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 .
Continued Examination
A Notice of Appeal and a Pre-Appeal Brief Request for Review were previously filed in this application. Upon review of the rejections of record and Applicant’s arguments, the Pre-Appeal Brief Conference panel determined that prosecution should be reopened for further examination. Accordingly, prosecution is hereby reopened, and the finality of the previous office action is withdrawn. Claims 1-2, 5-7, 10-12 and 15 remain pending in the application. The prior rejections of the pending claims under 35 U.S.C. § 103 set forth in the previous Final Office Action are hereby withdrawn. Withdrawal of those rejections should not be construed as an indication that the presently claimed subject matter has been found patentable. Rather, the previous grounds of rejection are superseded by the new grounds of rejection set forth below, which reflect further consideration of Applicant’s arguments and the additional prior art now of record.
Response to Arguments
Applicant's arguments filed May 29, 2026 with the Pre-Appeal Brief Request for Review have been fully considered but they are not persuasive in view of the newly applied grounds of rejection.
Applicant argues that Nelson’s dynamically generated and continuously updated process model does not disclose the claimed pretrained model because such a model is not trained from historical data before deployment. The Examiner agrees that Nelson alone does not expressly disclose pretraining in the manner argued by Applicant. The present rejection, however, no longer relies upon Nelson alone for this limitation. Cohen expressly teaches training a neural-network process model from off-line stored historical process data before the trained model is employed for control of a continuous industrial process. Cohen therefore directly addresses the distinction identified by Applicant between a model trained before deployment and Nelson’s adaptive model.
Applicant further argues that Nelson relies upon numerous heterogenous sensor inputs and does not disclose controlling the process using data gathered solely from a hardness analyzer sensor and a shaft speed sensor. The present rejection does not rely upon Nelson alone for the selection of those particular inputs. Zhang and Liu establish that hardness-related calcium concentration affects induction time and coal floatability, while Koh and Smith establish that agitation speed affects the same particle-bubble attachment behavior. Cohen further teaches that the particular variables supplied to a trained industrial process model may be selectively chosen and that the model may employ a very small number of variables, including fewer than five. Accordingly, the present rejection provides an articulated reason why one of ordinary skill would have selected the hardness-related and agitation-speed variables particularly relevant to the control relationship being optimized. Applicant’s additional arguments regarding deep-learning architectures, automatic hierarchical feature learning, computational resources, and manual feature engineering are not commensurate in scope with the pending claims, which do not require any particular deep-learning architecture, feature-learning technique, dataset size, or computational infrastructure.
With respect to claims 5, 10, and 15, Applicant correctly argued that Yan’s Bayesian Network is not the same as Bayesian optimization. The prior rejection relying upon Yan for that limitation has therefore been withdrawn. The present rejection instead relies upon Adams, which expressly teaches Bayesian optimization in connection with machine-learning systems, including neural networks, and teaches using Bayesian optimization to identify parameter values corresponding to improved performance of the trained system. Thus, deficiency identified by Applicant with respect to Yan is not present in the newly applied rejection. Applicant also argues that claim 5 requires a specific functional relationship in which the neural network constructs the objective function used by Bayesian optimization and that Nelson teaches away from such a combination by permitting selection of an individual analysis technique. These arguments are not persuasive. Pending claim 5 recites only that the pretrained model is “trained using a neural network technique and Bayesian optimization technique”; it does not recite that the neural network constructs the objective function for the Bayesian optimization technique. Arguments directed to such an unclaimed limitation are therefore not commensurate in scope with the claim. Further, Nelson does not criticize, discredit, or discourage combinations of analytical techniques; rather Nelson expressly permits use of “one or all” of several advanced analysis techniques and expressly contemplates combinations thereof. In any event, the present rejection relies upon Cohen for neural-network training and Adams for Bayesian optimization, and the reasons for combining those teachings with the modified Nelson system are set forth in the rejection below.
Accordingly, although the prior Yan-based rejection has been withdrawn and prosecution reopened, Applicant’s remaining arguments do not overcome the new grounds of rejection set forth in this Office action.
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 nonobviousness.
Claims 1-2, 6-7 and 11-12 are rejected under 35 U.S.C. 103 as being unpatentable over Nelson (WO-97/45203-A1) in view of Koh and Smith, "The Effect of Stirring Speed and Induction Time on Flotation", further in view of Zhang and Liu, "Effect of Calcium Ions on Induction Time Between a Coal Particle and Air Bubble", further in view of Gui et al., "Flotation process design based on energy input and distribution", and further in view of Cohen (US-20190236447-A1) hereafter collectively referred to as "modified Nelson".
Regarding claim 1, Nelson discloses a method of controlling a froth floatation process comprising receiving process and equipment information corresponding to water and an agitator involved in a froth flotation process. In particular, Nelson discloses a computerized monitoring and control system receiving information from process sensors and equipment sensors, including chemical or mineralogical composition, pH, and the rotational speed of the agitation mechanism (Nelson p. 12-14). Nelson further discloses analyzing sensed process and equipment values to determine operating conditions used to achieve desired flotation performance, wherein the control computer uses the sensed information and an internal process model to actuate control devices and adjust flotation-machine operation (Nelson p. 15-18 and p. 21). Nelson specifically identifies rotational speed of the agitation mechanism as an equipment parameter that is monitored and acted upon and teaches that operation of the flotation machine may thereby be adjusted, changed and preferably optimized. Nelson further discloses generating and maintaining a process model from historical flotation-operation data. Nelson explains that the computerized monitoring and control system may monitor various parameters with respect to time and thereby provide a detailed historical record of flotation-machine operation, and that this historical record may be used by the control computer to model flotation-machine operation, adjust models for flotation-machine operation, and generally learn how the flotation machine behaves in response to changes in various inputs (Nelson p. 15). Nelson additionally describes the information developed by the system as heuristic knowledge gained through operation of the flotation system and teaches that its process models may adapt based upon operational data (Nelson p. 17-18). Thus, Nelson teaches receiving and storing past values of sensed process variables and equipment variables, including chemical/mineralogical information and agitation speed, capturing process-performance information associated with those values, identifying operating conditions associated with desired flotation performance, and controlling the flotation process by adjustment of operational parameters including agitation speed.
Nelson, however, does not expressly disclose receiving a water hardness value from a hardness analyzer, analyzing agitator speed based specifically upon water-hardness, or controlling the process using data gather solely from a hardness analyzer sensor and a shaft speed sensor. Nelson also does not expressly identify the claimed target parameter values as an ash rejection percentage value, a combustible recovery percentage value, and an efficiency index value. Further, although Nelson teaches generating, learning, and adapting process models from historical operating information, Nelson does not expressly disclose that the process model is pretrained from stored historical process data before being implemented for process control.
Zhang and Liu disclose the known relationship between hardness related calcium-ion concentration and flotation performance. Zhang and Liu teach that particle-bubble attachment is essential to successful flotation and that induction time is strongly dependent upon the solution chemistry of the flotation system. Zhang and Liu experimentally demonstrate that the induction time between a coal particle and an air bubble increases as calcium-ion concentration increases, reporting an induction time of 17 ms in deionized water and 32 ms at a calcium-ion concentration of 1 mmol. L. Zhang and Liu further explain that the presence of calcium ions produces hydrophilic calcium-containing material on the coal surface, thereby weakening the hydrophobicity and floatability of the coal particles. Zhang and Liu therefore establish that calcium-ion concentration, a constituent of water hardness, was a known and relevant water-chemistry parameter affecting induction time and coal flotation performance.
Koh and Smith disclose the corresponding relationship between agitation speed and particle-bubble attachment. Koh and Smith teach that, for hydrophobic particles having short induction times, the particle-bubble attachment rate increases with increasing stirring speed because of increased collision rates, whereas for particles having lower hydrophobicity and longer induction times, attachment rate decreases as stirring speed becomes excessive. Koh and Smith therefore establish that agitation speed is another known result-effective variable affecting the same induction-time and particle-attachment behavior affected by the water-chemistry described by Zhang and Liu.
Gui further discloses controlling flotation energy input by adjusting flotation shaft speed and correlating shaft speed with flotation performance. Gui teaches that concentrate combustible matter recovery and ash content are primary evaluation indexes for a flotation process and uses a flotation efficiency index to evaluate combinations of those flotation results. Gui additionally demonstrates that different shaft-speed profiles produce different flotation efficiencies and identifies shaft-speed conditions suitable for improving flotation performance. Gui therefore establishes that agitator shaft speed was conventionally varied to optimize recognized flotation-performance parameters including ash content, combustible matter recovery, and flotation efficiency.
Cohen discloses generating a controller for a continuous industrial process using a model trained from historical process data before the trained model is employed in controlling the process. Cohen received from storage memory off-line stored values of controlled variables, manipulated variables, and in embodiments, disturbance variables of a continuous process over a plurality of time points; uses the stored historical values to train a neural network predictor of the process; trains a controller using the trained predictor; and thereafter employs the trained controller to control the continuous process (Cohen abstract and pars. [0017-0025]). Cohen explains that the method is applicable to a continuous production process in an industrial plant and that the predictor may be trained entirely from off-line historical plant data, including data collected and stored days, weeks, months or years before training begins (Cohen pars. [0037-0045]). Cohen further teaches testing the trained predictor against historical data and determining that the predictor is ready for operation before deployment (Cohen pars. [0057-0060]). Thus, Cohen expressly teaches the use of a pretrained process model generated from a plurality of past process-variable values and subsequently implemented for control of a continuous industrial process. Cohen additionally teaches that the particular variables supplied to the trained process model are selected according to the particular process. Cohen explains that the number of variables represented at each time point may be very small, including few than ten or even fewer than 5 variables (Cohen par. [0046]), and further teaches that, during model development, the specific controlled, manipulated, and/or disturbance variables used by the predictor may be adjusted when testing and refining the trained model (Cohen par. [0060]). Cohen therefore establishes that selecting a limited subset of the process variables most relevant to the desired control relationship, rather than supplying every available plant variable to the model, was a known implementation of model-based continuous process control.
It would have been obvious to one of ordinary skill in the art prior to the effective filing date of the claimed invention to modify Nelson’s historical-data, model-based flotation control method according to the teachings of Zhang and Liu, Koh and Smith, Gui and Cohen by using hardness-related water chemistry and agitation speed as the selected model inputs, training the process model from previously stored values of those inputs before deployment, and using the trained model to determine and adjust agitation speed to optimize the recognized flotation-performance targets of ash rejection, combustible recovery, and flotation efficiency. Nelson already provides the sensor-based flotation-control architecture, historical operating record, process modeling, and adjustment of agitation speed; Zhang and Liu establish that hardness-related calcium-concentration affects induction time and coal floatability; Koh and Smith establish that agitation speed affects the same induction-time and particle-attachment behavior; Gui establishes that shaft speed is adjusted to optimize ash, combustible recovery, and efficiency results; and Cohen teaches both pretraining a continuous process control model from stored historical process values before deployment and deliberately selecting a limited set of process variables relevant to the particular control objective. A person of ordinary skill in the art seeking to improve Nelson’s control of flotation performance therefore would have had reason to train Nelson’s model in advance using the historical hardness-related and agitator-speed information already relevant to the process, and to use those two directly related variables as the model inputs for determining an agitation speed responsive to the prevailing water condition, because doing so predictably applies known relationships between water chemistry, agitation, induction time, and flotation efficiency while permitting the controller to begin operation with a model already trained on prior process behavior. The combination would have had a reasonable expectation of success because each modification uses the respective references according to their established functions of measuring a flotation-affecting water chemistry condition, adjusting a known flotation-affecting operating variable, evaluating established flotation-performance metrics, and training an industrial-process controller from historical values of the variables selected for that control relationship, to obtain the predictable result of model-based adjustment of agitator speed in response to water hardness to improve or maintain desired flotation performance.
Regarding claim 2, modified Nelson teaches the method of claim 1, wherein the agitator speed value ("rotational speed of agitation mechanism" p.12 line 43) is measured by a shaft speed sensor communicatively coupled to the agitator. ("equipment parameters" and "equipment sensors" [p.14 line 1] and [Fig. 2 #44]) also described as "impeller speed" (p. 22 line 22) and corresponds to the agitator speed value. Nelson further teaches the use of these equipment sensors communicatively connected with the control system for obtaining the equipment parameters used to monitor and control operation of the flotation machine (Nelson p. 14 and Fig. 2).
Regarding claim 6, modified Nelson, as discussed above with respect to claim 1, discloses or renders obvious the corresponding control system for controlling a froth floatation process with data gathered solely from a hardness analyzer sensor and a shaft speed sensor (by suggestion of Zhang and Liu, Koh and Smith, for further explanation see rejection of claim 1 above), the control system comprising: a processor (Nelson pg. 19 line 2); and a memory communicatively coupled to the processor (Nelson pg. 10 line 26), wherein the processor is configured to: receive a water hardness value (Nelson pg. 21 par. 2 teaches equipment sensors for monitoring/measuring a variety of values, while Zhang and Liu provides motivation to configure the processor to receive a water hardness value) and an agitator speed value corresponding to water and an agitator involved in the froth floatation process (Nelson pg. 12 line 43 provides the parameter sensed while Koh and Smith provide the motivation to configure the processor to receive an agitator speed value); wherein the water hardness value is received from a hardness analyzer (LIBS and LIMS sensors[Nelson pg. 21 par. 2] are capable of analyzing chemical and mineralogical composition); analyze the agitator speed value vis-a-vis an optimal speed value required to achieve one or more target parameter values during the froth floatation process based on the water hardness value (the processor of Nelson is capable of receiving the chemical composition from the LIBS or LIMS sensors and adjusting the agitator speed required to meet one or more target parameter values. If an apparatus with elements similar to the elements in the instant application is capable of performing the intended use of the control system described by the instant claim, the intended use will be considered anticipated or made obvious. See MPEP 2114.); and implement a pretrained model, based on the analyzing, to adjust the agitator speed value to the optimal speed value in such a manner that the one or more target parameter values are achieved during the froth floatation process, (In response to one or more of the parameters sensed by the sensors [Nelson #42 and 44 shown in Fig. 2] the operation of the flotation machine and thereby its ultimate efficiency can be adjusted, changed, and preferably optimized [Nelson p.16 lines 22-25]) wherein the processor is configured to: receive a plurality of past water hardness values (via equipment sensors of Nelson by suggestion of Zhang and Lui see above rejection of claim 1) and a plurality of past agitator speed values corresponding to a past froth flotation process (Nelson p. 18 lines 1-7 describes a learning process applied to the models which are based on operational values which are collected by the sensors; Cohen abstract and pars. [0017-0025] teaches training the pretrained process model generated from historical process information to determine and adjust a process-control variable); capture a plurality of parameter values based on the plurality of past water hardness values and the plurality of past agitator speed values (Nelson p. 12 lists the plurality of values captured which includes these values; Cohen abstract and pars. [0017-0025] teaches training the pretrained process model generated from historical process information to determine and adjust a process-control variable); identify the one or more target parameter values among the plurality of parameter values such that the one or more target parameter values optimizes efficiency of the froth floatation process (Nelson p. 18 lines 1-7 describes the data collected as “heuristic knowledge”, which is knowledge gained by learning and doing, which is synonymous with observing the data collected in response to changes within the system and identifying target values to optimize the process); and determine the optimal speed value of the agitator corresponding to the one or more target parameter values (Nelson p. 12 table indicates that agitator speed is one of the many operational values which are monitored, recorded and acted upon), and wherein the one or more target parameter values comprises an ash rejection percentage value, a combustible recovery percentage value, and an efficiency index value (suggested by Gui ([p.66 section 3.2 par. 2] and Fig. 17).
Regarding claim 7, modified Nelson discloses the control system of claim 6, wherein the agitator speed value ("rotational speed of agitation mechanism" p.12 line 43) is measured by a shaft speed sensor communicatively coupled to the agitator. ("equipment parameters" and "equipment sensors" [p.14 line 1] and [Fig. 2 #44]) and the table on page 12 lists many equipment parameters sensed of which listed is the rotational speed of agitation mechanism later described as "impeller speed" (p. 22 line 22) the impeller of which will contain a shaft, the speed of which will coordinate with the rotational speed of the agitation mechanism.)
Regarding claim 11, modified Nelson discloses a non-transitory computer readable medium (central control computer[p.10 line 19]) including instructions stored thereon (computerized, "intelligent" systems for operating, controlling, monitoring and diagnosing various parameters [p.6 lines 4-6]) that when processed by at least one processor (microprocessor [p.19 line 2]) cause a control system to perform operations with data gathered solely from a hardness analyzer sensor and a shaft speed sensor (by suggestion of Zhang and Liu, Koh and Smith) comprising: receiving a water hardness value (via laser-induced breakdown spectroscopy sensors (LIBS) and /or laser induced mass spectroscopy sensors (LIMS sensors)[pg. 21 par. 2]) and an agitator speed value corresponding to water and an agitator involved in the froth floatation process (pg. 12 line 43), wherein the water hardness value is received from a hardness analyzer (LIBS and LIMS sensors [pg. 21 par. 2] are capable of analyzing chemical and mineralogical composition); analyzing the agitator speed value vis-a-vis an optimal speed value required to achieve one or more target parameter values during the froth floatation process based on the water hardness value (the processor of Nelson is capable of receiving the chemical composition from the LIBS or LIMS sensors and adjusting the agitator speed required to meet one or more target parameter values. If an apparatus with elements similar to the elements in the instant application is capable of performing the intended use of the control system described by the instant claim, the intended use will be considered anticipated or made obvious. See MPEP 2114.); and implementing a pretrained model, based on the analyzing (controller actuates at least one control device in response to the data received from the LIBS sensor and an internal process model [p.21 lines 24-25]), to adjust the agitator speed value to the optimal speed value in such a manner that the one or more target parameter values are achieved during the froth floatation process, (In response to one or more of the parameters sensed by the sensors [#42 and 44 shown in Fig. 2] the operation of the flotation machine and thereby its ultimate efficiency can be adjusted, changed, and preferably optimized.[p.16 lines 22-25]) wherein the instructions when processed by the at least one processor cause the control system to perform operations comprising: receiving a plurality of past water hardness values (chemical composition [Nelson p.28 line 1] which is analyzed by (Laser Induced Breakdown Spectroscopy (LIBS) [Nelson p. 21 line 8] by suggestion of Zhang and Lui comprise water hardness values, see rejection of claim 1 above) and a plurality of past agitator speed values corresponding to a past froth flotation process (Nelson pg. 12 line 43 lists that agitator speed as one of the monitored items which is sent to the processor and stored in memory [Nelson p. 10 line 26]); capturing a plurality of parameter values based on the plurality of past water hardness values and the plurality of past agitator speed values; identifying the one or more target parameter values among the plurality of parameter values such that the one or more target parameter values optimizes efficiency of the froth floatation process (Nelson p. 18 lines 1-7 describes the data collected as “heuristic knowledge”, which is knowledge gained by learning and doing, which is synonymous with observing the data collected in response to changes within the system and identifying target values to optimize the process); and determining the optimal speed value of the agitator corresponding to the one or more target parameter values (Nelson p. 12 table indicates that agitator speed is one of the many operational values which are monitored, recorded and acted upon), wherein the one or more target parameter values comprises an ash rejection percentage value, a combustible recovery percentage value, and an efficiency index value (Gui [p.66 section 3.2 par. 2] and Fig. 17).
Regarding claim 12, modified Nelson discloses the medium of claim 11, wherein the agitator speed value is measured (Nelson "rotational speed of agitation mechanism" p.12 line 43 corresponds to the agitator speed value) by a shaft speed sensor communicatively coupled to the agitator. (Nelson "equipment parameters" and "equipment sensors" [p.14 line 1] and [Fig. 2 #44]) and the table on page 12 lists many equipment parameters sensed of which listed is the rotational speed of agitation mechanism later described as "impeller speed" (Nelson p. 22 line 22) which will possess a shaft, the speed of which will coordinate with the rotational speed of the agitation mechanism.)
Claims 5, 10 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Nelson (WO-97/45203-A1) in view of Koh and Smith, "The Effect of Stirring Speed and Induction Time on Flotation", further in view of Zhang and Liu, "Effect of Calcium Ions on Induction Time Between a Coal Particle and Air Bubble", further in view of Gui et al., "Flotation process design based on energy input and distribution", and further in view of Cohen (US-20190236447-A1) hereafter collectively referred to as "modified Nelson" as applied to claim 1 and 6 above, and further in view of Adams (US-20140358831-A1).
Regarding claim 5, modified Nelson discloses the method of claim 1, wherein the pretrained model (Cohen abstract and pars. [0017-0025] teaches training the pretrained process model generated from historical process information to determine and adjust a process-control variable) is trained using a neural network technique. (Nelson p. 10, lines 2-5; and Cohen further explains that the predictor neural network may be trained entirely from historical data stored before training begins and that the neural network is repeatedly trained and tested until it reliably models operation of the controlled plant and is considered ready for operation thus modified Nelson teaches the claimed pretrained model being trained using a neural network technique).
Modified Nelson does not teach the additional use of a Bayesian optimization technique.
Adams expressly teaches Bayesian optimization as a technique for configuring and training machine-learning systems, including neural networks. Adams explains that a machine-learning system includes hyperparameters whose values affect the manner in which parameters are learned during training and the resulting performance of the trained system (Adams par. [0056-0059]). Adams specifically identifies a multilayer neural network as an exemplary machine-learning system having training-related hyperparameters including learning rates, dropout rates, weight norms, hidden-layer sizes, convolutional-kernel sizes and pooling sizes (Adams par. [0058]). Adams further teaches using Bayesian optimization to determine values of such hyperparameters. Adams explains that the relationship between machine-learning hyperparameter values and the resulting performance of the trained machine-learning system is treated as an objective function and that Bayesian optimization is used to identify hyperparameter values corresponding to improved or optimal system performance (Adams par. [0061-0064]). Adams teaches constructing a probabilistic model of the objective function from previous evaluations, using the probabilistic model together with an acquisition utility function to select additional parameter values for evaluation, updating the model based upon the resulting evaluations, and identifying an external value of the objective function. Adams further teaches that the probabilistic model of the objection function may itself comprise a neural network (Adams pars. [0011] and [0024]) and expressly states that its Bayesian optimization techniques are not limited to machine-learning hyperparameter optimization, but may also be applied to objective functions relating parameters of a nonlinear control system to performance of that control system (Adams par. [0077]).
It would have been obvious to one of ordinary skill in the art prior to the effective filing date of the claimed invention to further modify the pretrained neural-network process model of modified Nelson according to Adams by using Bayesian optimization during training to select model or training parameters that improve the performance of the trained model. Modified Nelson, particularly through Cohen, already teaches training a neural network model from historical process data for subsequent continuous-process control, while Adams teaches that the performance of a trained neural network depends upon training-related hyperparameters and that Bayesian optimization provides an efficient known technique for selecting values of those hyperparameters corresponding to improved model performance without requiring exhaustive evaluation of all possible parameter settings. A person of ordinary skill in the art seeking to improve the predictive accuracy and resulting control performance of the pretrained neural-network model therefore would have had reason to employ Adams’s Bayesian optimization technique during training of Cohen’s neural-network process model, with a reasonable expectation of success because Adams expressly teaches Bayesian optimization for neural-network training parameters and further recognizes its applicability to optimization of nonlinear control-system performance.
Regarding claim 10, modified Nelson, as applied to claim 6 above, discloses the control system of claim 6, wherein the pretrained model (Cohen abstract and pars. [0017-0025] teaches training the pretrained process model generated from historical process information to determine and adjust a process-control variable) is trained using a neural network technique (Nelson p. 10, lines 2-5; and Cohen further explains that the predictor neural network may be trained entirely from historical data stored before training begins and that the neural network is repeatedly trained and tested until it reliably models operation of the controlled plant and is considered ready for operation thus modified Nelson teaches the claimed pretrained model being trained using a neural network technique) and Bayesian optimization technique (Adams pars. [0056-0064]).
Regarding claim 15, modified Nelson discloses the medium as claimed in claim 11, wherein the pretrained model (Cohen abstract and pars. [0017-0025] teaches training the pretrained process model generated from historical process information to determine and adjust a process-control variable) is trained using a neural network technique (Nelson p. 10, lines 2-5; and Cohen further explains that the predictor neural network may be trained entirely from historical data stored before training begins and that the neural network is repeatedly trained and tested until it reliably models operation of the controlled plant and is considered ready for operation thus modified Nelson teaches the claimed pretrained model being trained using a neural network technique) and Bayesian optimization technique (Adams pars. [0056-0064]).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to WILLIAM ADDISON GEISBERT whose telephone number is (703)756-5497. The examiner can normally be reached Mon-Fri 7:30-5:00 EDT.
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, Bobby RAMDHANIE can be reached at (571)270-3240. 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.
/W.A.G./ Examiner, Art Unit 1779
/Bobby Ramdhanie/ Supervisory Patent Examiner, Art Unit 1779
Prosecution Insights