resulting in a difference; ”, “retraining, via the at least one processor, the machine learning multiplier model using the industry-specific indicators and the difference, resulting in an updated machine learning multiplier model; and ”, “retraining, via the at least one processor, the plurality of machine learning models using the plurality of data sets and the difference, resulting in an updated plurality of machine learning models, ”, and “wherein the updated machine learning multiplier model and the updated plurality of machine learning models are used in future iterations. ”
However, Lu teaches
receiving, at the computer system after generating the final prediction, an actual result; (see Lu [CONCLUSIONS: FINDINGS AND FUTURE DIRECTIONS] “Most existing drift detection and adaptation algorithms assume the ground true label is available after classification/prediction…”)
determining, via the at least one processor, that the actual result is distinct from the final prediction, resulting in a difference; (see Lu [introduction] “If the concept drift occurs ... the induced pattern of past data may not be relevant to the new data, leading to poor predictions…”, and see Lu [5.1 Training New Models for Global Drift] “When the classification error rate increases … more nodes are added to the network layers…”)
retraining, via the at least one processor, the machine learning multiplier model using the industry-specific indicators and the difference, resulting in an updated machine learning multiplier model; and (see Lu [5 DRIFT ADAPTATION] “This section focuses on strategies for updating existing learning models according to the drift ... There are three main groups of drift adaptation methods, namely simple retraining, ensemble retraining and model adjusting…”
retraining, via the at least one processor, the plurality of machine learning models using the plurality of data sets and the difference, resulting in an updated plurality of machine learning models, (see Lu [5.1 Training New Models for Global Drift] “retrain a new model with the latest data to replace the obsolete model ... A small window can better reflect the latest data distribution, but a large window provides more data for training a new model”
wherein the updated machine learning multiplier model and the updated plurality of machine learning models are used in future iterations. (See Lu [5.1 Training New Models for Global Drift] “the stable learner will be replaced with the reactive learner.”)
it would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the invention of Owen-Chowdhary, and Morshed-Bozorgdel to include receiving actual post-prediction results, determining prediction error from the difference between predicted and actual outcomes because Lu expressly teaches Section 5, “DRIFT ADAPTATION,” that predictive models should be updated after prediction when true outcomes become available and model performance degrades due to concept drift.
As to claim 4, Owen-Chowdhary-Morshed-Bozorgdel as modified by Lu teaches the method of claim 3,
wherein the retraining of the machine learning multiplier model occurs more frequently than the retraining of the plurality of machine learning models. (See Owen [0035] “This type of online learning may be particularly useful when observations may occur on a short timescale, such as observations that may occur at microsecond or nanosecond intervals.”)
As to claim 10, this is directed to a system that corresponds to the method of claim 3, See the rejection for claim 3 above, which also applies to claim 10.
As to claim 11, this is directed to a system that corresponds to the method of claim 4, See the rejection for claim 4 above, which also applies to claim 11.
As to claim 17, this is directed to a computer program product that corresponds to the method of claim 3, See the rejection for claim 3 above, which also applies to claim 17.
As to claim 18, this is directed to a computer program product that corresponds to the method of claim 4, See the rejection for claim 4 above, which also applies to claim 18.
Claim(s) 5, 12, and 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Owen et al. (US 20200279025 A1) in view of Chowdhary et al. (US 10332032 B2) and further in view of Lei et al. (US 20190130425 A1) and further in view Morshed-Bozorgdel et al. (A Novel Framework Based on the Stacking Ensemble Machine Learning (SEML) Method: Application in Wind Speed Modeling, 8 May 2022) and further in view of Lu et al. (Learning under Concept Drift: A Review, 18 October 2018), and further in view Stauder et al. (AI for in-line vehicle sequence controlling: development and evaluation of an adaptive machine learning artifact to predict sequence deviations in a mixed-model production line, 15 August 2021).
As to claim 5 Owen-Chowdhary, and Morshed-Bozorgdel as modified by Lu teaches the method of claim 3,
Owen-Chowdhary-Morshed-Bozorgdel-Lu do not explicitly teach “wherein the retraining of the machine learning multiplier model and the retraining of the plurality of machine learning models occur upon a predetermined number of iterations of the generating of the final prediction occurring.”.
However, Stauder teaches
wherein the retraining of the machine learning multiplier model and the retraining of the plurality of machine learning models occur upon a predetermined number of iterations of the generating of the final prediction occurring. (See Stauder [5.8 Model error estimation] “The model makes batchwise predictions … We decided to evaluate continual relearning with a fixed trigger that sets off relearning after receiving C new instances of the minority class.”
it would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the invention of Owen-Chowdhary- Morshed-Bozorgdel-Lu to include a fixed, count-based relearning trigger so that retraining of both the second-stage model and the first-stage models occurs after a preset number of prediction cycles to combine Owen-Chowdhary- Morshed-Bozorgdel-Lu with Stauder because Stauder expressly teaches the benefit of continual relearning with a fixed trigger, stating in Section 5.8, “Model error estimation,” that: “The model makes batchwise predictions … We decided to evaluate continual relearning with a fixed trigger that sets off relearning after receiving C new instances of the minority class.” to improve control over update timing, computational efficiency, and consistency of model maintenance, with a reasonable expectation of success.
As to claim 12, this is directed to a system that corresponds to the method of claim 5, See the rejection for claim 5 above, which also applies to claim 12.
As to claim 19, this is directed to a computer program product that corresponds to the method of claim 5, See the rejection for claim 5 above, which also applies to claim 19.
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ABDULLAH K ABOUD whose telephone number is (571)272-0025. The examiner can normally be reached Mon-Fri 8am-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, Li B Zhen, can be reached at (571) 272-3768. 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.
/ABDULLAH KHALED ABOUD/Examiner, Art Unit 2121
/Li B. Zhen/Supervisory Patent Examiner, Art Unit 2121
