DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claims 1-6, 8-10, and 12 are presented for examination.
Continued Examination under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on March 23, 2026 has been entered.
Response to Amendment
Applicant’s amendment has obviated the claim objections and the rejections under 35 USC § 101. Therefore, those objections and rejections are withdrawn (but see new ground of claim objection infra).
Priority
Receipt is acknowledged of certified copies of papers required by 37 CFR 1.55.
Claim Objections
Claim 10 is objected to because of the following informalities: “network and is trained” should be “network is trained”. Appropriate correction is required.
Claim Rejections - 35 USC § 103
Claims 1-2, 4, 6, 8-10, and 12 are rejected under 35 U.S.C. 103 as being unpatentable over Anisimov et al. (US 20200364539) (“Anisimov”) in view of Davies et al. (US 20200003659) (“Davies”) and further in view of Glugla (US 20180320611) (“Glugla”) and Hackworth et al. (US 20090242197) (“Hackworth”).
Regarding claim 1, Anisimov discloses “[a] method …, the method comprising:
receiving a sensor signal time series from a sensor (measurements of eye tracking data consist of a multichannel time series – Anisimov, paragraph 222; see also paragraph 90 (disclosing that the eye positions may be tracked with sensors)) …;
determining an evaluation signal time series within an evaluation time window of a sensor signal time series (data recorded by an EEG may be considered as a sequence of overlapping time interval windows, wherein the length of each window is similar to each other, and the time shift between the windows may be chosen from 0 ms to window length/2; each window consists of fragments of a multichannel EEG time series [evaluation signal time series = combination of data from two or more windows of time; evaluation time window = combination of two or more windows, but fewer than all windows, of the time series; sensor signal time series = whole time series including all the windows] – Anisimov, paragraph 223; EEG bioresponse measurement device has a number of sensors for measuring bioresponses of the user or the training user during the data recording sessions – id. at paragraph 118 [i.e., the time series is a sensor signal time series]);
determining sensor signal extracts from the evaluation signal time series, the sensor signal extracts being time-shifted with respect to one another or respectively offset from one another by a number of sensing steps, the sensor signal extracts being shorter in length than the evaluation signal time series (data recorded by an EEG may be considered as a sequence of overlapping time interval windows, wherein the length of each window is similar to each other, and the time shift between the windows may be chosen from 0 ms to window length/2; each window consists of fragments of a multichannel EEG time series [evaluation signal time series = combination of data from two or more windows of time; sensor signal extracts = single windows of the time series, which are shorter in length than the evaluation time series and are offset from each other by up to window length/2] – Anisimov, paragraph 223);
determining one or more frequency contributions from the sensor signal extracts using a fast Fourier transform (“FFT”) or a Goertzel algorithm (a feature extraction method based on FFT is applied to each window [sensor signal extract], generating a tensor of extracted features [frequency contributions] for each considered window of the EEG signal – Anisimov, paragraph 223; see also paragraph 132 (indicating that the features may include frequency domain features such as power)); … [and]
evaluating the one or more frequency contributions in a trained data-based sensor model (tensors produced by FFT [frequency contributions] are used as input for [i.e., evaluated by] a convolutional neural network of the training machine learning model [trained data-based sensor model] – Anisimov, paragraph 223) ….”
Anisimov appears not to disclose explicitly the further limitations of the claim. However, Davies discloses “determining a change-point time within the evaluation time window (method of training a machine-learning classifier comprises, inter alia, extracting an event section of the input signal comprising a section [evaluation time window] of the signal from a point prior to a change to a point after the change [i.e., including a change-point time], extracting pre- and post-event portions of the event section and transforming them into the frequency domain, determining a probability that an appliance is in a degradation state by applying the machine learning classifier to a feature vector comprising the frequency-domain data, and training the machine learning classifier using the annotated feature vectors – Davies, paragraphs 339-48) ...; [and]
monitoring a quantity … based on the change-point time (appliance operation signal [quantity] may be monitored for a change in magnitude, i.e., an event [change-point] – Davies, paragraph 147) ….”
Davies and the instant application both relate to analysis of time-series data using machine learning and are analogous. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Anisimov such that the system determines a change-point time, as disclosed by Davies, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would allow the system to pinpoint change times automatically, thereby allowing the user to make necessary corrections to the system being evaluated. See Davies, paragraphs 4 and 339-48.
Neither Anisimov nor Davies appears to disclose explicitly the further limitations of the claim. However, Glugla discloses a “method for operating a fuel injection system of an internal combustion engine (Glugla Fig. 4 depicts a method for adjusting fuel injection amounts based on estimations of airflow into a motor vehicle engine), the method comprising:
receiving a sensor signal … from a sensor arranged within … the fuel injection system (amount of exhaust gas recirculation (EGR) entering the intake manifold may be determined based on an EGR flow rate as measured with an EGR differential pressure sensor – Glugla, paragraph 72; fuel injection amount is calculated based on the cylinder air amount calculated from the EGR amount – id. at paragraphs 74-75 [i.e., the sensor is part of the injector system]); …
monitoring a quantity of fuel that has been injected (a fuel injection amount may be calculated [monitored] based on an air cylinder amount – Glugla, paragraph 75) …; and
controlling an opening or closing of the injector valve of the fuel injection system depending on the monitored quantity of fuel (fuel control correction may be determined based on exhaust gas oxygen sensor feedback and the fuel injection amount may be adjusted by, for instance, increasing a fuel pulse-width on the next fuel injection event [i.e., controlling a valve to let in more fuel] – Glugla, paragraph 76).….”
Glugla and the instant application both relate to computerized engine control systems and are analogous. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Anisimov and Davies to use the method to operate a fuel injector system of a vehicle, as disclosed by Glugla, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would improve fuel economy and extend the life of engine components. See Glugla, paragraph 10.
Hackworth discloses “a sensor arranged within an injector valve (sensors may be located inside an injection valve without the addition of a separate sensor carrier – Hackworth, paragraph 56) ….”
Hackworth and the instant application both relate to injector valves and are analogous. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Anisimov, Davies, and Glugla to arrange the sensor inside the injection valve, as disclosed by Hackworth, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would obviate the need for more complex systems to communicate with the sensor. See Hackworth, paragraph 56.
Regarding claim 2, Anisimov, as modified by Davies, Hackworth, and Glugla, discloses that “the trained data-based sensor model is trained to respectively associate a corresponding change-point time with the one or more frequency contributions from an evaluation-point time series (method of training a machine-learning classifier comprises, inter alia, extracting an event section of the input signal [evaluation-point time series] comprising a section of the signal from a point prior to a change to a point after the change, extracting pre- and post-event portions of the event section and transforming them into the frequency domain [i.e., associating the change-point time with frequency contributions], determining a probability that an appliance is in a degradation state by applying the machine learning classifier to a feature vector comprising the frequency-domain data, and training the machine learning classifier using the annotated feature vectors – Davies, paragraphs 339-48).” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Anisimov/Glugla/Hackworth to associate the change-point time with frequency-domain contributions, as disclosed by Davies, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would allow the system to pinpoint change times automatically, thereby allowing the user to make necessary corrections to the system being evaluated. See Davies, paragraphs 4 and 339-48.
Regarding claim 4, Anisimov, as modified by Davies/Glugla/Hackworth, discloses that “the trained data-based sensor model is configured as a single or multilayer neural network (invention relates to determining the correlation of the eye-movements and/or facial expressions of a user who is consuming visual information; the determination of the correlation may include training one or more machine learning models that include deep [multilayer] neural networks within their architecture – Anisimov, paragraph 18; tensors produced by FFT are used as input for a convolutional neural network of the training machine learning model – id. at paragraph 223).”
Regarding claim 6, Anisimov, as modified by Davies/Glugla/Hackworth, discloses “[a] device comprising: a computer configured to carry out the method according to claim 1 (system may include a processor [computer] that is connected to memory with a variety of bridges and controllers that may reside between the processor and memory – Anisimov, paragraph 92).”
Regarding claim 8, Anisimov, as modified by Davies/Glugla/Hackworth, discloses “[a] non-transitory machine-readable storage medium comprising instructions which, when executed by a computer, cause the computer to execute the method according to claim 1 (system may include a processor that is connected to memory with a variety of bridges and controllers that may reside between the processor and memory [non-transitory machine-readable storage medium] – Anisimov, paragraph 92).”
Regarding claim 9, Anisimov discloses “[a] method for training a deep neural network to determine change-point times in a sensor signal, the method comprising:
providing a plurality of training datasets, each training dataset including an evaluation signal time series (separate normalization of each word from different sources may provide average values for a gaze tracks feature for each normalized word over all occurrences in the training data set; for example, the sum of the gaze tracks features of all n-occurrences of the word “water” are averaged over the number of occurrences n – Anisimov, paragraph 195; see also paragraph 221 (describing the dataset gathered as a multidimensional time series; note that the dataset may be divided into subsets to form a plurality)) …, each evaluation signal time series corresponding to a respective evaluation time window of a sensor signal time series from a sensor (data recorded by an EEG may be considered as a sequence of overlapping time interval windows, wherein the length of each window is similar to each other, and the time shift between the windows may be chosen from 0 ms to window length/2; each window consists of fragments of a multichannel EEG time series [evaluation signal time series = combination of data from two or more windows of time; sensor signal extracts = single windows of the time series] –Anisimov, paragraph 223; see also paragraph 90 (disclosing the use of sensors to collect the raw data)) …;
determining sensor signal extracts from each evaluation signal time series of the plurality of training datasets, the sensor signal extracts being time-shifted with respect to one another or respectively offset from one another by a number of sensing steps, the sensor signal extracts being shorter in length than the evaluation signal time series (data recorded by an EEG may be considered as a sequence of overlapping time interval windows, wherein the length of each window is similar to each other, and the time shift between the windows may be chosen from 0 ms to window length/2; each window consists of fragments of a multichannel EEG time series [evaluation signal time series = combination of data from two or more windows of time; sensor signal extracts = single windows of the time series, which are shorter in length than the evaluation signal time series and are offset from each other by up to window length/2] – Anisimov, paragraph 223);
determining one or more frequency contributions from the sensor signal extracts of each evaluation signal time series using a fast Fourier transform (“FFT”) or a Goertzel algorithm (a feature extraction method based on FFT is applied to each window [sensor signal extract], generating a tensor of extracted features for each considered window of the EEG signal – Anisimov, paragraph 223; see also paragraph 132 (indicating that the features may include frequency domain features such as power)); and
training the deep neural network … using the one or more frequency contributions (tensors produced by FFT [frequency contributions] are used as input for [i.e., evaluated by] a convolutional neural network of the training machine learning model [deep learning model] – Anisimov, paragraph 223; see also paragraph 18 (disclosing training of the models)) ….”
Anisimov appears not to disclose explicitly the further limitations of the claim. However, Davies discloses “providing training datasets … including a change-point time (method of training a machine-learning classifier comprises, inter alia, extracting an event section of the input signal comprising a section of the signal from a point prior to a change to a point after the change, extracting pre- and post-event portions of the event section and transforming them into the frequency domain, determining a probability that an appliance is in a degradation state by applying the machine learning classifier to a feature vector comprising the frequency-domain data, and training the machine learning classifier using the annotated feature vectors [i.e., the training dataset comprises feature vectors derived from change-point times] – Davies, paragraphs 339-48); … [and]
training the … model to determine change-point times in the evaluation signal time series using the … change-point times associated therewith (method of training a machine-learning classifier comprises, inter alia, extracting an event section of the input signal comprising a section of the signal from a point prior to a change to a point after the change, extracting pre- and post-event portions of the event section and transforming them into the frequency domain, determining a probability that an appliance is in a degradation state by applying the machine learning classifier to a feature vector comprising the frequency-domain data, and training the machine learning classifier using the annotated feature vectors [i.e., the training dataset comprises feature vectors derived from change-point times and is used to determine change-point times] – Davies, paragraphs 339-48).” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Anisimov to train the model using change-point times, as disclosed by Davies, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would allow the system to pinpoint change times automatically, thereby allowing the user to make necessary corrections to the system being evaluated. See Davies, paragraphs 4 and 339-48.
Neither Anisimov nor Davies appears to disclose explicitly the further limitations of the claim. However, Glugla discloses that the “sensor [is] arranged within … a fuel injection system of an internal combustion engine (amount of exhaust gas recirculation (EGR) entering the intake manifold may be determined based on an EGR flow rate as measured with an EGR differential pressure sensor – Glugla, paragraph 72; fuel injection amount is calculated based on the cylinder air amount calculated from the EGR amount – id. at paragraphs 74-75 [i.e., the sensor is part of the injector system]) ….” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Anisimov and Davies to perform the method using sensor data from a fuel injector, as disclosed by Glugla, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would improve fuel economy and extend the life of engine components. See Glugla, paragraph 10.
Hackworth discloses “a sensor arranged within an injector valve (sensors may be located inside an injection valve without the addition of a separate sensor carrier – Hackworth, paragraph 56) ….” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Anisimov, Davies, and Glugla to arrange the sensor inside the injection valve, as disclosed by Hackworth, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would obviate the need for more complex systems to communicate with the sensor. See Hackworth, paragraph 56.
Regarding claim 10, the rejection of claim 9 is incorporated. Anisimov further discloses a “deep neural network (invention relates to determining the correlation of the eye-movements and/or facial expressions of a user who is consuming visual information; the determination of the correlation may include training one or more machine learning models that include deep neural networks within their architecture – Anisimov, paragraph 18) ….”
Anisimov/Glugla/Hackworth appears not to disclose explicitly the further limitations of the claim. However, Davies discloses that “the … network … is trained using a backpropagation-based training method (training process may involve, inter alia, back-propagating the error and updating the neural network parameters accordingly – Davies, paragraphs 241-45).” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Anisimov/Glugla/Hackworth to train the model using backpropagation, as disclosed by Davies, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would increase the accuracy of the network by ensuring that it has the correct weights. See Davies, paragraph 253.
Regarding claim 12, Anisimov, as modified by Davies, Hackworth, and Glugla, discloses that “the sensor is a piezo sensor configured to measure pressure changes in fuel being conducted by the injector valve (pressure-sensitive disc such as a piezo ceramic disc may be included in the differential pressure sensor between upstream and downstream pressure ports – Glugla, paragraph 31).” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Anisimov, Hackworth, and Davies to use a piezo sensor, as disclosed by Glugla, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would improve fuel economy and extend the life of engine components. See Glugla, paragraph 10.
Claim 3 is rejected under 35 U.S.C. 103 as being unpatentable over Anisimov in view of Davies, Glugla, and Hackworth and further in view of Prater et al. (US 20200309681) (“Prater”).
Regarding claim 3, neither Anisimov, Glugla, Hackworth, nor Davies appears to disclose explicitly the further limitations of the claim. However, Prater discloses that “the one or more frequency contributions are determined based on one or more predetermined frequencies or a phase state of an underlying sine or cosine signal (each spectrum is decomposed into a sum of appropriate transform basis functions with associated coefficients; one example of a commonly used transform is the FFT, where a signal is decomposed into a sum of sines and cosines of different frequencies and associated coefficients – Prater, paragraph 72).”
Prater and the instant application both relate to frequency-domain analysis of signals and are analogous. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Anisimov, Glugla, Hackworth, and Davies to use the FFT to decompose the signal into sines and cosines whose frequencies can then be analyzed, as disclosed by Prater, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would allow the frequencies to be calculated efficiently in closed-form fashion. See Prater, paragraph 72.
Claim 5 is rejected under 35 U.S.C. 103 as being unpatentable over Anisimov in view of Davies, Hackworth, and Glugla and further in view of Adlakha et al. (WO 2017214613) (“Adlakha”).
Regarding claim 5, the rejection of claim 1 is incorporated. Davies further discloses that “the trained data-based sensor model is configured to indicate the change-point time as a classification vector (method of training a machine-learning classifier comprises, inter alia, extracting an event section of the input signal comprising a section of the signal from a point prior to a change to a point after the change, extracting pre- and post-event portions of the event section and transforming them into the frequency domain, determining a probability that an appliance is in a degradation state by applying the machine learning classifier to a feature vector comprising the frequency-domain data, and training the machine learning classifier using the annotated feature vectors [i.e., the change-point information is transformed into a feature vector/classification vector] – Davies, paragraphs 339-48) ….” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Anisimov/Glugla/Hackworth to organize the change-point information as a feature vector, as disclosed by Davies, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would allow the system to pinpoint change times automatically, thereby allowing the user to make necessary corrections to the system being evaluated. See Davies, paragraphs 4 and 339-48.
Neither Anisimov, Glugla, Hackworth, nor Davies appears to disclose explicitly the further limitations of the claim. However, Adlakha discloses that “the change-point time is indicated as an argmax of the classification [data] (change point (e.g., the point where a different state or behavior is recognized as being distinct from another or previous state) is given by an argmax of a log likelihood ratio [classification data] – Adlakha, paragraph 83).”
Adlakha and the instant application both relate to change-point detection and are analogous. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Anisimov, Hackworth, and Davies to calculate the change-point time using an argmax, as disclosed by Adlakha, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would allow the system to monitor performance problems related to the changes, thereby enhancing the performance of the system. See Adlakha, paragraph 3.
Response to Arguments
Applicant's arguments filed March 23, 2026 (“Remarks”) have been fully considered but they are, except insofar as rendered moot by the withdrawal of a ground of rejection, not persuasive.
Applicant argues that the cited references allegedly do not render the amended claims obvious because (a) the EGR sensor of Glugla is not arranged within an injector valve; (b) the combination does not teach determining a change-point time using a data-based sensor model because Anisimov fails to disclose determining a change-point time and Davies fails to disclose that a data-based sensor model is used; and (c) the combination does not teach operating the fuel injection system depending on the change-point time because the fuel injection amount of Glugla is not based on a change-point time and Davies does not relate to fuel-quantity monitoring. Remarks at 13-16.
Regarding (a), without conceding that Glugla does not teach this element and purely in the interest of expediting prosecution, newly cited reference Hackworth is used to teach a sensor inside an injector valve. Therefore, this argument is moot.
Regarding (b), this argument is a classic case of attacking references individually when the rejection is based on the combination. It is not necessary that Anisimov teach determining change-point times because Davies does, and it is not necessary that Davies teach using a data-based sensor model because Anisimov does. Applicant does not attack the reason for combining the two, but rather argues that each reference fails to teach an element that the rejection does not say it teaches. One cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986).
Regarding (c), here again Applicant is improperly attacking references individually. Davies teaches performing monitoring based on a change-point time, Glugla teaches monitoring a quantity of fuel, and it would have been obvious to an ordinary artisan before the effective filing date to combine the two to arrive at a system that monitors fuel quantity based on a change-point time for the reasons given in the rejection.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to RYAN C VAUGHN whose telephone number is (571)272-4849. The examiner can normally be reached M-R 7:00a-5:00p ET.
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, Kamran Afshar, can be reached at 571-272-7796. 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.
/RYAN C VAUGHN/ Primary Examiner, Art Unit 2125
Prosecution Insights