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 .
DETAILED ACTION
This office action is in response to the claims filed on 03/26/2026.
Claims 1-3, 6-20 are presented for examination.
Response to Argument
In reference to applicant’s argument regrading rejections under 35 U.S.C. § 112(f):
Applicant’s Argument:
In reply to the interpretation of claims 1-5, 7-18, and 20 under 35 U.S.C. 112(f) as containing means-plus-function claim recitations, the Applicant respectfully requests reconsideration. The Applicant respectfully submits that the there is no intention for any of the claims to be interpreted as having means-plus- function recitations and the Applicant believes that the claims pre-amendment do not invoke 35 U.S.C. 112(f). Notwithstanding, in the spirit of expediting patent prosecution, the Applicant has amended the claims for clarification to include the recitations of "at least one processor" or a "computer" as structural elements, as well as other clarifying claim amendments. In view of these amendments, the statement of the Applicant intentions, and the substance of the claims, the Applicant respectfully submits that none of the pending claims should be interpreted as containing means-plus-function claim recitations under 35 U.S.C. 112(f).
Examiner’s Response:
The 112(f) rejection is withdrawn in view of the claim amendment filed on 03/26/2026.
In reference to applicant’s argument regrading rejections under 35 U.S.C. § 112(a):
Applicant’s Argument:
In reply to the rejection of claims 1-18 and 20 under 35 U.S.C. 112(a) as purportedly failing to comply with the written description requirement, the Applicant respectfully requests reconsideration. This rejection is effectively an extension of the interpretation of the claims under 35 U.S.C. 112(f) as containing means-plus-function recitations. It appears that the theory of this rejection is that since the claims allegedly include means-plus-function recitations, then the claims are interpreted according to the embodiments disclosed in the specification and the embodiments disclosed in the specification are not specific enough for one of ordinary skill in the art to practice the invention. The Applicant respectfully traverses all aspects of this rejection. First, the claims do not include means-plus-function recitations (at least post-amendment), as discussed above. Second, in arguendo, the disclosure in the specification is adequate to enable those of ordinary skill in the art to practice the invention. At least for these reasons, this rejection should be withdrawn.
Examiner’s Response:
The 112(a) rejection is withdrawn in view of the claim amendment filed on 03/26/2026.
In reference to applicant’s argument regrading rejections under 35 U.S.C. § 101:
Applicant’s Argument regarding the 101 on pages 10-17 in the remark filed on 03/26/2026:
On December 5, 2025, the following paragraph was added to the end of MPEP 2106.04(d), subsection III: In Ex Parte Desjardins, Appeal No. 2024- 000567 (PTAB September 26, 2025, Appeals Review Panel Decision) (precedential), the claimed invention was a method of training a machine learning model on a series of tasks. The Appeals Review Panel (ARP) overall credited benefits including reduced storage, reduced system complexity and streamlining, and preservation of performance attributes associated with earlier tasks during subsequent computational tasks as technological improvements that were disclosed in the patent application specification… Ex Parte Desjardins, Appeal No. 2024-000567 (PTAB September 26, 2025, Appeals Review Panel Decision) (precedential). In view of these significant revisions to the MPEP that reflect both a change of law and a change of policy at the U.S. Patent and Trademark Office, the Applicant respectfully submits that this rejection under 35 U.S.C. 101 is hereby moot. It is notable that the circumstances in the patent application under examination are analogous to the facts set forth in Ex Parte Desjardins. In arguendo and in view of MPEP 2106.05(g), at a minimum, the claims (as amended) relate to a "significant extra-solution activity' because: 1) The recitations of the claims are not well known, as evidenced by the weakness of the prior art rejections (discussed below). 2) The recitations of the claims are significant, as evidence from both the weakness of the prior art rejections and the practical applications of the embodiments covered by the claims. 3) The recitations of the claims are significantly more than "necessary data
gathering and outputting". In view of the above, favorable reconsideration and withdraw of the rejections under 35 U.S.C. 101 are respectfully requested.
Examiner’s Response:
In light of the amendments made, the rejection of claims 1-20 under 35 USC 101 have been withdrawn.
In reference to applicant’s argument regrading rejections under 35 U.S.C. § 103:
Applicant’s Argument:
The applicant’s argument regarding the 103 rejection based on the claim amendment filed on 03/26/2026.
Examiner’s Response:
This argument includes the newly amended limitations. It has been fully considered but is moot in view of the new grounds of rejection presented below necessitated by the amendment. See 20200410163
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 for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1, 2, 3, 6-8, 11-20 are rejected under 35 U.S.C. 103 as being unpatentable over Watanabe et al. (Pub. No. US 20190137969-hereinafter, Watanabe) in view of Martinson et al. (PUB. No. US 20180053102 -hereinafter, Martinson) and of OGAWA et al. (PUB. No. US 20180107947 -hereinafter, OGAWA) and further in view of Satou et al. (PUB. No. US 20190056718 -hereinafter, Satou) and further in view of Yoshida et al. (PUB. No. US 20220143823 -hereinafter, Yoshida) and further in view of Satou et al. (PUB. No. US 20190056718 -hereinafter, Satou).
Regrading claim 1, Watanabe teaches Watanabe teaches a learning device comprising at least one processor wherein: the at one processor acquire, initialization data (Watanabe, [Par.0036-0041, Fig.2], “The data acquisition unit 2 is functional means that acquires data including at least one of internal data and external data indicating an environment from a manufacturing machine or a factory corresponding to an environment for the machine learning device 100. For example, the data acquisition unit 2 acquires data of a current value of the servomotor 50 or the spindle motor 62 driving each unit of the manufacturing machine controlled by the controller 1, a detection value of a temperature, vibration, etc. detected by a sensor installed in each unit of the manufacturing machine, etc…[0040] The input safety circuit 120 is functional means that detects an abnormal value of input data observed by the state observation unit 110 and input to the machine learning unit 130, executes an operation at the time of detecting an abnormal value when the abnormal value of the input data is detected, and outputs the input data as safety measurement data to the machine learning unit 130 when the abnormal value of the input data is not detected…[0041], At the time of detecting an abnormal value of input data, the input safety circuit 120 executes an operation related to the input data determined in advance such as reacquisition of the input data, correction of the input data, or discard of the input data.” Examiner’s note, the data acquisition unit to collect the data before the input safety circuit input into the machine learning unit.),
initialization data including state data indicating a state of the equipment (Watanabe, [Par.0036], “The data acquisition unit 2 is functional means that acquires data including at least one of internal data and external data indicating an environment from a manufacturing machine or a factory corresponding to an environment for the machine learning device 100. For example, the data acquisition unit 2 acquires data of a current value of the servomotor 50 or the spindle motor 62 driving each unit of the manufacturing machine controlled by the controller 1, a detection value of a temperature, vibration, etc. detected by a sensor installed in each unit of the manufacturing machine, etc…” Examiner’s note, the data indicating the environment from the manufacturing machine or factory, is considered as the initialization data. )
and action data indicating an action on the control target (Watanabe, [Par.0040-0041], “[0040], The input safety circuit 120 is functional means that detects an abnormal value of input data observed by the state observation unit 110 and input to the machine learning unit 130, executes an operation at the time of detecting an abnormal value when the abnormal value of the input data is detected, and outputs the input data as safety measurement data to the machine learning unit 130 when the abnormal value of the input data is not detected…[0041], ] At the time of detecting an abnormal value of input data, the input safety circuit 120 executes an operation related to the input data determined in advance such as reacquisition of the input data, correction of the input data, or discard of the input data. For example, the input safety circuit 120 may execute the operation related to the input data determined in advance when a predetermined condition is satisfied. As an example, when there is a time margin for reacquiring the input data, the input safety circuit 120 may command the state observation unit 110 (furthermore, the data acquisition unit 2) to reacquire the input data. In this case, for example, when an abnormal value is continuously detected a predetermined number of times at the time of reacquiring the input data, the input safety circuit 120 may abandon reacquisition of the input data. As another example, when safety measurement data in a current operation cycle of the controller 1 is indispensable (when the machine learning device 100 needs to necessarily output safety inference data in the current operation cycle of the controller 1, etc.), the input safety circuit 120 may correct a value of a data item of input data in which an abnormal value is detected based on input data input in the past, and output the corrected input data as safety measurement data.” Examiner’s note, the measurement data is considered as the action data, that indicates the action of the input safety circuit whether reacquisition of the input data, correction of the input data, or discard of the input data.);
the at least one processor acquires the initialization data before the control of the control target by a machine learning model; (Watanabe, [Par.0036-0041, Fig.2], “The data acquisition unit 2 is functional means that acquires data including at least one of internal data and external data indicating an environment from a manufacturing machine or a factory corresponding to an environment for the machine learning device 100. For example, the data acquisition unit 2 acquires data of a current value of the servomotor 50 or the spindle motor 62 driving each unit of the manufacturing machine controlled by the controller 1, a detection value of a temperature, vibration, etc. detected by a sensor installed in each unit of the manufacturing machine, etc….” Examiner’s note, the data acquisition unit to collect the data before the input safety circuit input into the machine learning unit.),
However, Watanabe does not teach the control target is provided in the equipment; a machine learning model outputs an action corresponding to a state of the equipment to control the control target; the at least one processor defines an option for the machine learning model to choose the action; and the at least one processor extracts, as the sample data, a combination of the state data included in the initialization data and the action included in the option. and a preliminary learning unit configured to initialize the machine learning model by performing preliminary learning on a basis of the initialization data before start of reinforcement learning of the machine learning model,
On the other hand, Martinson teaches and the at least one processor extracts, as the sample data (Martinson, [Par.0008], “In a first step, Jain describes extracting features from input sensor data, such as high-level features from a driver-facing camera to detect a driver's head pose, object features from a road-facing camera to determine a road occupancy status, etc. However, Jain's approach requires a substantial amount of human involvement, which makes it impractical for dynamic systems and possibly dangerous. Further, the number of sensory inputs considered by Jain is not representative of typical human driving experiences, and the model is unable to consider important features affecting a driver's action, such as steering patterns, local familiarity,” [Par.0014], “[0014] According to another innovative aspect of the subject matter described in this disclose, a system may include one or more computer processors and one or more non-transitory memories storing instructions that, when executed by the one or more computer processors, cause the computer system to perform operations comprising: aggregating local sensor data from a plurality of vehicle system sensors during operation of vehicle by a driver; detecting, during the operation of the vehicle, a driver action using the local sensor data; and extracting, during the operation of the vehicle, features related to predicting driver action from the local sensor data. The operations may also include adapting, during operation of the vehicle, a stock machine learning-based driver action prediction model to a customized machine learning-based driver action prediction model using one or more of the extracted features and the detected driver action,” and [Par.0061], “In some implementations, the processor(s) 213 may be capable of generating and providing electronic display signals to a display device (not shown), supporting the display of images, capturing and transmitting images, performing complex tasks including various types of feature extraction and sampling” Examiner’s note, the features (position of the driver’s head) is extracted from the input sensor data (sample data) by the processor (extracting unit) to predict the driver’s action by the machine learning model.),
Watanabe and Martinson are analogous in arts because they have the same field of endeavor of generating the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to a learning device comprising at least one processor wherein: the at one processor acquire, initialization data, as taught by Watanabe to include the and the at least processor extract, as the sample data, as taught by Martinson. The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the road safety, (Martinson, [par.0004], “Being able to predict a driver action a few seconds ahead of the action may greatly improve the efficiency and usefulness of such advance driver assistance systems. In particular, an advance driver assistance system that can predict actions further in advance and with greater accuracy will enable new advance driver assistance system functionality, such as automatic turn and braking signals, which can further improve road safety”).
However, neither Watanabe nor Martinson teaches the control target is provided in the equipment; a machine learning model outputs an action corresponding to a state of the equipment to control the control target; the at least one processor defines an option for the machine learning model to choose the action; as the sample data, a combination of the state data included in the initialization data and the action included in the option. and a preliminary learning unit configured to initialize the machine learning model by performing preliminary learning on a basis of the initialization data before start of reinforcement learning of the machine learning model,
On the other hand, OGAWA teaches the at least one processor defines an option for the machine learning model to choose the action (OGAWA, [Par.0058-0059], “[0058], The machine learning apparatus has the function of extracting useful regularity, knowledge representation, criteria and the like from the aggregate of data input to the apparatus by analysis, outputting the results of the determination, and performing knowledge learning (machine learning). There are various machine learning techniques, which are broadly classified into “supervised learning”, “unsupervised learning”, and “reinforcement learning”, for example. Furthermore, in order to implement these techniques, a technique called “deep learning” is available that learns extraction of a feature quantity itself. [0059Note that “reinforcement learning (Q-learning)” is applied to the machine learning apparatus 2 that is illustrated in FIG. 6. A general-purpose computer or processor can be used for the machine learning apparatus 2. Alternatively, when general-purpose computing on graphics processing units (GPGPU), a large scale PC cluster, or the like is applied to the machine learning apparatus 2,” [Par.0070-0072], “Here, reinforcement learning (Q-learning) is a technique that learns actions as well as determination and classification so as to learn an appropriate action in consideration of interaction provided to the environment by an action, that is, the technique that learns the method for maximizing the reward acquired in the future. Q-learning is exemplified in the following description, however, the invention is not limited to the case to which the Q-learning is adopted.[0071] The Q-learning is a method for learning value Q(s, a) for selecting an action a under a given environment state s. That is, the action a that has highest value Q(s, a) is selected as an optimal action in the given state s. However, a correct value of value Q(s, a) for a combination of the state s and the action a is completely unknown in the beginning. Thus, an agent (a subject of actions) selects various actions a under the given state s, and a reward is provided to each of the selected actions a. In this way, the agent learns selection of a better action, that is, correct value Q(s, a).[0072] Furthermore, in order to maximize the total of rewards acquired in the future as a result of actions, the final objective is to satisfy the equation: Q(s, a)=E[Σ(γ.sup.t)r.sub.t]. Here, expectation value is acquired when a state changes in response to an optimal action.” Examiner’s note, the q learning method for selecting the correct action under the environment state, for example, the performance of the Q-learning/ reinforcement learning for selecting an action a under a given environment state s. That is, the action that has highest value Q(s, a) is selected as an optimal action in the given state s, wherein, the q learning is performed by the processor (definition unit). The Q learning to select the action. ),
as the sample data a combination of the state data included in the initialization data and the action included in the option. (Ogawa, [Par.0072-0074], “[0072] Furthermore, in order to maximize the total of rewards acquired in the future as a result of actions, the final objective is to satisfy the equation: Q(s, a)=E[Σ(γ.sup.t)r.sub.t]. Here, expectation value is acquired when a state changes in response to an optimal action. The expectation value, which is unknown, is learned through search. The update equation of value Q(s, a) is represented by, for example, the following Equation (1):..[0073] In the Equation (1), s.sub.t represents the state of the environment at time t, and a.sub.t represents the action at time t. The state is changed to s.sub.t+1 by the action a.sub.t. The reward that is acquired by the change of the state is represented by r.sub.t+1. The term with “max” is a Q value that is multiplied by γ when the action a that has the highest known Q value is selected under the state s.sub.t+1. Here, γ is a parameter of 0<γ≤1, called discount factor. Symbol α is a learning coefficient in the range of 0<α≤1.[0074] The above-described Equation (1) represents a method for updating the evaluation value Q(s.sub.t, a.sub.t) of the action a.sub.t in the state s.sub.t on the basis of the reward r.sub.t+1 that is returned as a result of the action a.sub.t. That is, this indicates that if the sum of the reward r.sub.t+1 and an evaluation value Q(s.sub.t+1, max a.sub.t+1) of the best action max a in a subsequent state caused by the action a is greater than the evaluation value Q(s.sub.t, a.sub.t) of the action a in the state s, Q(s.sub.t, a.sub.t) is increased. In contrast, if the sum is less than the evaluation value Q(s.sub.t, a.sub.t), Q(s.sub.t, a.sub.t) is decreased. In other words, value of a given action in a given state is brought closer to a reward immediately returned as a result and value of the best action in a subsequent state caused by the given action.” Examiner’s note, the state data include the particular state data at the particular time and the action data.).
Watanabe, Martinson and Ogawa are analogous in arts because they have the same field of endeavor of generating the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the combined teaching of Watanabe and Martinson of the learning device according to claim 2, the at least one processor extract, as the sample data, as set forth above, to include the processor define an option for the machine learning model to choose the action, a combination of the state data included in the initialization data and the action included in the option, as taught by OWAGA. The modification would have been obvious because one of the ordinary skills in art would be motivated to select the best action, (Ogawa, [0071] The Q-learning is a method for learning value Q(s, a) for selecting an action a under a given environment state s. That is, the action a that has highest value Q(s, a) is selected as an optimal action in the given state s. However, a correct value of value Q(s, a) for a combination of the state s and the action a is completely unknown in the beginning. Thus, an agent (a subject of actions) selects various actions a under the given state s, and a reward is provided to each of the selected actions a. In this way, the agent learns selection of a better action, that is, correct value Q(s, a)).
However, Watanabe and Martinson do not teach the control target is provided in the equipment; a machine learning model outputs an action corresponding to a state of the equipment to control the control target; and a preliminary learning unit configured to initialize the machine learning model by performing preliminary learning on a basis of the initialization data before start of reinforcement learning of the machine learning model,
On the other hand, Satou teaches the control target is provided in the equipment, (Satou, [Par.0039], “A controller 1 for a conveying machine can be embodied as a controller that controls a conveying machine such as a conveyor (not shown) for conveying a conveyance article, a machine (not shown) for conveying a container, a pack, or the like filled with a liquid, an automatic conveying vehicle (not shown) for conveying a conveyance article from a predetermined position to another position, or a robot (not shown),” Examiner’s note, the conveyance article is controlled by the controller at the conveyor, therefore, the conveyance article is considered as the control target is provided in the conveyor (equipment).
a machine learning model outputs an action corresponding to a state of the equipment to control the control target (Satou, [Par.0054, 0064], “By executing this learning cycle repeatedly, the learning unit 110 can automatically identify a characteristic that implies a correlation between the state of the conveyance article (the conveyance article state data S2) and the conveyance operation of the conveying machine 70 relative to the state. At the start of the learning algorithm, the correlation between the conveyance article state data S2 and the conveyance operation of the conveying machine 70 is substantially unknown, but the learning unit 110 interprets the correlation by identifying the characteristic gradually as learning progresses. Once the correlation between the conveyance article state data S2 and the conveyance operation of the conveying machine 70 has been analyzed to a relatively reliable standard, the learning result output iteratively by the learning unit 110 can be used to select an action (i.e., for decision-making), or in other words to determine the control of the conveyance operation of the conveying machine 70 relative to the current state (in other words, the state in which the conveyance article is conveyed by the conveying machine)…[Par.0064], “Reinforcement learning is a method of iteratively executing, by trial and error, a cycle of observing a current state of an environment in which a learning subject exists (i.e. input), executing a predetermined action in the current state (i.e. output), and rewarding the action, and then learning a policy (in the machine learning device according to this application, determining the conveyance operation of the conveying machine 70) with which the total reward is maximized as an optimal solution.” Examiner’s note, the machine learning determines the action in the current state to control the conveyance operation at the conveying machine.) .;
Watanabe, Martinson, Ogawa and Satou are analogous in arts because they have the same field of endeavor of generating the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the initialization data including state data indicating the a state of the a equipment and action data indicating an action on the a control target, as taught by Watanabe, to include the the control target is provided in the equipment; a machine learning model outputs an action corresponding to a state of the equipment to control the control target, as taught by Satou. The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the seft-determined machine operation, (Satou, [Par.0095], “he decision-making unit 122 outputs a command value C for setting the conveyance operation of the conveying machine 70 determined on the basis of the learning result to the controller 2 of the conveying machine. By implementing this control period repeatedly, the machine learning device 120 advances learning of the conveyance operation of the conveying machine 70, and as a result, the reliability of the self-determined conveyance operation of the conveying machine 70 gradually improves.”).
However Watanabe, Martinson, Ogawa and Satou do not teach and a preliminary learning unit configured to initialize the machine learning model by performing preliminary learning on a basis of the initialization data before start of reinforcement learning of the machine learning model,
On the other hand, Yoshida teaches a preliminary learning unit configured to initialize the machine learning model by performing preliminary learning on a basis of the initialization data before start of reinforcement learning of the machine learning model (Yoshida, [Par.0014], “In order to solve the above-mentioned problems, the present invention employs the means indicated below. That is, the present invention provides a learning system for an operation inference learning model for controlling an automatic driving robot, the learning system training the operation inference learning model by reinforcement learning, and comprising the operation inference learning model, which infers operations of a vehicle for making the vehicle run in accordance with a defined command vehicle speed based on a running state of the vehicle including a vehicle speed, and the automatic driving robot, which is installed in the vehicle and which makes the vehicle run based on the operations, wherein the learning system comprises a vehicle learning model that has been trained by machine learning to simulate actions of the vehicle based on an actual running history of the vehicle, and that outputs a simulated running state, which is the running state simulating the vehicle based on the operations inferred by the operation inference learning model; and the operation inference learning model is pre-trained by reinforcement learning by applying the simulated running state output by the vehicle learning model to the operation inference learning model, and after the pre-training by reinforcement learning has ended, the operation inference learning model is further trained by reinforcement learning by applying, to the operation inference learning model, the running state acquired by the vehicle being run based on the operations inferred by the operation inference learning model.” Examiner’s note, a vehicle learning model that has been trained by machine learning to simulate actions of the vehicle based on an actual running history of the vehicle before the start of the reinforcement learning by applying the simulated running state output by the vehicle learning model to the operation inference learning model.).
Watanabe, Martinson, Ogawa, Satou and Yoshida are analogous in arts because they have the same field of endeavor of generating the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the learning device comprising: a data acquisition unit configured to acquire, before control of a control target provided in equipment by a machine learning model, as taught by Watanabe to include the at least one processor initializes the machine learning model by performing preliminary learning on a basis of the initialization data before start of reinforcement learning of the machine learning model, as taught by Yoshida. The modification would have been obvious because one of the ordinary skills in art would be motivated to reducing undesirable vehicle operation, (Yoshida , [0016] The present invention can provide a learning system and a learning method for an operation inference learning model for controlling an automatic driving robot (drive robot) that can reduce stress on an actual vehicle by reducing undesirable vehicle operation outputs by the operation inference learning model during reinforcement learning, and that can improve the accuracy of operations output by the operation inference learning model.”).
Regarding claim 2, Watanabe teaches the learning device according to claim 1, but it does not teach further comprising: an extraction unit configured to extract sample data to be used for initialization of the machine learning model from the initialization data.
On the other hand, Watanabe teaches further comprising: an extraction unit configured to extract sample data (Martinson, [Par.0008], “In a first step, Jain describes extracting features from input sensor data, such as high-level features from a driver-facing camera to detect a driver's head pose, object features from a road-facing camera to determine a road occupancy status, etc. However, Jain's approach requires a substantial amount of human involvement, which makes it impractical for dynamic systems and possibly dangerous. Further, the number of sensory inputs considered by Jain is not representative of typical human driving experiences, and the model is unable to consider important features affecting a driver's action, such as steering patterns, local familiarity,” [Par.0014], “[0014] According to another innovative aspect of the subject matter described in this disclose, a system may include one or more computer processors and one or more non-transitory memories storing instructions that, when executed by the one or more computer processors, cause the computer system to perform operations comprising: aggregating local sensor data from a plurality of vehicle system sensors during operation of vehicle by a driver; detecting, during the operation of the vehicle, a driver action using the local sensor data; and extracting, during the operation of the vehicle, features related to predicting driver action from the local sensor data. The operations may also include adapting, during operation of the vehicle, a stock machine learning-based driver action prediction model to a customized machine learning-based driver action prediction model using one or more of the extracted features and the detected driver action,” and [Par.0061], “In some implementations, the processor(s) 213 may be capable of generating and providing electronic display signals to a display device (not shown), supporting the display of images, capturing and transmitting images, performing complex tasks including various types of feature extraction and sampling” Examiner’s note, the features (position of the driver’s head) is extracted from the input sensor data (sample data) by the processor (extracting unit) to predict the driver’s action by the machine learning model.)
Watanabe and Martinson are analogous in arts because they have the same field of endeavor of generating the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify learning device, as taught by Watanabe to include the extraction unit configured to extract sample data, as taught by Martinson. The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the road safety, (Martinson, [par.0004], “Being able to predict a driver action a few seconds ahead of the action may greatly improve the efficiency and usefulness of such advance driver assistance systems. In particular, an advance driver assistance system that can predict actions further in advance and with greater accuracy will enable new advance driver assistance system functionality, such as automatic turn and braking signals, which can further improve road safety”).
Regrading claim 3, Watanabe in view of Martinson teaches the learning device according to claim 2, wherein the extraction unit includes a selection unit configured to select the initialization data, and the extraction unit is configured to extract the sample data from the selected initialization data and the extraction unit is configured to extract the sample data from the selected initialization data (Martinson,[Par.0008], “In a first step, Jain describes extracting features from input sensor data, such as high-level features from a driver-facing camera to detect a driver's head pose, object features from a road-facing camera to determine a road occupancy status, etc. However, Jain's approach requires a substantial amount of human involvement, which makes it impractical for dynamic systems and possibly dangerous. Further, the number of sensory inputs considered by Jain is not representative of typical human driving experiences, and the model is unable to consider important features affecting a driver's action, such as steering patterns, local familiarity,” [Par.0014], “[0014] According to another innovative aspect of the subject matter described in this disclose, a system may include one or more computer processors and one or more non-transitory memories storing instructions that, when executed by the one or more computer processors, cause the computer system to perform operations comprising: aggregating local sensor data from a plurality of vehicle system sensors during operation of vehicle by a driver; detecting, during the operation of the vehicle, a driver action using the local sensor data; and extracting, during the operation of the vehicle, features related to predicting driver action from the local sensor data. The operations may also include adapting, during operation of the vehicle, a stock machine learning-based driver action prediction model to a customized machine learning-based driver action prediction model using one or more of the extracted features and the detected driver action,”, [Par.0028], “The technology described herein may efficiently and effectively model a driver's behavior based on the sensor data capturing the internal and external environments of a moving platform 101.” and [Par.0061], “In some implementations, the processor(s) 213 may be capable of generating and providing electronic display signals to a display device (not shown), supporting the display of images, capturing and transmitting images, performing complex tasks including various types of feature extraction and sampling” Examiner’s note, the features related to the driver action (position of the driver’s head) is selected and extracted by the processor \from the input sensor data (sample data), wherein, the sensor data is collected by the sensor.).
Watanabe and Martinson are analogous in arts because they have the same field of endeavor of generating the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify learning device, as taught by Watanabe to include the extraction unit includes a selection unit configured to select the initialization data, and the extraction unit is configured to extract the sample data from the selected initialization data and the extraction unit is configured to extract the sample data from the selected initialization data, as taught by Martinson. The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the road safety, (Martinson, [par.0004], “Being able to predict a driver action a few seconds ahead of the action may greatly improve the efficiency and usefulness of such advance driver assistance systems. In particular, an advance driver assistance system that can predict actions further in advance and with greater accuracy will enable new advance driver assistance system functionality, such as automatic turn and braking signals, which can further improve road safety”).
Regarding claim 6, the combined teaching of Watanabe and Martinson teach the learning device according to claim 4, but they do not teach wherein the machine learning model is configured to output the action corresponding to the state of the equipment on a basis of each weight for combinations of the state data included in the initialization data and actions included in the option.
On the other hand, Ogawa teaches wherein the machine learning model is configured to output the action corresponding to the state of the equipment on a basis of each weight for combinations of the state data included in the initialization data and actions included in the option (Ogawa, [0044] The state observation unit 21 acquires the thermal displacement amount of the machine tool 1 during the period of machining the workpiece, as a state variable” and [Par.0074-0075], “The above-described Equation (1) represents a method for updating the evaluation value Q(s.sub.t, a.sub.t) of the action a.sub.t in the state s.sub.t on the basis of the reward r.sub.t+1 that is returned as a result of the action a.sub.t. That is, this indicates that if the sum of the reward r.sub.t+1 and an evaluation value Q(s.sub.t+1, max a.sub.t+1) of the best action max a in a subsequent state caused by the action a is greater than the evaluation value Q(s.sub.t, a.sub.t) of the action a in the state s, Q(s.sub.t, a.sub.t) is increased. In contrast, if the sum is less than the evaluation value Q(s.sub.t, a.sub.t), Q(s.sub.t, a.sub.t) is decreased. In other words, value of a given action in a given state is brought closer to a reward immediately returned as a result and value of the best action in a subsequent state caused by the given action. [0075] Here, Q(s, a) is represented on a computer by a method in which the values of all state and action pairs (s, a) are stored in a table or by a method in which a function of approximating Q(s, a) is prepared. By the latter method, the above-described Equation (1) can be achieved by adjusting a parameter of an approximate function by a technique such as stochastic gradient descent. A neural network, which will be described later, can be used for the approximate function.” Examiner’s note, the best action is selected based on the Q value corresponding to the state data regarding the state of the machine and the actions.) .
Watanabe, Martinson and Ogawa are analogous in arts because they have the same field of endeavor of generating the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the combined teaching of Watanabe and Martinson of the learning device, as set forth above, to include the the extraction unit includes the machine learning model is configured to output the action corresponding to the state of the equipment on a basis of each weight for combinations of the state data included in the initialization data and actions included in the option, as taught by Ogawa. The modification would have been obvious because one of the ordinary skills in art would be motivated to select the best action, (Ogawa, [0071] The Q-learning is a method for learning value Q(s, a) for selecting an action a under a given environment state s. That is, the action a that has highest value Q(s, a) is selected as an optimal action in the given state s. However, a correct value of value Q(s, a) for a combination of the state s and the action a is completely unknown in the beginning. Thus, an agent (a subject of actions) selects various actions a under the given state s, and a reward is provided to each of the selected actions a. In this way, the agent learns selection of a better action, that is, correct value Q(s, a)).
Regarding claim 7, the combined teaching of Watanabe and Martinson teach the learning device according to claim 4, but they do not teach wherein the definition unit is configured to define the option on a basis of a distribution of actions indicated by the action data included in the initialization data.
On the other hand, Ogawa teaches wherein the definition unit is configured to define the option on a basis of a distribution of actions indicated by the action data included in the initialization data (Ogawa, [Par.0059], “Note that “reinforcement learning (Q-learning)” is applied to the machine learning apparatus 2 that is illustrated in FIG. 6. A general-purpose computer or processor can be used for the machine learning apparatus 2. Alternatively, when general-purpose computing on graphics processing units (GPGPU), a large scale PC cluster, or the like is applied to the machine learning apparatus 2,” and [Par.0071-0075], “The Q-learning is a method for learning value Q(s, a) for selecting an action a under a given environment state s. That is, the action a that has highest value Q(s, a) is selected as an optimal action in the given state s. However, a correct value of value Q(s, a) for a combination of the state s and the action a is completely unknown in the beginning. Thus, an agent (a subject of actions) selects various actions a under the given state s, and a reward is provided to each of the selected actions a. In this way, the agent learns selection of a better action, that is, correct value Q(s, a)…[Par.0075], Here, Q(s, a) is represented on a computer by a method in which the values of all state and action pairs (s, a) are stored in a table or by a method in which a function of approximating Q(s, a) is prepared. By the latter method, the above-described Equation (1) can be achieved by adjusting a parameter of an approximate function by a technique such as stochastic gradient descent. A neural network, which will be described later, can be used for the approximate function.” Examiner’s note, the action selection is performed by the processor. ).
Watanabe, Martinson and Ogawa are analogous in arts because they have the same field of endeavor of generating the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the combined teaching of Watanabe and Martinson of the learning device, as set forth above, to include the definition unit is configured to define the option on a basis of a distribution of actions indicated by the action data included in the initialization data, as taught by Ogawa. The modification would have been obvious because one of the ordinary skills in art would be motivated to select the best action, (Ogawa, [0071] The Q-learning is a method for learning value Q(s, a) for selecting an action a under a given environment state s. That is, the action a that has highest value Q(s, a) is selected as an optimal action in the given state s. However, a correct value of value Q(s, a) for a combination of the state s and the action a is completely unknown in the beginning. Thus, an agent (a subject of actions) selects various actions a under the given state s, and a reward is provided to each of the selected actions a. In this way, the agent learns selection of a better action, that is, correct value Q(s, a)).
Regarding the claim 8 is rejected for the same reason as the claim 7, since these claims recite the same limitation.
Regarding claim 11, the combined teaching of Watanabe and Martinson teach the learning device according to claim 4, but they do not teach the definition unit is configured to define a plurality of the options corresponding to the state of the equipment.
On the other hand, Ogawa teaches the definition unit is configured to define a plurality of the options corresponding to the state of the equipment (Ogawa, [Par.0059, 0071-0074], “[Par.0059], Note that “reinforcement learning (Q-learning)” is applied to the machine learning apparatus 2 that is illustrated in FIG. 6. A general-purpose computer or processor can be used for the machine learning apparatus 2. Alternatively, when general-purpose computing on graphics processing units (GPGPU), a large scale PC cluster, or the like is applied to the machine learning apparatus 2,” “[0071], The Q-learning is a method for learning value Q(s, a) for selecting an action a under a given environment state s. That is, the action a that has highest value Q(s, a) is selected as an optimal action in the given state s. However, a correct value of value Q(s, a) for a combination of the state s and the action a is completely unknown in the beginning. Thus, an agent (a subject of actions) selects various actions a under the given state s, and a reward is provided to each of the selected actions a. In this way, the agent learns selection of a better action, that is, correct value Q(s, a)…[0074], The above-described Equation (1) represents a method for updating the evaluation value Q(s.sub.t, a.sub.t) of the action a.sub.t in the state s.sub.t on the basis of the reward r.sub.t+1 that is returned as a result of the action a.sub.t. That is, this indicates that if the sum of the reward r.sub.t+1 and an evaluation value Q(s.sub.t+1, max a.sub.t+1) of the best action max a in a subsequent state caused by the action a is greater than the evaluation value Q(s.sub.t, a.sub.t) of the action a in the state s, Q(s.sub.t, a.sub.t) is increased. In contrast, if the sum is less than the evaluation value Q(s.sub.t, a.sub.t), Q(s.sub.t, a.sub.t) is decreased. In other words, value of a given action in a given state is brought closer to a reward immediately returned as a result and value of the best action in a subsequent state caused by the given action.” Examiner’s note, selecting the best action with the higher q value corresponding to state data, wherein the state data is the state of the equipment, as it can be seen at [0044] The state observation unit 21 acquires the thermal displacement amount of the machine tool 1 during the period of machining the workpiece, as a state variable”.).
Watanabe, Martinson and Ogawa are analogous in arts because they have the same field of endeavor of generating the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the combined teaching of Watanabe and Martinson of the learning device, as set forth above, to include the definition unit is configured to define a plurality of the options corresponding to the state of the equipment, as taught by Ogawa. The modification would have been obvious because one of the ordinary skills in art would be motivated to select the best action, (Ogawa, [0071] The Q-learning is a method for learning value Q(s, a) for selecting an action a under a given environment state s. That is, the action a that has highest value Q(s, a) is selected as an optimal action in the given state s. However, a correct value of value Q(s, a) for a combination of the state s and the action a is completely unknown in the beginning. Thus, an agent (a subject of actions) selects various actions a under the given state s, and a reward is provided to each of the selected actions a. In this way, the agent learns selection of a better action, that is, correct value Q(s, a)).
Regarding the claim 12 is rejected for the same reason as the claim 11, since these claims recite the same limitation.
Regarding claim 13, Watanabe teaches the learning device according to claim 1, wherein the data acquisition unit is configured to acquire the state data in response to control of the control target by the machine learning model (Watanabe, [Par.0036-0041, Fig.2], “The data acquisition unit 2 is functional means that acquires data including at least one of internal data and external data indicating an environment from a manufacturing machine or a factory corresponding to an environment for the machine learning device 100. For example, the data acquisition unit 2 acquires data of a current value of the servomotor 50 or the spindle motor 62 driving each unit of the manufacturing machine controlled by the controller 1, a detection value of a temperature, vibration, etc. detected by a sensor installed in each unit of the manufacturing machine, etc…[0040] The input safety circuit 120 is functional means that detects an abnormal value of input data observed by the state observation unit 110 and input to the machine learning unit 130, executes an operation at the time of detecting an abnormal value when the abnormal value of the input data is detected, and outputs the input data as safety measurement data to the machine learning unit 130 when the abnormal value of the input data is not detected…[0041], At the time of detecting an abnormal value of input data, the input safety circuit 120 executes an operation related to the input data determined in advance such as reacquisition of the input data, correction of the input data, or discard of the input data.” Examiner’s note, the data acquisition unit to collect the data before the input safety circuit input into the machine learning unit.)
the learning device further comprising: a reinforcement learning unit configured to update the machine learning model by performing reinforcement learning using, as learning data, the state data (Watanabe, [Par.0039], “Further, for example, when the learning model 136 is a learning model for so-called supervised learning, the state observation unit 110 observes state data and label data as input data for learning by the learning unit 132, and observes state data as input data for inference by the inference unit 134. In addition, for example, when the learning model 136 is a learning model for so-called reinforcement learning, the state observation unit 110 observes state data and determination data as input data for learning by the learning unit 132, and observes state data as input data for inference by the inference unit 134.” And [Par.0044], “The learning unit 132 included in the machine learning unit 130 is functional means that executes learning of the learning model 136 based on the safety measurement data output from the input safety circuit 120. The learning unit 132 updates the learning model 136 by performing learning according to an algorithm of the learning model 136 based on the safety measurement data. The learning unit 132 may switch ON/OFF a learning process by manipulation of a worker of a manufacturer or an operator, etc. In addition, the learning unit 132 may not be an indispensable configuration of the machine learning device 100 after learning of the learning model 136 is completed. For example, when the manufacturer ships the controller 1 to a customer, management using only an inference function may be performed without causing extra learning at a factory of the customer by incorporating only the inference unit 134 and the learning model 136 as the machine learning unit 130 and performing shipping.”).
However, Watanabe does not teach and the action data acquired from the machine learning model in response to input of the state data to the machine learning model,
On the other hand, Ogawa teaches the action data acquired from the machine learning model in response to input of the state data to the machine learning model (OGAWA, [Par.0071-0075], “The Q-learning is a method for learning value Q(s, a) for selecting an action a under a given environment state s. That is, the action a that has highest value Q(s, a) is selected as an optimal action in the given state s. However, a correct value of value Q(s, a) for a combination of the state s and the action a is completely unknown in the beginning. Thus, an agent (a subject of actions) selects various actions a under the given state s, and a reward is provided to each of the selected actions a. In this way, the agent learns selection of a better action, that is, correct value Q(s, a)…[0074], The above-described Equation (1) represents a method for updating the evaluation value Q(s.sub.t, a.sub.t) of the action a.sub.t in the state s.sub.t on the basis of the reward r.sub.t+1 that is returned as a result of the action a.sub.t. That is, this indicates that if the sum of the reward r.sub.t+1 and an evaluation value Q(s.sub.t+1, max a.sub.t+1) of the best action max a in a subsequent state caused by the action a is greater than the evaluation value Q(s.sub.t, a.sub.t) of the action a in the state s, Q(s.sub.t, a.sub.t) is increased. In contrast, if the sum is less than the evaluation value Q(s.sub.t, a.sub.t), Q(s.sub.t, a.sub.t) is decreased. In other words, value of a given action in a given state is brought closer to a reward immediately returned as a result and value of the best action in a subsequent state caused by the given action.”). .
Watanabe and Ogawa are analogous in arts because they have the same field of endeavor of generating the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the data acquisition unit is configured to acquire the state data in response to control of the control target by the machine learning model, the learning device further comprising: a reinforcement learning unit configured to update the machine learning model by performing reinforcement learning using, as learning data, the state data, as taught by Watanabe, to include the action data acquired from the machine learning model in response to input of the state data to the machine learning model, as taught by Ogawa. The modification would have been obvious because one of the ordinary skills in art would be motivated to optimize the action at the given state, (Ogawa, [0071] The Q-learning is a method for learning value Q(s, a) for selecting an action a under a given environment state s. That is, the action a that has highest value Q(s, a) is selected as an optimal action in the given state s. However, a correct value of value Q(s, a) for a combination of the state s and the action a is completely unknown in the beginning. Thus, an agent (a subject of actions) selects various actions a under the given state s, and a reward is provided to each of the selected actions a. In this way, the agent learns selection of a better action, that is, correct value Q(s, a)).
Regarding the claim 14 is rejected for the same reason as the claim 13, since these claims recite the same limitation.
Regarding claim 15 Watanabe in view of Ogawa teaches the learning device according to claim 13, wherein the preliminary learning unit is configured to initialize the machine learning model on a basis of the initialization data to choose an action closer to the action data corresponding to the state data in response to input of the state data (OGAWA, [Par.0071-0075], “The Q-learning is a method for learning value Q(s, a) for selecting an action a under a given environment state s. That is, the action a that has highest value Q(s, a) is selected as an optimal action in the given state s. However, a correct value of value Q(s, a) for a combination of the state s and the action a is completely unknown in the beginning. Thus, an agent (a subject of actions) selects various actions a under the given state s, and a reward is provided to each of the selected actions a. In this way, the agent learns selection of a better action, that is, correct value Q(s, a)…[0074], The above-described Equation (1) represents a method for updating the evaluation value Q(s.sub.t, a.sub.t) of the action a.sub.t in the state s.sub.t on the basis of the reward r.sub.t+1 that is returned as a result of the action a.sub.t. That is, this indicates that if the sum of the reward r.sub.t+1 and an evaluation value Q(s.sub.t+1, max a.sub.t+1) of the best action max a in a subsequent state caused by the action a is greater than the evaluation value Q(s.sub.t, a.sub.t) of the action a in the state s, Q(s.sub.t, a.sub.t) is increased. In contrast, if the sum is less than the evaluation value Q(s.sub.t, a.sub.t), Q(s.sub.t, a.sub.t) is decreased. In other words, value of a given action in a given state is brought closer to a reward immediately returned as a result and value of the best action in a subsequent state caused by the given action.”).,
and the reinforcement learning unit is configured to update the machine learning model to further increase a reward obtained by a series of actions (OGAWA, [Par.0071-0075], “The Q-learning is a method for learning value Q(s, a) for selecting an action a under a given environment state s. That is, the action a that has highest value Q(s, a) is selected as an optimal action in the given state s. However, a correct value of value Q(s, a) for a combination of the state s and the action a is completely unknown in the beginning. Thus, an agent (a subject of actions) selects various actions a under the given state s, and a reward is provided to each of the selected actions a. In this way, the agent learns selection of a better action, that is, correct value Q(s, a)…[0074], The above-described Equation (1) represents a method for updating the evaluation value Q(s.sub.t, a.sub.t) of the action a.sub.t in the state s.sub.t on the basis of the reward r.sub.t+1 that is returned as a result of the action a.sub.t. That is, this indicates that if the sum of the reward r.sub.t+1 and an evaluation value Q(s.sub.t+1, max a.sub.t+1) of the best action max a in a subsequent state caused by the action a is greater than the evaluation value Q(s.sub.t, a.sub.t) of the action a in the state s, Q(s.sub.t, a.sub.t) is increased. In contrast, if the sum is less than the evaluation value Q(s.sub.t, a.sub.t), Q(s.sub.t, a.sub.t) is decreased. In other words, value of a given action in a given state is brought closer to a reward immediately returned as a result and value of the best action in a subsequent state caused by the given action.”)..
Watanabe and Ogawa are analogous in arts because they have the same field of endeavor of generating the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify the learning device, as taught by Watanabe, to include the learning device according to claim 13, wherein the preliminary learning unit is configured to initialize the machine learning model on a basis of the initialization data to choose an action closer to the action data corresponding to the state data in response to input of the state data, and the reinforcement learning unit is configured to update the machine learning model to further increase a reward obtained by a series of actions., as taught by Ogawa. The modification would have been obvious because one of the ordinary skills in art would be motivated to optimize the action at the given state, (Ogawa, [0071] The Q-learning is a method for learning value Q(s, a) for selecting an action a under a given environment state s. That is, the action a that has highest value Q(s, a) is selected as an optimal action in the given state s. However, a correct value of value Q(s, a) for a combination of the state s and the action a is completely unknown in the beginning. Thus, an agent (a subject of actions) selects various actions a under the given state s, and a reward is provided to each of the selected actions a. In this way, the agent learns selection of a better action, that is, correct value Q(s, a)).
Regarding claim 16, Watanabe teaches a control device comprising: the learning device according to claim 1; and a control unit configured to control the control target by the machine learning model (Watanabe, [Par.0036-0041, Fig.2], “The data acquisition unit 2 is functional means that acquires data including at least one of internal data and external data indicating an environment from a manufacturing machine or a factory corresponding to an environment for the machine learning device 100. For example, the data acquisition unit 2 acquires data of a current value of the servomotor 50 or the spindle motor 62 driving each unit of the manufacturing machine controlled by the controller 1, a detection value of a temperature, vibration, etc. detected by a sensor installed in each unit of the manufacturing machine, etc…[0040] The input safety circuit 120 is functional means that detects an abnormal value of input data observed by the state observation unit 110 and input to the machine learning unit 130, executes an operation at the time of detecting an abnormal value when the abnormal value of the input data is detected, and outputs the input data as safety measurement data to the machine learning unit 130 when the abnormal value of the input data is not detected…[0041], At the time of detecting an abnormal value of input data, the input safety circuit 120 executes an operation related to the input data determined in advance such as reacquisition of the input data, correction of the input data, or discard of the input data.”.
Regarding the claim 17 is rejected for the same reason as the claim 16, since these claims recite the same limitation.
Regarding the claim 18 is rejected for the same reason as the claim 16, since these claims recite the same limitation.
Regarding the claim 19 is rejected for the same reason as the claim 1, since these claims recite the same limitation.
Regarding the claim 20 is rejected for the same reason as the claim 1, since these claims recite the same limitation.
Claims 9, 10 are rejected under 35 U.S.C. 103 as being unpatentable over Watanabe (Pub. No. US 20190137969-hereinafter, Watanabe) in view of Martinson et al. (PUB. No. US 20180053102 -hereinafter, Martinson) and of OGAWA et al. (PUB. No. US 20180107947 -hereinafter, OGAWA) and further in view of Satou et al. (PUB. No. US 20190056718 -hereinafter, Satou) and further in view of Yoshida et al. (PUB. No. US 20220143823 -hereinafter, Yoshida) and further in view of Satou et al. (PUB. No. US 20190056718 -hereinafter, Satou) and further in view of Vah et al. (PUB. No. US 20210263792 -hereinafter, Vah).
Regarding claim 9, Watanabe and Ogawa teach the learning device according to claim 4, wherein the definition unit, but it does not teach wherein the definition unit is configured to define a common option regardless of the state of the equipment.
However, Vah teaches wherein the at least one processor defines a common option regardless of the state of the equipment (Vah, [0062-0064], “FIGS. 3A-3D illustrate a workflow for pre-screening troubleshooting action recommendations using an ensemble machine learning model approach. The workflow begins with a user (e.g., a technician) invoking an expert troubleshooting system 301 in path 1. The user in step 302 requests a troubleshooting recommendation for a particular asset (e.g., a computing device such as a laptop). A first machine learning model in step 303 generates a recommended troubleshooting action. The first machine learning model, in some embodiments, is a conversational machine learning model that treats the steps of a troubleshooting process as a conversation. In other embodiments, however, various other types of machine learning models may be utilized to generate the recommended troubleshooting action. From here, the workflow branches into path 2 in step 304 if the recommended troubleshooting action is a repair action or into path 3 in step 305 if the recommended troubleshooting action is a diagnostic action.” Examiner’s note, the machine learning model to predict whether the action is repair actions, or diagnostic actions .).
action.”)..
Watanabe, Ogawa and Vah are analogous in arts because they have the same field of endeavor of generating the machine learning model.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to modify combined teaching of Watanabe and Ogawa of the definition unit, as set forth above, to include the definition unit is configured to define a common option regardless of the state of the equipment, as taught by Vah. The modification would have been obvious because one of the ordinary skills in art would be motivated to predict the success of the action , (Vah, [Par.0071], “In the FIG. 3A-3D system flow, the first machine learning model provides high-level functions of generating recommended troubleshooting actions (e.g., repair actions, diagnostic actions, etc.). The second machine learning model takes as input information of the given asset being troubleshooted (e.g., asset description such as product platform, symptoms and error descriptions, the recommended troubleshooting action generated by the first machine learning model) and generates an output indicating whether the outcome of the recommended troubleshooting action is desirable or not as per various KPIs. The KPIs may include whether a repair action is likely to result in VFF or NFF, whether a diagnostic action is likely to be effective or ineffective, whether a repair or diagnostic action is complex (e.g., time-consuming, requiring special training or knowledge) or trivial, etc.”).
Regarding the claim 10 is rejected for the same reason as the claim 9, since these claims recite the same limitation.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to EM N TRIEU whose telephone number is (571)272-5747. The examiner can normally be reached on Mon-Fri from 9:00-5:00.
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, Omar Fernandez Rivas can be reached on (571) 272-2589. 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.
/E.T./Examiner, Art Unit 2128
/OMAR F FERNANDEZ RIVAS/Supervisory Patent Examiner, Art Unit 2128
Prosecution Insights