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 Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 8/6/2026 has been entered.
Response to Amendment
The amendment filed July 20, 2026 has been entered.
Claims 1, 4-7, 9, and 11-19 remain pending and examined in the application. Claims 2-3, 8, and 10 are canceled.
Applicant’s amendments to the Specification and Claims have overcome each and every objection and 112(b) rejection previously set forth in the Final Office Action mailed May 18, 2026.
Based on Applicant’s amendments and remarks, the previous prior art rejection and 101 rejection have been modified to address the claim amendments.
Claim Objections
Claim 15 is objected to because of the following informalities:
Regarding claim 15, Ln. 9 recites, “the a cost function”, which is grammatically incorrect. The above limitation should be amended to recite, “a cost function” to be grammatically correct. Appropriate correction is required.
Claim Rejections - 35 USC § 101
The text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
Claims 1, 4-7, 9, and 11-19 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The subject matter eligibility test for the claims is shown below:
Subject Matter Eligibility Test, Step 1
Independent claim 1 and its dependent claims are drawn to a method. Independent claim 13 and its dependent claim are drawn to a system. Claim 15, although it incorporates claim 1 by reference, has been treated as an independent claim and is drawn to a method. All claims are therefore drawn to a statutory category.
Subject Matter Eligibility Test, Step 2A Prong One
In Step 2A Prong One, it is determined if the claims recite an abstract idea, law of nature, or natural phenomenon. Independent claims 1, 13, and 15 each recite setting one or more values using a model-based deep reinforcement learning and the consideration of a cost function. This cost function is a mathematical equation, which falls under the mathematical concepts group of abstract ideas. See MPEP 2106.04(a). Further, independent claims 1 and 13 each recite predicting a future behavior of the process system using a neural network to determine one or more prediction values. This a determination/evaluation step, and is a mental process-type abstract idea, that recites using a generic neural network to perform the abstract idea. See MPEP 2106.04(a)(2)(III)(C). Further, independent claims 1 and 13 each recite comparing the one or more prediction values for the one or more operating parameters for future instants to real values obtained at the one or more future instants to determine a prediction quality representing an accuracy of a prediction. This is a determination/evaluation step, and is a mental process-type abstract idea. Still further, independent claims 13 and 15 recite either switching to a fallback control process based on the prediction quality falling below a specified minimum quality (claim 13), or transferring an existing control process to a self-optimizing control process (claim 15). The decision to switch control processes based on a prediction quality threshold, or to switch from an existing control process to a self-optimizing control process, is a determination/evaluation step, and is a mental process-type abstract idea. All independent claims therefore recite an abstract idea.
Subject Matter Eligibility Test, Step 2A Prong Two
In step 2A Prong Two, it is determined if the claims recite additional elements that integrate the judicial exception into a practical application. Independent claim 1 further recites i) one or more actuators including one or more mass flows and/or valves, ii) controlling one or more operating parameters of the process system using the one or more actuators, iii) a model-based deep reinforcement learning process, iv) a neural network, v) setting of one or more manipulated variable values in a second operating phase either a) by means of the self-optimizing process, where the system is operated in a first operating phase preceding the second operating phase, or b) by means of a further control process, and vi) training the neural network by means of training data obtained in the first operating phase. Independent claim 13 recites i) the one or more actuators including one or more mass flows and/or valves, ii) controlling one or more operating parameters of the process system via one or more manipulated variable values, iii) the model-based deep reinforcement learning process, and iv) the neural network, similarly to claim 1, and additionally recites v) a control device configured to carry out the setting of one or more manipulated variable values. Independent claim 15 recites i) the one or more actuators, ii) the model-based deep reinforcement learning, and iii) the neural network as in claim 1. The dependent claims also recite: further training the neural network in a second phase following the initial phase and/or assigning operating parameters to specific manipulated variable values (claim 4), taking consumption parameters into account with the cost function (claim 5), using past values to determine prediction values for specifying the one or more manipulated variable values (claim 6), exploring new control strategies in repeated loops by the neural network (claim 7), assessing manipulated variable values for suitability (claim 9), adapting the self-optimizing control process (claim 11), applying the method to a system where a cryogenic separation of component mixtures takes place (claim 12), a system where cryogenic separation of component mixtures takes place (claim 14), switching to a fallback control process based on a determined prediction quality falling below a minimum threshold (claim 16), replacing the self-optimizing control process by a different control process if the determined prediction quality falls below a specified minimum quality (claim 17), applying the method to a system where a cryogenic separation of component mixtures in an air fractionation plant takes place (claim 18), and a system where cryogenic separation of components in an air fractionation plant takes place (claim 19). Other than the limitations present in claims 12, 14, and 18-19, the limitations recited in the dependent claims merely further describe the limitations already present in the independent claims, and do not recite additional features. The recited limitations do not actually apply the abstract idea judicial exceptions into a practical application. Rather, the model-based deep reinforcement learning and neural network are stated at a high level of generality. There are no particular details about a particular model-based deep reinforcement learning or neural network, or how the model-based deep reinforcement learning or neural network work to provide a self-optimizing control process. These components are used to generally apply the abstract idea. The claims invoke a general model-based deep reinforcement learning/neural network as a tool for performing the self-optimizing control process in combination with the cost function, rather than purporting to improve the technology or a computer. See MPEP 2106.05(f). Therefore, the limitations of the model-based deep reinforcement learning/neural network are nothing more than an attempt to generally link the use of the judicial exception to the technological environment of computers. Further, the acts of controlling one or more operating parameters of the process system using the one or more actuators, either via the one or more manipulated variable values or otherwise, amounts to merely reciting the words “apply it” or an equivalent, and would naturally be expected to be performed in order to operate the process system. See MPEP 2106.04(d)(I). Additionally, the other limitations present in the claims, i.e. i) one or more actuators, iv) a control device configured to carry out the setting of one or more manipulated variable values, and the system where a cryogenic separation of component mixtures takes place either in an air fractionation plant or otherwise (present in claims 12, 14 and 18-19), generally link the use of the judicial exception to a laboratory environment. Further, regarding the switching to a fallback control process based on the determined prediction quality in claim 13, and the replacement of the existing control process with the self-optimizing control process in claim 15, these limitations amount to merely reciting the words “apply it” or an equivalent that would naturally result based on the decision to switch control processes based on a prediction quality, or to switch from an existing control process to a self-optimizing control process, which is a determination/evaluation step, and is a mental process-type abstract idea. See MPEP 2106.04(d)(I). Still further, regarding elements v) and vi) in independent claim 1, these limitations are gathering training data obtained in a first operating phase in order to train a neural network, which amounts to mere data gathering, which is insignificant extra-solution activity. See MPEP 2106.05(g).
Subject Matter Eligibility Test, Step 2B
In step 2B, it is determined if the claim recites additional elements that amount to significantly more than the judicial exception. In this case, the claims recite one or more actuators comprising one or more mass flows and/or valves, a self-optimizing control process comprising the use of model-based deep reinforcement learning, a neural network, a control device, a system that is operated in which a cryogenic separation of component mixtures takes place, and an air fractionation plant. These generically recited elements are nothing more than well-understood, routine, and conventional components that are well-known in the art, particularly as all claims have been rejected over the prior art. Further, the application of the abstract idea-type judicial exception into a laboratory environment is nothing more than generally linking the mental process judicial exception to a particular technological environment or field of use. See MPEP 2106.05(d) and 2106.05(e).
Further, with regards to the generically recited actuator(s), control process comprising model-based deep reinforcement learning, neural network, control device, cryogenic separation system, and air fractionation plant being nothing more than well-understood, routine, and conventional components that are well-known in the art, the following prior art is relied upon to show that the above elements are well-understood, routine, and conventional:
Badgwell et al. (US Pub. No. 2019/0187631; hereinafter Badgwell; already of record) teaches one or more actuators comprising one or more mass flows and/or valves ([0032]), a self-optimizing control process comprising the use of model-based deep reinforcement learning, a neural network, and a control device ([0003], [0028], the reinforcement learning agent acts as a model, [0060]).
Okada (US Pub. No. 2018/0218262; already of record) teaches a self-optimizing control process comprising the use of model-based deep reinforcement learning, a neural network, and a control device ([0007], [0028], [0084]).
Zapp et al. (Translation of WO Pub. No. 2015/158431; hereinafter Zapp; already of record) teaches a system that is operated in which a cryogenic separation of component mixtures takes place, and an air fractionation plant (Pg. 1 1st Para., Pg. 4 4th Para.).
Claims 4-7, 9, 11-12, 14, and 16-19 are rejected as depending on a claim rejected under 35 U.S.C. 101 without including additional elements sufficient to make the claims subject matter eligible.
Claim Rejections - 35 USC § 103
The text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
Claims 1, 4-7, 9, 11, and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Cohen et al. (US Pub. No. 2019/0236447; hereinafter Cohen) in view of Badgwell.
Regarding claim 1, Cohen discloses a method for operating a process system ([0017], [0002]). The method comprises:
setting one or more actuators in the process system by means of one or more manipulated variable values ([0017], [0020]-[0021], [0042]).
Controlling one or more operating parameters of the process system using the one or more actuators, wherein the one or more operating parameters include one or more mass flows and/or substance concentrations and/or one or more temperatures for the process system ([0017], [0020]-[0021], [0042]).
whereby the one or more operating parameters of the process system are influenced by the one or more manipulated variable values, and wherein the setting of the one or more manipulated variable values is carried out at least in a process phase by means of a self-optimizing control process ([0017], [0020]-[0021], [0042]).
The self-optimizing control process comprises using model-based deep reinforcement learning and consideration of a cost function, and wherein one or more components of the process system are represented in a model by means of a neural network, wherein the neural network represents a behavior of the process system and is used in the model-based deep reinforcement learning ([0017], [0020]-[0021], [0042], [0047], [0072]).
Predicting a future behavior of the process system over a specified time horizon for one or more future instants by means of the neural network to determine one or more prediction values to control the one or more operating parameters of the process system using the one or more actuators ([0028], [0058]-[0060], [0072], see also Claim 6).
Comparing the one or more prediction values for the one or more operating parameters predicted by the neural network for the one or more future instants to real values of the one or more operating parameters obtained at the one or more future instants, wherein a prediction quality representing an accuracy of the prediction by the neural network is determined on the basis of the comparison ([0028], [0058]-[0060], [0072], see also Claim 6).
The setting of the one or more manipulated variable values is carried out in a second operating phase (a) by means of the self-optimizing control process, wherein the system is operated in a first operating phase, which precedes the second operating phase, or (b) by means of a further control process ([0017]-[0021], [0042], [0047], [0072]).
training the neural network is by means of training data obtained in the first operating phase ([0017]-[0021], [0042], [0047], [0072]).
Cohen fails to explicitly disclose:
the one or more actuators include one or more mass flows and/or valves, and the one or more manipulated variable values of the one or more mass flows and/or valves.
Badgwell is in the analogous field of adaptive PID controller tuning via deep reinforcement learning (Badgwell [0026]). Badgwell teaches one or more actuators including one or more mass flows and/or valves (Badgwell; [0032], [0049], [0063]). It would have been obvious to one having ordinary skill in the art before the effective filing date of the invention to modify the method of Cohen with the teachings of Badgwell so that the method comprises that the one or more actuators include one or more mass flows and/or valves, so that the manipulated variable values are of the one or more mass flows and/or valves, to be able to use the model in order to optimize a chemical process and ensure that controlled variables are maintained at target values for optimal production (Badgwell; [0003], [0032], [0049], [0063]).
Regarding claim 4, modified Cohen discloses the method according to Claim 1. Cohen further discloses that the neural network is subsequently trained by means of training data obtained in the second operating phase, and/or wherein the training data in each case comprise operating parameters assigned to specific manipulated variable values (Cohen; [0017]-[0021], see also Claim 7).
Regarding claim 5, modified Cohen discloses the method according to Claim 1. Cohen further discloses that consumption parameters are taken into account by means of the cost function and are assessed with respect to respective target parameters (Cohen; [0072], the reward function depends on CVs, MVs, and/or DVs, [0042], the disturbance variables include a flow rate of feed stock entering the plant).
Regarding claim 6, modified Cohen discloses the method according to Claim 1. Modified Cohen further discloses that one or more actual values of the one or more operating parameters are acquired for one or more past instants at which one or more prediction values for the one or more operating parameters are determined for one or more future instants using the one or more actual values by means of the self-optimizing control process, and wherein the one or more manipulated variable values are specified by means of one or more setpoint values for the one or more operating parameters and by means of the one or more prediction values by means of the self- optimizing control process (Cohen; [0028], [0058]-[0060], the neural network is trained to provide for time t, t+1, and t-T for a period of T time points before the calculated time point, [0072], see also Claim 6).
Regarding claim 7, modified Cohen discloses the method according to Claim 1. Cohen further discloses exploring new control strategies by means of the neural network in repeated exploration loops (Cohen; [0058]-[0060], [0079]-[0080]).
Regarding claim 9, modified Cohen discloses the method according to Claim 1, and all limitations recited therein.
Modified Cohen fails to explicitly disclose that the one or more manipulated variable values are assessed for their suitability prior to their use to set the one or more actuators.
Badgwell further teaches that one or more manipulated variable values are assessed for their suitability prior to their use to set the one or more actuators (Badgwell; [0032], [0049], [0063]). It would have been obvious to one having ordinary skill in the art before the effective filing date of the invention to modify the method of modified Cohen with the further teachings of Badgwell so that the one or more manipulated variable values are assessed for their suitability prior to their use to set the one or more actuators. The motivation would have been to ensure that the manipulated variables are near a reference value or reference trajectory (Badgwell; [0032], [0049], [0063]), in order to ensure that the process system is operating as desired.
Regarding claim 11, modified Cohen discloses the method according to Claim 1. Cohen further discloses that an adaptation of the self-optimizing control process is performed (Cohen; [0058]-[0060], [0079]-[0080]).
Regarding claim 15, modified Cohen discloses the self-optimizing control process as recited in claim 1. Modified Cohen further discloses a method for converting a process system, which system is configured to set one or more actuators in the process system by means of one or more manipulated variable values and thereby influence one or more operating parameters of the system, wherein the one or more actuators include one or more mass flows and/or control valves wherein the method comprises:
replacing an existing control process, by means of which the one or more control values of the process system are set, by a self-optimizing control process as recited in claim 1,
the self-optimizing control process comprising using model-based deep reinforcement learning and consideration of a cost function, and the one or more components of the process system is represented in a model by means of a neural network, the neural network representing a behavior of the process system and being used in the model-based deep reinforcement learning, and
wherein replacement of the existing control process with the self-optimizing control process comprises subsequently transferring control functions of the existing control process to the self-optimizing control process (the details of the self-optimizing control process as claimed have been previously recited in Claim 1. Further, Cohen teaches replacing an old process with a self-optimizing control process in [0038]).
Claims 12 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Cohen in view of Badgwell, as applied to claims 1, 4-7, 9, 11, and 15 above, in view of Zapp.
Regarding claim 12, modified Cohen discloses the method according to Claim 1.
Modified Cohen fails to explicitly disclose that a cryogenic separation of component mixtures takes place in the process system.
Zapp is in the analogous field of controlling process systems (Zapp Pg. 4 4th Para.), and teaches a method of controlling a process system that is operated in which a cryogenic separation of component mixtures takes place, particularly an air fractionation plant (Zapp Pg. 4 4th Para.). It would have been obvious to one having ordinary skill in the art before the effective filing date of the invention to modify the method of modified Cohen with the teachings of Zapp so that a cryogenic separation of component mixtures takes place in the process system. The motivation would have been to apply the teachings of Badgwell, which provide for improved automatic tuning of controllers to reduce or minimize the amount of required manual intervention (Badgwell [0006]), into a cryogenic separation environment as in Zapp (Zapp Pg. 4 4th Para.), thereby optimizing the process of cryogenic separation and minimizing the need for oversight.
Regarding claim 18, modified Cohen discloses the method according to Claim 1, and all limitations recited therein.
Modified Cohen fails to explicitly disclose that a cryogenic separation of component mixtures in an air fractionation plant takes place in the process system.
Zapp teaches a cryogenic separation of component mixtures in an air fractionation plant that takes place in the process system (Zapp Pg. 4 4th Para.). It would have been obvious to one having ordinary skill in the art before the effective filing date of the invention to modify the method of modified Cohen with the teachings of Zapp so that a cryogenic separation of component mixtures in an air fractionation plant that takes place in the process system. The motivation would have been to apply the teachings of Badgwell, which provide for improved automatic tuning of controllers to reduce or minimize the amount of required manual intervention (Badgwell [0006]), into a cryogenic separation environment as in Zapp (Zapp Pg. 4 4th Para.), thereby optimizing the process of cryogenic separation and minimizing the need for oversight.
Claims 13 and 16-17 are rejected under 35 U.S.C. 103 as being unpatentable over Cohen in view of Badgwell and Ryu et al. (US Pub. No. 2021/0328630; hereinafter Ryu; already of record).
Regarding claim 13, Cohen discloses a process system ([0017], [0002]). The process system is configured to
set, by means of one or more manipulated variable values, one or more actuators of the process system ([0017], [0020]-[0021], [0042]).
the process system comprising a control device which is configured to
control one or more operating parameters of the process system, wherein the one or more operating parameters include one or more mass flows and/or substance concentrations or one or more temperatures for the process system, wherein the one or more operating parameters of the process system are influenced by the one or more manipulated variable values ([0017], [0020]-[0021], [0042]).
Carry out the setting of the one or more manipulated variable values, at least in a process phase, by means of a self-optimizing control process and to carry out the self- optimizing control process by means of model-based deep reinforcement learning and consideration of a cost function, one or more components of the process system being represented in a model by means of a neural network, the neural network representing a behavior of the process system and being used in the model-based deep reinforcement learning ([0017], [0020]-[0021], [0042], [0047], [0072]).
Predict a future behavior of the process system over a specified time horizon for one or more future instants by means of the neural network to determine one or more prediction values to control the one or more operating parameters of the process system using the one or more actuators ([0028], [0058]-[0060], [0072], see also Claim 6).
Compare the one or more prediction values for the one or more operating parameters predicted by the neural network for the one or more future instants to real values of the one or more operating parameters obtained at the one or more future instants, wherein a prediction quality representing an accuracy of the prediction by the neural network is determined on the basis of the comparison ([0028], [0058]-[0060], [0072], see also Claim 6).
Cohen fails to explicitly disclose:
the one or more actuators include one or more mass flows and/or valves, and the one or more manipulated variable values of the one or more mass flows and/or valves; and
that the control device is configured to switch to a fallback control process from the self-optimizing control process if the prediction quality determined based on the comparison falls below a specified minimum quality.
Badgwell is in the analogous field of adaptive PID controller tuning via deep reinforcement learning (Badgwell [0026]). Badgwell teaches one or more actuators including one or more mass flows and/or valves (Badgwell; [0032], [0049], [0063]). It would have been obvious to one having ordinary skill in the art before the effective filing date of the invention to modify the process system of Cohen with the teachings of Badgwell so that the one or more actuators include one or more mass flows and/or valves, so that the manipulated variable values are of the one or more mass flows and/or valves, to be able to use the model in order to optimize a chemical process and ensure that controlled variables are maintained at target values for optimal production (Badgwell; [0003], [0032], [0049], [0063]).
Modified Cohen fails to explicitly disclose that the control device is configured to switch to a fallback control process from the self-optimizing control process if the prediction quality determined based on the comparison falls below a specified minimum quality.
Ryu is in the analogous field of machine learning models (Ryu [0011]). Ryu teaches switching to a fallback control process from a first control process based on a determined prediction quality falling below a specified minimum quality (Ryu; [0012], [0028]). It would have been obvious to one having ordinary skill in the art before the effective filing date of the invention to modify the process system of modified Cohen with the teachings of Ryu so that the control device is configured to switch to a fallback control process from the self-optimizing control process if the prediction quality determined based on the comparison falls below a specified minimum quality, in order to discard models that are ineffective at predicting real events, thereby ensuring that the model that is used in the control process is effective at predicting real events.
Regarding claim 16, modified Cohen discloses the method according to Claim 1, and all limitations recited therein.
Modified Cohen fails to explicitly disclose switching to a fallback control process from the self-optimizing control process if the prediction quality determined based on the comparison falls below a specified minimum quality.
Ryu is in the analogous field of machine learning models (Ryu [0011]). Ryu teaches switching to a fallback control process from a first control process based on a determined prediction quality falling below a specified minimum quality (Ryu; [0012], [0028]). It would have been obvious to one having ordinary skill in the art before the effective filing date of the invention to modify the method of modified Cohen with the teachings of Ryu to comprise switching to a fallback control process from the self-optimizing control process if the prediction quality determined based on the comparison falls below a specified minimum quality, in order to discard models that are ineffective at predicting real events, thereby ensuring that the model that is used in the control process is effective at predicting real events.
Regarding claim 17, modified Cohen discloses the method according to Claim 1, and all limitations recited therein.
Modified Cohen fails to explicitly disclose that the self-optimizing control process is replaced by a different control process if the determined prediction quality falls below a specified minimum quality.
Ryu teaches a process that is replaced by a different control process if the determined prediction quality falls below a specified minimum quality (Ryu; [0012], [0028]). It would have been obvious to one having ordinary skill in the art before the effective filing date of the invention to modify the method of modified Cohen with the teachings of Ryu so that the self-optimizing control process is replaced by a different control process if the determined prediction quality falls below a specified minimum quality, in order to discard models that are ineffective at predicting real events, thereby ensuring that the model that is used in the control process is effective at predicting real events.
Claims 14 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Cohen in view of Badgwell and Ryu, as applied to claim 13 above, in view of Zapp.
Regarding claim 14, modified Cohen discloses the process system according to Claim 13.
Modified Cohen fails to explicitly disclose that the process system is adapted to perform a cryogenic separation of component mixtures.
Zapp is in the analogous field of controlling process systems (Zapp Pg. 4 4th Para.), and teaches a method of controlling a process system that is adapted to perform a cryogenic separation of component mixtures (Zapp Pg. 4 4th Para.). It would have been obvious to one having ordinary skill in the art before the effective filing date of the invention to modify the process system of modified Cohen with the teachings of Zapp so that the process system is adapted to perform a cryogenic separation of component mixtures. The motivation would have been to apply the teachings of Badgwell, which provide for improved automatic tuning of controllers to reduce or minimize the amount of required manual intervention (Badgwell [0006]), into a cryogenic separation environment as in Zapp (Zapp Pg. 4 4th Para.), thereby optimizing the process of cryogenic separation and minimizing the need for oversight.
Regarding claim 19, modified Cohen discloses the process system according to Claim 13, and all limitations recited therein.
Modified Cohen fails to explicitly disclose that the process system is an air fractionation plant adapted to perform a cryogenic separation of components of an air mixture.
Zapp teaches an air fractionation plant adapted to perform a cryogenic separation of components of an air mixture (Zapp Pg. 4 4th Para.). It would have been obvious to one having ordinary skill in the art before the effective filing date of the invention to modify the process system of modified Cohen with the teachings of Zapp so that the process system is an air fractionation plant adapted to perform a cryogenic separation of components of an air mixture. The motivation would have been to apply the teachings of Badgwell, which provide for improved automatic tuning of controllers to reduce or minimize the amount of required manual intervention (Badgwell [0006]), into a cryogenic separation environment as in Zapp (Zapp Pg. 4 4th Para.), thereby optimizing the process of cryogenic separation and minimizing the need for oversight.
Response to Arguments
Applicant's arguments filed July 20, 2026 have been fully considered but they are not persuasive.
Applicant argues on Pg. 8 of their Remarks that the claims are not directed to an abstract idea under Step 2B, Prong 1, as the claims recite training a neural network, and training a neural network is not a judicial exception, pointing to Example 39 of the Subject Matter Eligibility Examples. The Examiner respectfully disagrees. In Example 39, it is not the training of the neural network that is not a judicial exception, it is applying transformations to digital facial images that cannot be practicably performed in the human mind. In the case of the instant application, the claims do recite a judicial exception, i.e. a cost function, which is a mathematical equation-type judicial exception, and the neural network is recited at a high level of generality to perform the judicial exception. A claim that requires a computer may still recite a mental process- See MPEP 2106.04(a)(2)(III)(C). See Example 48, Claim 1 for a counter example demonstrating that reciting training a neural network does not automatically render a claim patent-eligible.
Applicant further argues on Pg. 8 of their Remarks that independent claims 1, 13, and 15 integrate any alleged abstract idea into a practical application under Step 2A, Prong Two, as the claims require setting one or more actuators in the process system, which is a concrete control of a physical process system, not a generic invocation of a tool. The Examiner respectfully disagrees, as the limitations in question amount to applying the judicial exception and are insignificant extra-solution activity, particularly at the high level of generality recited.
Applicant further argues on Pgs. 8-9 of their Remarks that the comparing a neural network’s own predictions of physical operating parameters against later-measured real plant values to gauge a neural networks’ prediction accuracy is not something that can be practically performed in the human mind. The Examiner respectfully disagrees, as such a comparison can be readily performed in a human mind, and is a determination/evaluation-type mental process abstract idea. The neural network is again applied at a high level of generality, and does not amount to integration into a practical application.
Applicant further argues on Pg. 9 of their Remarks that claim 13, by reciting that the control device is configured to switch to a fallback control process if the prediction quality falls below a specified minimum quality, does not amount to merely apply it, and is a practical application. The Examiner respectfully disagrees, as determining that an operation should be switched to a different control process if the existing control process is inaccurate or faulty is a determination/evaluation that can be practically performed in the human mind, and the act of actually switching to a fallback control process amounts to merely “apply it”.
Applicant further argues on Pg. 9 of their Remarks that the claimed control approach provides a concrete technological improvement, and points to parts of the Specification to support this argument. The Examiner does not find this argument persuasive, as the excerpts form the Specification cited are vague and conclusory, and it is unclear specifically which part(s) of the claimed subject matter, if any, actually results in the improvements discussed in the Specification. In order for such an argument to be persuasive, the improvement as discussed in the Specification must be clearly reflected in the claims.
Applicant further argues on Pgs. 10-11 of their Remarks that Wen does not teach the neural network recited in claims 1, 13, and 15. While the Examiner agrees with this argument, the claimed neural network has been rejected using the reference Cohen.
Applicant further argues on Pgs. 11-12 of their Remarks that Wen does not teach the comparing limitations of claims 1 and 13, as amended. While the Examiner agrees with this argument, the claimed limitations have been rejected using the reference Cohen.
Applicant further argues on Pg. 12 of their Remarks that Ryu does not teach switching to a fallback control process from a self-optimizing control process. The Examiner respectfully disagrees. Ryu does teach switching to a fallback control process, and the other prior art of record already teaches the claimed self-optimizing control process. Therefore, Ryu, in combination with the other prior art of record, does teach the claimed limitation.
Applicant further argues on Pg. 12 of their Remarks that Ryu is non-analogous art. The Examiner respectfully disagrees. Ryu is concerned with predictive modeling, which is reasonably pertinent to the problem faced by the inventor. Ryu is therefore analogous art.
Applicant further argues on Pgs. 12-13 of their Remarks that Wen does not teach replacing an existing control process by a self-optimizing control process, and subsequently transferring control functions of the existing control process to the self-optimizing control process. Without acceding to Applicant’s argument, the Examiner notes that the instant claims have been rejected using Cohen as a primary reference, which does teach the claimed limitation.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to John McGuirk whose telephone number is (571)272-1949. The examiner can normally be reached M-F 8am-530pm.
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, Charles Capozzi can be reached at (571) 270-3638. 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.
/JOHN MCGUIRK/Primary Examiner, Art Unit 1798
Prosecution Insights