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 .
Claim Amendments
This Non-Final Rejection is filed in response to Applicant Arguments/Remarks Made in an amendment filed 09/08/2026.
Claims 1 and 25 are amended.
Claims 1-27 remain pending.
Repones to Arguments
Argument 1, Applicant argues in Applicant Arguments/Remarks Made in an amendment filed 09/08/2026 pg. 8-11, that prior art fails to teach the limitations,
“assisting a pilot in control of an aircraft”, “instructions expressed by the pilot via a human-machine interface in a predefined semantic framework comprising expressions allowing to express functional objectives and intentions regarding the conducting of the mission", "said technical parameters being adapted to the mission context, said context
comprising current data obtained from measurements by sensors on the aircraft and in
its environment", or "a step of developing control orders for an aircraft flight management system".
Response to Argument 1, applicant’s arguments have been considered, However in light of the amendments, a newly found combination of prior art (U.S. Patent Application Publication NO. 20210092078 “Lee”, U.S. Patent Application Publication NO. 20190302276 “Sandford”, and further in light of U.S. Patent Application Publication NO. 20070204323 “Wilkinson”) is applied to updated rejections.
However, the examiner notes While the Lee may not explicitly teach that a user is a pilot and that the information being output information is aircraft specific control information for a pilot, it should be noted that that Lee sets the stage for an NLP based artificial intelligence based decision making system capable of having a predefined semantic framework for processing a user’s requests related to flight adjacent information, wherein the BRI for a mission context encompasses how the AI assistance displays data related to the user’s request. In response to applicant’s argument that there is no teaching, suggestion, or motivation to combine the references, the examiner recognizes that obviousness may be established by combining or modifying the teachings of the prior art to produce the claimed invention where there is some teaching, suggestion, or motivation to do so found either in the references themselves or in the knowledge generally available to one of ordinary skill in the art. See In re Fine, 837 F.2d 1071, 5 USPQ2d 1596 (Fed. Cir. 1988), In re Jones, 958 F.2d 347, 21 USPQ2d 1941 (Fed. Cir. 1992), and KSR International Co. v. Teleflex, Inc., 550 U.S. 398, 82 USPQ2d 1385 (2007). In this case, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to add a pilot in control of an aircraft and the sensor data associated with the aircraft, to the HMI with semantic processing of Lee, with a pilot in the control of an aircraft that may have an artificial intelligence system processing aircraft sensor data in order to output an alert for the control path of said aircraft to the pilot, as taught by Sandford. One would have been motivated to the combine the aircraft specific sensor processing and display of aerial navigation orders of Sandford with HMI of Lee as the combination provides users with key information to vehicle guidance, navigation and control systems, specifically, velocity vectors and range, with derivable information about surface relative attitude as such parameters measured with high accuracy enable safe and reliable human driven and autonomous cars and trucks and enable aerial vehicles (with and without pilots) to navigate without GPS or other external signal (Sandford, para. 0052]).
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.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
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) 1-3, 20, 22, 24, 25, & 27 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication NO. 20210092078 “Lee”, U.S. Patent Application Publication NO. 20190302276 “Sandford”, and further in light of U.S. Patent Application Publication NO. 20070204323 “Wilkinson”.
Claim 1:
Lee teaches a computer-implemented method for assisting a (i.e. para. [0037], a dialogue system (e.g., a conversational computing system, a chatbox system, a digital assistant, etc.) is a computer system that can employ artificial intelligence (AI) and/or natural language processing (NLP) to facilitate human-computer interaction), the method being implemented between a (i.e. para. [0067], “the user input at the step 604 can correspond to the statement data 112. Furthermore, the user input at the step 604 can include a phrase such as, for example, “I want to fly to someplace for my family vacation”, wherein the BRI for a mission encompasses a task for a user), and comprising: the following steps performed by a translation system called a top-down translator: receiving, at input of said top-down translator, data with a high level of abstraction, the data with a high level of abstraction being instructions expressed by the (i.e. para. [0067], the user input at the step 604 can correspond to the statement data 112. Furthermore, the user input at the step 604 can include a phrase such as, for example, “I want to fly to someplace for my family vacation”, wherein the BRI for a top-down translator encompasses how a dialogue system receives input of a user’s request, wherein the BRI for a high level of abstraction encompasses how the user input phase is a general command in pre-defined semantics that expresses an intention to execute an understood task. Wherein the BRI for a predefined semantic framework comprising natural language expressions encompasses how analyzing the question results in the derivation of the semantic meaning that is embedded in the task that include the requirements on how to complete the task. Wherein the BRI for expressions encompasses dialogue keywords such as “fly” and “family vacation” that express an intention corresponding to “FLIGHT_REQUEST” and “DESTINATION_REQUEST”); and
translating the said instructions into technical parameters (i.e. para. [0064], the dialogue component 102 can additionally or alternatively generate the machine learning model 508 based on intent information associated with the statement data 112), by one or more universal approximators (i.e. para. [0064], “The contextual information 510 can be data that provides context to the statement data 112. In an aspect, the contextual information 510 can provide knowledge regarding the statement data 112”, wherein it is noted that applicant’s specification cites that a universal approximator generally denotes a neural network. Wherein it is noted that the statement data is translated into a dialogue component by a machine learning component that may be a recurrent neural network model and word embedding model which may be used by the dialogue system to output a recommendation), said universal approximators being one or more neural networks and/or fuzzy logic decision trees that have been trained via machine learning (i.e. para. [0057], the machine learning component 302 can employ deep learning, expert systems, fuzzy logic… other non-linear training techniques, data fusion, utility-based analytical systems) to map high-level abstract data expressed in the predefined semantic framework into technical parameters that can be manipulated by a decision assistance system, said technical parameters being adapted to the mission context (i.e. para. [0060], “The dialogue system 404 can include the dialogue component 102. In an embodiment, the dialogue component 102 can include the contextual information component 104, the dialogue routing component 106, the communication component 202, the machine learning component 302”, wherein it is noted that the dialogue component maps the user input to semantic meaning as part of the machine learning system that may use fuzzy logic and be trained to learn one or more patterns, one or more inferences, one or more correlations, one or more features and/or information related to the statement data and/or the contextual data as part of outputting an appropriate response back to a user),
a step performed by a decision assistance system of the aircraft platform, said decision assistance system being able to manipulate said technical parameters to produce raw technical parameters (i.e. para. [0071], “the contextual information can be determined via one or more NLP techniques and/or one or more NLU techniques. For example, a language model associated with NLP and/or a language model associated with NLP can be employed to extract the contextual information from the statement.”, wherein the BRI for raw technical parameters encompasses the extracted contextual data, such as date or a time or a location) as low-level abstraction data (i.e. para. [0064], “the contextual information 510 can provide knowledge regarding the statement data 112. In an embodiment, the contextual information component 104 can determine contextual information 510 based on the statement data 112. For instance, the contextual information 510 can include … a feature… a date, a time) representing technical, quantifiable and measurable task recommendations for achieving the functional objectives and intentions expressed by the pilot (i.e. para. [0056, 0063, 0069], “The machine learning component 302 can perform learning explicitly or implicitly with respect to learning one or more patterns, one or more inferences, one or more correlations, one or more features and/or information related to the statement data 112 and/or the contextual data… a skill from the set of skills 506.sub.1-N can be a software program that represents a particular capability of a task and/or an answer in a specific domain”, wherein the decision making is affected by the contextual data which directs the system to infer and to share a skill, such as a first skill “FLIGHT_REQUEST” and a second skill “ATTRACTION_REQUEST, to help a user accomplish a required task within a specific domain. Wherein it is noted that the recommendation output by the dialogue system is based on the path traversal restricted by technical, quantifiable and measurable task recommendations for a user’s flight request for destination for a family vacation or a flight request to specific location of Chicago as seen in Fig. 7),
the following steps performed by a translation system called a bottom-up translator: receiving, at input of said bottom-up translator, said raw technical parameters from said decision assistance system of the aircraft platform, (i.e. para. [0072], “a path traversal for a dialogue system is generated, by the system (e.g., by the dialogue routing component 106), based on the contextual information.”, wherein the model receive the context data. Wherein the BRI for raw technical parameters encompasses contextual data that is used to infer a corresponding domain specific skill for an action); and
translating said received raw technical parameter, into expressions comprised in said predefined semantic framework, (i.e. para. [0072-0073], “A skill associated with the dialogue system can be, for example, a model (e.g., a machine learning model, a generative model, etc.) that generates a response (e.g., an answer) in response to the statement… In an embodiment, the skill (e.g., the skill associated with the path traversal for the dialogue system) that receives the statement can generate the response. For example, the response can be output generated by the model associated with the skill”, wherein the BRI for low level of abstraction encompasses how context data is translated into a high level response in the form of a conversational answer from the model in response to a user’s statement) by (i.e. para. [0057], the machine learning component 302 can employ deep learning, expert systems, fuzzy logic), said expressions being in natural language (i.e. para, [0067], “the candidate skills at the step 606 can include two or more skills from the set of skills 506.sub.1-N to potentially route the user input (e.g., the statement data 112). In an example, the candidate skills at the step 606 can include a first skill “FLIGHT_REQUEST” and a second skill “DESTINATION_REQUEST.” “, wherein it is noted that the user input is to a computer system that can employ artificial intelligence (AI) and/or natural language processing (NLP) to facilitate human-computer interaction) and representing proposal of tasks to be realized to conduct the mission according to the functional objection and intentions expressed by the (i.e. para. [0037], “a dialogue system (e.g., a conversational computing system, a chatbox system, a digital assistant, etc.) is a computer system that can employ artificial intelligence (AI) and/or natural language processing (NLP) to facilitate human-computer interaction”, wherein the model produces a recommended answer expression through NLP which characterizes the answer as dialogue to the user’s mission of, “I want to fly to someplace for my family vacation.”); and
While Lee teaches an HMI that translates high level user input dialogue into low level context and skills that are used by the machine learning to generate a recommended answer within a technical domain, which is translated back into high level dialogue using NLP, Lee may not explicitly teach
a pilot in the control of an aircraft;
said context comprising current data obtained from measurements by sensors on the aircraft and in its environment; and
a step of developing control orders for an aircraft flight management system.
the bottom-up translation being executed by one or more white boxes comprising one or more fuzzy logic decision trees.
However, Sandford teaches
a pilot in the control of an aircraft (i.e. para. [0054], The present invention may be implemented in ADAS 3-5 (Advanced Driver Assistance) vehicles, both civilian and military as well as piloted and unpiloted aircraft, especially those requiring VTOL (Vertical Take Off and Landing));
said context comprising current data obtained from measurements by sensors on the aircraft and in its environment (i.e. para. [0087], FIG. 9 is an illustration 114 of the displays that convey navigation information to the pilot of the vehicle 12. Surface relative velocity is presented on instruments that show Vx 116, Vy 118 and Vz 119. FIG. 9 also depicts other navigation information for the vehicle 12, including surface relative altitude 120, flight path angle 122, the velocity vector 124, the angle of attack 126 and the surface relative pitch angle 128); and
a step of developing control orders for an aircraft flight management system (i.e. para. [0178], “A traffic collision avoidance system or traffic alert and collision avoidance system . . . is an aircraft collision avoidance system designed to reduce the incidence of mid-air collisions between aircraft. It monitors the airspace around an aircraft for other aircraft equipped with a corresponding active transponder, independent of air traffic control, and warns pilots of the presence of other transponder-equipped aircraft which may present a threat of mid-air collision (MAC)”, wherein it is noted Sandford describes a system, such as in Fig. 9-10, that processes aircraft specific sensor data and outputs displayed navigational alerts in order to provide situational awareness information suitable for artificial intelligence decision making, and avoidance of stationary or mobile hazards and hazard-relative navigation which includes giving warning to aircraft pilots in the control of an aircraft via a GUI)
It would have been obvious to one of ordinary skill in the art before the effective filing date to add a pilot and an aircraft platform, and a step of developing control orders for an aircraft flight management system, with the HMI of Lee, with a pilot in the control of an aircraft may have an artificial intelligence system processing aircraft sensor data in order to output an alert for the control path of said aircraft to the pilot, as taught by Sandford. One would have been motivated to the combine the aircraft specific sensor processing and display of aerial navigation orders of Sandford with HMI of Lee as the combination provides users with key information to vehicle guidance, navigation and control systems, specifically, velocity vectors and range, with derivable information about surface relative attitude as such parameters measured with high accuracy enable safe and reliable human driven and autonomous cars and trucks and enable aerial vehicles (with and without pilots) to navigate without GPS or other external signal (Sandford, para. 0052]).
While Lee-Sandford teach a specialized HMI, Lee-Sandford may not explicitly teach
the bottom-up translation being executed by one or more white boxes comprising one or more fuzzy logic decision trees
However, Wilkinson teaches
the bottom-up translation being executed by (i.e. para. [0066], “A classifier is a function that maps an input attribute vector, x=(x1, x2, x3, x4, xn), to a confidence that the input belongs to a class, that is, f(x)=confidence (class). Such classification can employ a probabilistic and/or statistical-based analysis (e.g., factoring into the analysis utilities and costs) to prognose or infer an action that a user desires to be automatically performed”, wherein the BRI for the bottom-up translation encompasses the attribute vectors that are fed into a classifier model in order to infer a recommended user action. The examiner notes that Wilkinson teaches a human machine interface where raw data that consists of input attribute vectors may be used to display and inform a potential action according to an identified constraint) one or more white boxes (i.e. para. [0054], the disclosed techniques can be extensible to software on a white box in accordance with some embodiments) comprising one or more fuzzy logic decision trees (i.e. para. [0067], directed and undirected model classification approaches include, e.g., naive Bayes, Bayesian networks, decision trees, neural networks, fuzzy logic models)
It would have been obvious to one of ordinary skill in the art before the effective filing date to add executed by one or more white boxes comprising one or more fuzzy logic decision trees, with the HMI of Lee-Sandford, with the recommendations based on technical policy data that uses one or more white-boxes, as taught by Wilkinson. One would have been motivated combine white boxes of Wilkinson with HMI of Lee-Sandford as the combination provides a device that behaves in a certain manner when introduced into a certain environment, which takes advantage of all the environment features available.
Claim 2:
Lee, Sandford, and Wilkinson teach the method as claimed in claim 1.
Lee teaches further comprising the steps of: receiving the received data with a low level of abstraction (12) from the machine in response to the input data (11) (i.e. para. [0072], “A skill associated with the dialogue system can be, for example, a model (e.g., a machine learning model, a generative model, etc.) that generates a response (e.g., an answer) in response to the statement”, wherein the BRI for low level of abstraction encompasses context and skill data is input into a model and in response an answer is generated); comparing the input data with a high level of abstraction (16) captured by an HMI of a pilot and the translated data with a high level of abstraction (17) (i.e. para. [0073], “The response can be an answer related to the statement. Furthermore, the response can include audio data and/or textual data. In an embodiment, the skill (e.g., the skill associated with the path traversal for the dialogue system) that receives the statement can generate the response. For example, the response can be output generated by the model associated with the skill”, wherein the high level user input captured by the HMI may be compared as relevant to translated model answer most associated with the determined skill).
Claim 3:
Lee, Sandford, and Wilkinson teach the method as claimed in claim 2.
Lee teaches comprising
the step of selecting output data from among multiple output data, through filtering and/or thresholding (i.e. para. [0057], “The machine learning component 302 can employ any suitable machine-learning based techniques, statistical-based techniques and/or probabilistic-based techniques. For example, the machine learning component 302 can employ deep learning, expert systems, fuzzy logic”. Wherein the BRI for filtering or thresholding encompasses the probabilistic fuzzy logic of the machine learning system for selecting an appropriate answer routing), or notably by traversing the bottom-up translation consisting of GFT fuzzy logic decision trees.
Claim 20:
Lee, Sandford, and Wilkinson teach the method as claimed in any one of the preceding claims.
Lee further teaches wherein machine learning is performed online (i.e. para. [0026], Measured service: cloud systems automatically control and optimize resource use by leveraging a metering capability at some level of abstraction appropriate to the type of service (e.g., storage, processing, bandwidth, and active user accounts)).
Claim 22:
Lee, Sandford, and Wilkinson teaches the method as claimed in any one of the preceding claims.
Lee further teaches wherein the machine learning comprises one or more algorithms selected from among the algorithms comprising: support vector machines (i.e. para. [0057], the machine learning component 302 can perform… a set of support vector regression machine learning computations); classifiers (i.e. para. [0064], he machine learning model 508 can be, for example, a classification model); neural networks (i.e. para. [0055], the machine learning component 302 can combine a recurrent neural network model and a word embedding mode; decision trees and/or steps in statistical methods such as the Gaussian mixture model, logistic regression, linear discriminant analysis and/or genetic algorithms (i.e. para. [0057], he machine learning component 302 can … a set of logistic regression machine learning computations).
Claim 24:
Lee, Sandford, and Wilkinson teach a computer program product.
Lee further teaches said computer program comprising code instructions for performing the steps of the method as claimed in any one of claims 1 to 23 when said program is executed on a computer (i.e. para. [0104], The computer program product can include a computer readable storage medium (or media) having computer readable program instructions thereon for causing a processor to carry out aspects of the present invention).
Claim 25:
Claim 25 is the system claim reciting similar limitations to Claim 1 and is rejected for similar reasons.
Claim 27:
Lee, Sandford, and Wilkinson teach the method as claimed in claim 1.
Lee further teaches
Wherein the top-down translator machine learning is based on the bottom-up translator machine learning (i.e. para. [0038], “a classification model can be generated to infer a correct skill for the user input statement to be routed via the dialogue system. As such, an improved dialogue system can be provided. For example, performance for routing input of a dialogue system to a corresponding skill of the dialogue system can be improved. Accuracy and/or quality of a response provided by a dialogue system can also be improved. Moreover, accuracy of data generated by a machine learning model for dialogue routing can be improved and/or quality of data generated by a machine learning model for dialogue routing can be improved”, wherein the top-down NLP or NLU model that extracts context information from a user’s statement can be based on and improved by the quality of the bottom-up classification machine learning model responses to a user).
Claim(s) 4 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication NO. 20210092078 “Lee”, U.S. Patent Application Publication NO. 20190302276 “Sandford”, and further in light of U.S. Patent Application Publication NO. 20070204323 “Wilkinson”, as applied to claim 1 above, and further in light of U.S. Patent Application Publication NO. 20200341438 “Alsadah”.
Claim 4:
Lee, Sandford, and Wilkinson teach the method as claimed in claim 1.
While Wilkinson teaches a white-box comprising one or more GFT fuzzy logic decision trees, Lee, Sandford, and Wilkinson may not explicit teach further comprising
the step of controlling (125) at least one black box (1101) using at least one white box (1301), the top-down translator comprising one or more black boxes, a white box comprising one or more GFT fuzzy logic decision trees.
However, Alsadah teaches
the step of controlling (125) at least one black box (1101) using at least one white box (1301), the top-down translator comprising one or more black boxes, a white box comprising one or more GFT fuzzy logic decision trees (i.e. para. [0079], “the system modeler 146 (shown in FIG. 4), which can include a black-box, grey-box, white-box identification or another systems identification modeling solution suitable for the control system”, wherein it is noted in Fig. 7 that the HMI functions as a top-down translator for user operator input being input into a system modeler that may comprise a black box and white box for correlating an identifying corrective action from user input data)
It would have been obvious to one of ordinary skill in the art before the effective filing date to add the step of controlling (125) at least one black box (1101) using at least one white box (1301), the top-down translator comprising one or more black boxes, a white box, with the HMI using white box fuzzy logic of Lee-Sandford-Wilkinson, with the use of a black box and white box combination, as taught by Alsadah. One would have been motivated to the system models of Alsadah with HMI of Lee-Sandford-Wilkinson as the combination attributes in the structure of interest might not be known, thus a system model can be constructed based on insight into or experimental data for the structure of interest, which can be utilized in determining relationships between data sources.
Claim(s) 5 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication NO. 20210092078 “Lee”, U.S. Patent Application Publication NO. 20190302276 “Sandford”, and in light of U.S. Patent Application Publication NO. 20070204323 “Wilkinson”, and further in light of U.S. Patent Application Publication NO. 20200341438 “Alsadah”, as applied to claim 4 above, and further in light of U.S. Patent NO. 10629191 “Cheng”.
Claim 5:
Lee, Sandford, Wilkinson, and Alsadah teach the method as claimed in claim 4.
Lee, Sandford, Wilkinson, and Alsadah may not explicitly teach
further comprising a step of controlling a network (110) of top-down black boxes using a bottom-up white box, the control step consisting in optimizing (125) said black boxes through machine learning.
However, Cheng teaches
further comprising a step of controlling a network (110) of top-down black boxes using a bottom-up white box, the control step consisting in optimizing (125) said black boxes through machine learning (i.e. Col. 46, lines 35-45, “This data set can be used, using methods known in the art, including but not limited to standardized machine learning methods such as parametric classification methods, non-parametric methods, decision tree learning, neural networks, methods combining both inductive and analytic learning, and modeling approaches such as regression models, to train the machine learning system and to evaluate and optimize the performance of the trained system… In one embodiment, the database includes data, for example, of sample conversations with end user utterances and optimal action agent selection, which can assist in the evaluation of a trained system”, wherein that the BRI for top down encompasses the user input into a black box machine learning algorithm and the BRI for bottom up encompasses how the black box output is transformed back up to the user in a human understandable bot response. This end user utterance can then be evaluating and used to optimize the performance of the trained black box system).
It would have been obvious to one of ordinary skill in the art before the effective filing date to add further comprising a step of controlling a network (110) of top-down black boxes using a bottom-up white box, the control step consisting in optimizing (125) said black boxes through machine learning, with the HMI of Lee-Sandford-Wilkinson-Alsadah, with the use of a black box being optimized by white box algorithm feedback, as taught by Cheng. One would have been motivated to the system models of Cheng with HMI of Lee-Sandford-Wilkinson-Alsadah as the combination may improve the quality of the database populating the database with cases in which the user utterances were correctly recognized.
Claim(s) 6 & 8 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication NO. 20210092078 “Lee”, U.S. Patent Application Publication NO. 20190302276 “Sandford”, and further in light of U.S. Patent Application Publication NO. 20070204323 “Wilkinson”, as applied to claim 1 above, and further in light of U.S. Patent Application Publication NO. 20200285741 “Kashyap”.
Claim 6:
Lee, Sandford, and Wilkinson teach the method as claimed in claim 1.
Lee, Sandford, and Wilkinson may not explicitly teach further comprising
a step of selecting a network of universal approximators from among a plurality thereof through machine learning.
However, Kashyap teaches
as a step of selecting a network of universal approximators from among a plurality thereof through machine learning (i.e. para. [0035], “there are a library of available ML packs 134 from which specific ML packs 134 can be selected and implemented as part of the EDR system and/or for use by the monitoring device 130. In some cases, a software tool can provide such recommendations (either by automatically exploring the network topology 100 and characterizing same or by receiving data characterizing the network topology 100). In other variations, there can be segment-specific ML packs 134 (i.e., medical device ML packs, etc.) that can be manually selected and implemented”, wherein the BRI for a universal approximator encompasses selecting a specific ML pack from among a library of ML packs through machine learning recommendations based on data characteristics).
It would have been obvious to one of ordinary skill in the art before the effective filing date to add a step of selecting a network of universal approximators from among a plurality thereof through machine learning, neural network HMI recommendations of Lee-Sandford-Wilkinson, with how machine learning is used to select a recommended ML pack based on learned data characteristic, as taught by Kashyap. One would have been motivated to the Kashyap with Lee-Sandford-Wilkinson as the combination greatly reduce the likelihood and/or amount of sensitive data loss within an enterprise computing topology by choosing an appropriate neural network.
Claim 8:
Lee, Sandford, and Wilkinson teach method as claimed in claim 1.
Lee, Sandford, and Wilkinson may not explicitly teach
wherein a universal approximator is a parameterized function and/or a neural network and/or a CMA-ES algorithm.
However, Kashyap teaches
wherein a universal approximator is a parameterized function and/or a neural network (i.e. para. [0035], “there are a library of available ML packs 134 from which specific ML packs 134 can be selected and implemented as part of the EDR system and/or for use by the monitoring device 130. In some cases, a software tool can provide such recommendations (either by automatically exploring the network topology 100 and characterizing same or by receiving data characterizing the network topology 100). In other variations, there can be segment-specific ML packs 134 (i.e., medical device ML packs, etc.) that can be manually selected and implemented”, wherein the BRI for a universal approximator being a neural network encompasses selecting a specific ML pack from among a library of ML packs through machine learning recommendations based on data characteristics) and/or a CMA-ES algorithm.
Claim(s) 7 & 9-11 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication NO. 20210092078 “Lee”, U.S. Patent Application Publication NO. 20190302276 “Sandford”, and further in light of U.S. Patent Application Publication NO. 20070204323 “Wilkinson”, as applied to claim 4 above, and further in light of U.S. Patent NO. 10489713 “Ernest”, hereinafter Ernest 713’.
Claim 7:
Lee, Sandford, and Wilkinson teach the method as claimed in claim 1.
Lee, Sandford, and Wilkinson may not explicitly teach further comprising
a step of optimizing (125) a graph of fuzzy inference systems FIS of a GFT through machine learning.
However, Ernest 713’ teaches
a step of optimizing (125) a graph of fuzzy inference systems FIS of a GFT through machine learning (i.e. Col. 2 lines, 38-42, “a GA creates and optimizes a Fuzzy Inference System (FIS), or a controller based on fuzzy control. Each chromosome in this GA defines an FIS, and the FIS it creates must be evaluated to evaluate each chromosome's fitness”, wherein the GFT methodology allows for a FIS of a GFT to be optimized through a genetic algorithm, wherein the BRI for machine learning encompasses a genetic algorithm).
It would have been obvious to one of ordinary skill in the art before the effective filing date to add a step of optimizing (125) a graph of fuzzy inference systems FIS of a GFT through machine learning, neural network HMI of Lee-Sandford-Wilkinson, with a type of AI GFT may use fuzzy control optimization to intelligently solve problems as an alternative to more mathematical modes, as taught by Ernest 713’. One would have been motivated to the Ernest 713’ with Lee-Sandford-Wilkinson as the combination would have the algorithms focus on high performance in the most important portion(s) of the solution at each stage of the overall process.
Claim 9:
Lee, Sandford, and Wilkinson teach the method as claimed in any one of the preceding claims.
Lee, Sandford, and Wilkinson may not explicitly teach
wherein the machine learning comprises implementing a genetic algorithm, which determines a configuration of GFT fuzzy decision trees used in the bottom-up and/or top-down translators, notably a configuration of the membership functions and fuzzy rule bases of each fuzzy inference system FIS making up the GFT fuzzy decision tree.
However, Ernest 713’ teaches
wherein the machine learning comprises implementing a genetic algorithm (i.e. Col. 1, lines 33-35, If we implement a search heuristic such as a Genetic Algorithm (GA) over this fuzzy solution space, we can even forgo the expert knowledge. This methodology is referred to as a GFS), which determines a configuration of GFT fuzzy decision trees used in the bottom-up and/or top-down translators (i.e. Wherein it is noted in Fig. 1 that the genetic algorithm takes unmutated parameters and finds a best chromosome, wherein the BRI for bottom-up translators encompasses the saving and loading of best chromosome fitness and population run algorithms the BRI for top-down translators encompasses the steps to use unmutated parameters as input to create a best chromosome and population output), notably a configuration of the membership functions and fuzzy rule bases of each fuzzy inference system FIS making up the GFT fuzzy decision tree (i.e. Col. 4, lines 50-54, Lastly, this system has FISs as part of its makeup; namely, in this embodiment, a fuzzy tree 117. The if-then rule bases and shape of the (preferably triangular) membership functions are defined here as well).
It would have been obvious to one of ordinary skill in the art before the effective filing date to add wherein the machine learning comprises implementing a genetic algorithm, which determines a configuration of GFT fuzzy decision trees used in the bottom-up and/or top-down translators, notably a configuration of the membership functions and fuzzy rule bases of each fuzzy inference system FIS making up the GFT fuzzy decision tree, and the search systems neural network HMI of Lee-Sandford-Wilkinson, with the genetic algorithm and optimization of GFTs, as taught by Ernest 713’. One would have been motivated to the Ernest 713’ with Lee-Sandford-Wilkinson as the combination discretizes the control solution space into a range of possible linguistic terms and their combinations, as altering a predefined parameter is much more computationally efficient than creating a parameter. In many embodiments, this process drastically increases the performance of the present embodiment compared to a GA, and it does so in an unbiased manner, minimizing presumptions regarding the solution space of the AI that the system is optimizing.
Claim 10:
Lee, Sandford, Wilkinson, and Ernest 713’ teach the method as claimed in claim 9.
Ernest 713’ further teaches
wherein said configuration is performed by breaking the membership functions and the rule bases down into a plurality of associated genes, and then randomly mixing them and/or randomly replacing one or more genes with others (i.e. Col. 4, lines 30-45, The present system utilizes tournament-style polling, where a number of chromosomes are selected from the generation at random, and the chromosome with the highest fitness in this selection is chosen for breeding… The if-then rule bases and shape of the (preferably triangular) membership functions are defined here as well).
Claim 11:
Lee, Sandford, Wilkinson, and Ernest 713’ teach the method as claimed in claim 9.
Ernest teaches further comprising a step of using a genetic algorithm to optimize the structure of a fuzzy logic decision tree (i.e. Col. 8, lines 21-22 The system being disclosed has been used to create and optimize a host of GFTs).
Claim(s) 12 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication NO. 20210092078 “Lee”, U.S. Patent Application Publication NO. 20190302276 “Sandford”, further in light of U.S. Patent Application Publication NO. 20070204323 “Wilkinson”, and further in light of U.S. Patent Application Publication NO. 20200341438 “Alsadah”, as applied to claim 4 above, and further in light of Button, K., Miller, A., & Seck, H. H. (2019, January 1). A.I. in the Cockpit. Aerospace America. https://aerospaceamerica.aiaa.org/features/a-i-in-the-cockpit/, hereinafter “Button”.
Claim 12:
Lee, Sandford, Wilkinson, and Alsadah teach the method as claimed in claim 4.
While Alsadah teaches updating current data of a system model in a case where current readings include bad data, Lee, Sandford, Wilkinson, and Alsadah may not explicitly teach
further comprising a step of updating (120) current data relating to an aircraft or its environment, said data independently modifying data from the white boxes and/or black boxes.
However, Button teaches
further comprising a step of updating (120) current data relating to an aircraft or its environment, said data independently modifying the data from white boxes and/or black boxes (i.e. para. [Adapting on the fly], “The algorithms merge real-time aerodynamic modeling with adaptive controls and real-time guidance, and then model the aerodynamics to steadily improve the control of the vehicle… these could help tune autopilots, provide a health monitoring function by detecting aerodynamic changes caused by icing, for example, or help update and tweak the control laws for the plane for optimal performance.”, wherein real-time time data may be used to update the white or black box algorithms used to assist a pilot in a suggestion machine learning model).
It would have been obvious to one of ordinary skill in the art before the effective filing date to add updating (120) current data relating to an aircraft or its environment, said data independently modifying the data from white boxes and/or black boxes, with the HMI using white box fuzzy logic of Lee-Sandford-Wilkinson-Alsadah, with how a black or white box algorithm may be updated with current data relating to an aircraft or its environment, as taught by Button. One would have been motivated to combine Button with of Lee-Sandford-Wilkinson-Alsadah as the combination would help explain to regulators how the software reacts in specific situations as no one as yet wants to turn aircraft or automobiles over to neural nets in which the reasoning logic is impossible to validate.
Claim(s) 13-14 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication NO. 20210092078 “Lee”, U.S. Patent Application Publication NO. 20190302276 “Sandford”, and further in light of U.S. Patent Application Publication NO. 20070204323 “Wilkinson”, and further in light of U.S. Patent NO. 11580358 “Ernest”, hereinafter Ernest 358’.
Claim 13:
Lee, Sandford, and Wilkinson teach the method as claimed in any one of the preceding claims.
Lee, Sandford, and Wilkinson may not explicitly teach
further comprising a step of accessing one or more intermediate values manipulated in the GFT fuzzy logic decision trees.
However, Ernst 358’ teaches
further comprising a step of accessing one or more intermediate values manipulated in the GFT fuzzy logic decision trees (i.e. Col. 14, lines 53-60, “some implementations may additionally or alternatively evaluate only the particular sets of inputs, intermediate values, or other parameters that are encountered during simulation or determination of the objective function”, wherein it is noted that the rule based system for evaluation may be a fuzzy logic system).
It would have been obvious to one of ordinary skill in the art before the effective filing date to add a step of accessing one or more intermediate values manipulated in the GFT fuzzy logic decision trees, to the search systems neural network HMI of Lee-Sandford-Wilkinson, with a step of accessing one or more intermediate values manipulated in the GFT fuzzy logic decision trees, as taught by Ernst 358’. One would have been motivated to the Ernst 358’ with Lee-Sandford-Wilkinson as the combination provides a partial evaluation may be completed far more quickly than an evaluation across the entire input space.
Claim 14:
Lee, Sandford, Wilkinson, and Ernst 358’ teach the method as claimed in any one of the preceding claim.
Ernst 358’ teaches further comprising a step of displaying one or more intermediate values manipulated in the GFTs (i.e. Col. 21 lines 21-25 “In some implementations, instead of or in addition to performing additional training (block 466) automatically, the system may notify (block 468) one or more administrators or other users of the training set deficiency by sending an electronic communication describing the current status of training”, wherein intermediate values may be part of the particular set of inputs used in a deficient training set that would be displayed in a communication).
Claim(s) 15-17 & 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication NO. 20210092078 “Lee”, U.S. Patent Application Publication NO. 20190302276 “Sandford”, and further in light of U.S. Patent Application Publication NO. 20070204323 “Wilkinson”, and further in light of Gunning, D. (2019, February 2). Explainable artificial intelligence (XAI). Nasa.gov. https://asd.gsfc.nasa.gov/conferences/ai/program/003-XAIforNASA.pdf /, hereinafter “Gunning”.
Claim 15:
Lee, Sandford, and Wilkinson teach the method as claimed in any one of the preceding claims.
Lee, Sandford, and Wilkinson may not explicitly teach
wherein one or more of the bottom-up translation white boxes are displayed on demand in a human- machine interface.
However, Gunning teaches,
wherein one or more of the bottom-up translation white boxes are displayed on demand in a human- machine interface (i.e. the examiner notes on pg. 11, that the goal of explainable artificial intelligence is to, “Create a suite of machine learning techniques to produce more explainable models and enable human users to understand, trust, and effectively manage the emerging generation of artificially intelligent partners”, wherein the BRI for a bottom-up translation white box encompasses the machine learning model’s explanation interface that may be transmitted to a human computer interface after processing a result).
It would have been obvious to one of ordinary skill in the art before the effective filing date to add wherein one or more of the bottom-up translation white boxes are displayed on demand in a human- machine interface, search systems neural network HMI of Lee-Sandford-Wilkinson, with how an XAI model produces an explanation interface to a user on demand to evaluate the training of the model as part of its human computer interface, as taught by Gunning. One would have been motivated to the Gunning with Lee-Sandford-Wilkinson as the combination discretizes the control solution space into a range of possible linguistic terms and their combinations, as altering a predefined parameter is much more computationally efficient than creating a parameter. In many embodiments, this process drastically increases the performance of the present embodiment compared to a GA, and it does so in an unbiased manner, minimizing presumptions regarding the solution space of the AI that the system is optimizing.
Claim 16:
Lee, Sandford, Wilkinson, and Gunning teach the method as claimed in any one of the preceding claims.
Gunning further teaches wherein the bottom-up translation is configured through supervised learning (i.e. the examiner notes that in pg. 18, the operator may evaluate the correctability of the XAI system, wherein the BRI for how the bottom-up translation is configured through supervised learning encompasses how the machine to human performance and explanation may be supervised by an operator as it undergoes continuous training).
Claim 17:
Lee, Sandford, Wilkinson, and Gunning teach the method as claimed in any one of the preceding claims.
Gunning further teaches wherein the top- down translation is adjusted or configured through reinforcement learning based on the configured or trained bottom-up translator (i.e. the examiner notes in pg. 13, that the XAI model may undergo a Reinforcement Learning Task Wherein the BRI for a top-down translation encompasses the human to machine steps of inputting a simulated mission and the BRI for a bottom-up translator encompasses the machine to human steps for outputting a recommendation and explained behavior back to an operator. The examiner notes that the future human to machine input is adjusted as the notes that as the operator views the trained XAI model output and decides on which future tasks to delegate to the XAI model).
Claim 19:
Lee, Sandford, Wilkinson, and Gunning teach the method as claimed in any one of the preceding claims.
Gunning teaches wherein one or more of intermediate computing results, information relating to root causes and/or computing context of one or more of the steps of the method are displayed in a human-machine interface (i.e. the examiner notes in pg. 11, that the XAI model has an explanation interface displaying one or more information bits relating to computing context and steps of the method related to the task of the model).
Claim(s) 18 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication NO. 20210092078 “Lee”, U.S. Patent Application Publication NO. 20190302276 “Sandford”, in light of U.S. Patent Application Publication NO. 20070204323 “Wilkinson”, and further in light of U.S. Patent Application Publication NO. 20200341438 “Alsadah”, as applied to claim 4 above, and further in light of U.S. Patent NO. 10489713 “Ernest”, hereinafter Ernest 713’.
Claim 18:
Lee, Sandford, Wilkinson, and Alsadah teach the method as claimed in claim 4.
Lee, Sandford, Wilkinson, and Alsadah may not explicitly teach
wherein the top-down black boxes are put into competition.
However, Ernest 713’ teaches
wherein the top-down black boxes are put into competition (i.e. Col. 6 lines 62-67, “Combined Performance FIS 250 takes into account recent performance history input 230 in order to adapt to those changes. Additionally, if a client takes more than three times the amount of time as the slowest returned client to return its fitness values (in some embodiments, these times are scaled to account for the relative size of assigned chunks), its chunk is re-assigned”, wherein the BRI for top down black-boxes encompasses how any of the GFT algorithm chunks taking input data may be top-down black boxes, wherein the BRI for competition encompasses how the performance history of each chunk of the GFTs may be evaluated against each other).
It would have been obvious to one of ordinary skill in the art before the effective filing date to add wherein the top-down black boxes are put into competition, to the neural network HMI of Lee-Sandford-Wilkinson-Alsadah, with a how black box models are compared against each other, as taught by Ernest 713’. One would have been motivated to the Ernest 713’ with Lee-Sandford-Wilkinson-Alsadah as the combination would have the algorithms focus on high performance in the most important portion(s) of the solution at each stage of the overall process.
Claim(s) 21 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication NO. 20210092078 “Lee”, U.S. Patent Application Publication NO. 20190302276 “Sandford”, and further in light of U.S. Patent Application Publication NO. 20070204323 “Wilkinson”, as applied to claim 1 above, and further in light of U.S. Patent NO. 5983144 “Bonissone”.
Claim 21:
Lee, Sandford, and Wilkinson teach the method as claimed in claim 1.
Lee, Sandford, and Wilkinson may not explicitly teach
wherein the fuzzy logic decision trees use words from a finite dictionary and carrying semantics.
However, Bonissone teaches
wherein the fuzzy logic decision trees use words from a finite dictionary and carrying semantics (i.e. Col. 5, lines 20-40, “The fuzzy logic PI controller as shown in FIG. 3 comprises a knowledge base 28 having a rule set, term sets… The rule set maps linguistic descriptions of state vectors such as e and .DELTA.e into the incremental control actions .DELTA.u; the term sets define the semantics of the linguistic values used in the rule sets”, wherein the BRI for a finite dictionary encompasses defined linguistic rule set with associated semantics).
It would have been obvious to one of ordinary skill in the art before the effective filing date to add wherein the fuzzy logic decision trees use words from a finite dictionary and carrying semantics, to the fuzzy logic HMI of Lee-Sandford-Wilkinson, with wherein the fuzzy logic uses words from a finite dictionary and carrying semantics, as taught by Bonissone. One would have been motivated to the Bonissone with Lee-Sandford-Wilkinson as the combination increases the fine control of the fuzzy logic controller.
Claim(s) 23 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication NO. 20210092078 “Lee”, U.S. Patent Application Publication NO. 20190302276 “Sandford”, and further in light of U.S. Patent Application Publication NO. 20070204323 “Wilkinson”, as applied to claim 1 above, and further in light of U.S. Patent Application Publication NO. 20070011108 “Benson”.
Claim 23:
Lee, Sandford, and Wilkinson teach the method as claimed in any one of the preceding claims.
Lee, Sandford, and Wilkinson may not explicitly teach wherein one or more data processing operations are governed by a certified avionics flight management system FMS internalizing predefined constraints.
However, Benson teaches
wherein one or more data processing operations are governed by a certified avionics flight management system FMS internalizing predefined constraints (i.e. para. [0167]. “Inside the flight deck the system constantly updates the information provided by the prediction system, along with readings from the avionics systems aboard the aircraft. Current conditions and flight system information is run through the fuzzy rule process based on information from the rule library”, wherein the information processing operations are governed by an avionics system internalizing pre-defined information from a rule library).
It would have been obvious to one of ordinary skill in the art before the effective filing date to add wherein one or more data processing operations are governed by a certified avionics flight management system FMS internalizing predefined constraints, to the fuzzy logic HMI of Lee-Sandford-Wilkinson, with the fuzzy logic is applied to an avionics system data and affected by associated flight rules, as taught by Benson. One would have been motivated to combine Benson with Lee-Sandford-Wilkinson as the combination synchronized data collection which results in an improved system for making decisions and for carrying out related tasks.
Claim(s) 26 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication NO. 20210092078 “Lee”, U.S. Patent Application Publication NO. 20190302276 “Sandford”, and further in light of U.S. Patent Application Publication NO. 20070204323 “Wilkinson”, as applied to claim 25 above, and further in light of U.S. Patent Application Publication NO. 20200202723 “Pierre”.
Claim 26:
Lee, Sandford, and Wilkinson teach the system as claimed in claim 25.
Lee teaches further comprising one or more neural networks configured for machine learning, said one or more neural networks being chosen from among neural networks comprising: an artificial neural network (i.e. para. [0055], the machine learning component 302 can combine a recurrent neural network model and a word embedding mode); an acyclic artificial neural network; a recurrent neural network; a forward propagation neural network; a convolutional neural network; a generative adversarial neural network; said one or more neural networks being emulated in software form and/or being physical circuits (i.e. para. [0109], As another example, a component can be an apparatus with specific functionality provided by mechanical parts operated by electric or electronic circuitry).
While Lee, Sandford, and Wilkinson teach that a neural network may be selected from a recurrent neural network model being a physical circuit, Lee, Sandford, and Wilkinson may not explicitly teach
said one or more neural networks being chosen from among neural networks comprising: an acyclic artificial neural network; a forward propagation neural network; a convolutional neural network; a generative adversarial neural network;
However, Pierre teaches said one or more neural networks being chosen from among neural networks comprising: an acyclic artificial neural network; a forward propagation neural network; a convolutional neural network; a generative adversarial neural network (i.e. para. [0018], the system further comprises one or more neural networks chosen from the neural networks comprising: an artificial neural network; an acyclic artificial neural network; a recurrent neural network; a feedforward neural network; a convolutional neural network; and/or a generative adversarial neural network)
It would have been obvious to one of ordinary skill in the art before the effective filing date to add said one or more neural networks being chosen from among neural networks comprising: an acyclic artificial neural network; a forward propagation neural network; a convolutional neural network; a generative adversarial neural network, to the fuzzy logic HMI of Lee-Sandford-Wilkinson, with neural network selection pool types, as taught by Pierre. One would have been motivated to Pierre or with Lee-Sandford-Wilkinson as the combination provide a more varied system for aircraft performance to be predicted, and independently of models provided by manufacturers.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
U.S. Patent Application Publication NO. 20170076600 “Scofield” teaches in para. [0069], FIG. 12 presents an illustration of a set of exemplary scenarios 1200 whereby the device 202 may advise the user 102 as to the selected lane 218. As a first variation 1202 of this fifth aspect, a visual and/or audial indicator may be presented to the user 102 by the device 202 and/or vehicle 104, such as a light on the dashboard of the vehicle 104 or an audio or voice cue 1204 prompting the user to select a particular lane 112.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DAVID H TAN whose telephone number is (571)272-7433. The examiner can normally be reached M-F 7:30-4:30.
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, Cesar Paula can be reached at (571) 272-4128. 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.
/D.T./Examiner, Art Unit 2145
/CESAR B PAULA/Supervisory Patent Examiner, Art Unit 2145
