DETAILED ACTION
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA . This action is responsive to pending claims 1-20 filed 1/2/2026.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 1-5, 7-16, 18-20 are rejected under 35 U.S.C. 103 as being unpatentable over Tang ("Joint modeling of local and global temporal dynamics for multivariate time series forecasting with missing values", published 2020) in view of Lee (US 20200172989 A1).
For claim 1, Tang discloses: a computer-implemented method (fig.2 gives overview of computer-implemented training process) comprising:
identifying a first set of features of training data and a second set of features of the training data (§Introduction ¶1-2 contemplates application to multivariate time series (MTS) data, with §Problem Formulation ¶1 contemplating d-dimensional data, hence, the data having various feature sets);
training a deep learning model using the training data (fig.2 gives training process overview),
wherein training the deep learning model comprises training a first function to determine a relationship between the first set of features and the second set of features (§Modeling Global Dynamics (p.5959) eq.6-8 contemplates weighted correlations of input variables (eq.6: x, z, z’) with each other in order to access global patterns, hence, the memory module keyed by local statistics and local variables constitutes a set of learned cross-feature relations, see also p.5959 col.2 ¶4: “Besides, since variables at the same time interval interact with each other in Equation 6, ai also preserves inner correlations of variables at the same time interval), and training a second function to determine a relationship between missing data of a first period of time and complete data of a second period of time that follows the first period of time (fig.2: local statistics in a both forward and backward temporal direction and global dynamics are fed into an LSTM in order to generate predictions, hence, the LSTM constitutes a trained function for determining relationship between missing data of a first period of time (such as denoted via mask vectors, see §Problem Formulation ¶1) and complete data of a second period of time (such as the non-missing data); furthermore, memory modules for recognizing global dynamics constitutes learning a relationship between a first period of time (possibly including missing data) and a second period of time, such as learned via §Modeling Global Dynamics eq.6);
generating imputation time series data and forecasted time series data using the trained deep learning model (fig.2: imputation data is generated by the LSTM, see p.5959 eq.5, ¶3: providing estimates for missing variables; forecast data is generated, see p.5959 col.2 ¶4: “The forecasting results … are generated after the n-th iteration”),
wherein the imputation time series data is generated based on the trained deep learning model performing an imputation task on input data (ibid: p.5959 eq.5 shows learned parameters U, b for performing imputation task),
wherein the forecasted time series data is generated based on the trained deep learning model performing a forecasting task on the input data (ibid: likewise, the forecasting data is generated via a forecasting task), and
wherein the trained deep learning model is a multi-task model that simultaneously performs the imputation task and the forecasting task on the input data (…
Tang fig.2 gives overview of the process of imputing data via local variables, with p.5959 eq.5 disclosing the LSTM naturally providing imputations for missing values during the local statistics phase. Although Tang’s LSTM (fig.2) appears iterative rather than simultaneous, one of ordinary skill in the art would understand Tang as a simultaneous impute and forecast model for 2 reasons:
(1) In Tang, the imputation step is performed in order to generate values for the future forecasting step which occurs later in the iterations of the LSTM network (see fig.2, top). A LSTM is a type of iterative or recurrent neural network. According to Applicant’s Specs 0049, 0055-56, simultaneous imputation and forecasting is described as a second function that would determine Zshift (see fig.1C) comprising both the imputed current time window and the forecasting task, hence, providing simultaneous impute and forecast in one pass. However, 0055 discloses the use of an LSTM (as well as a CNN) to determine Zshift, which would be similar to how Tang performs impute and forecast. Hence, the Specs / BRI understanding of “simultaneous” impute and forecast would include the type of iterative impute and forecast via LSTM of Tang.
(2) Informational reference Zuo ("Graph convolutional networks for traffic forecasting with missing values", published 12/13/2022) is cited as reflecting the art’s understanding of simultaneous imputation and forecasting, in particular, p.3 ¶1 disclosing the “joint modeling of Spatio-temporal patterns and missing values in a one-step process”, §2 def.1 and def. 2 conceiving of the difference between “one-step processing”, where impute and forecast are “jointly modeled in one single step” (hence, simultaneously), in contrast with two-step processing, where impute is handled in a preprocessing step. Hence, a POSITA’s understanding of joint or simultaneous technique would include those of Tang and Zuo, which perform impute and forecast via a single model, even if the model is iterative in nature.
In summary, further clarification is needed, as a POSITA would understand Tang to be simultaneous, e.g., clarifying that a CNN is used, or that impute and forecast are performed non-iteratively, etc.
In the interests of compact prosecution, attention is directed to Liu ("Naomi: Non-autoregressive multiresolution sequence imputation", published 2019), which describes a non-autoregressive approach to impute and forecast, in particular, §4.4 “Forward Prediction” where a forecasting is performed in a single step by “masking” future values. See the “Response to Arguments” section for a potential mapping of Liu.
… ).
Tang does not disclose: controlling an operation of a system to control one or more products manufactured by the system;
wherein the operation is controlled based on the imputation time series data and the forecasted time series data.
Lee discloses: controlling an operation of a system to control one or more products manufactured by the system; wherein the operation is controlled based on the imputation time series data and the forecasted time series data (Abstract, fig.2: imputing and forecasting hot metal temperature (HMT) data for triggering control actions in a manufacturing process).
It would have been obvious before the effective filing date to a person of ordinary skill in the art to modify the method of Tang by incorporating the industrial control integration of Lee. Both concern the art of imputation and prediction systems for time series data via machine learning, and the incorporation would have, according to Lee, allowed improved quality and better control of manufacturing processes where data availability is irregular (0002-4).
For claim 2, Tang modified by Lee discloses the method of claim 1, as described above. Tang modified by Lee further discloses: obtaining the input data from a data server system (Tang §Experiments: Datasets (p.5960) contemplates obtaining data such as Beijing Air, PhysioNet, etc. datasets from the webservers of the footnotes; Lee fig.2:204, 0028);
generating control information to control the operation of the system associated with the data server system (Lee fig.2:218-220, 0038-39),
wherein the control information is generated using the imputation time series data and the forecasted time series data (Lee fig.2,:214-216, 0035-37); and
controlling the operation of the system using the control information (Lee fig. 2:220, 0039).
For claim 3, Tang modified by Lee discloses the method of claim 1, as described above. Tang further discloses: wherein the input data is provided with missing values (§Problem Formulation ¶1 (p.5957)),
wherein performing the imputation task on the input data to obtain the imputation time series data comprises:
performing the imputation task on the input data to obtain the missing values for a first period of time (fig.2, p.5959 eq.5: estimates of missing values are provided by the LSTM), and
wherein performing the forecasting task on the input data to obtain the forecasted time series data comprises:
performing the forecasting task on the input data to forecast time series data for a second period of time that follows the first period of time (fig.2, p.5959 col.2 ¶4: future time periods are performed via forecasting).
For claim 4, Tang modified by Lee discloses the method of claim 3, as described above. Tang further discloses: providing the input data to cause the trained deep learning model to simultaneously perform the imputation task and the forecasting task (fig.2, p.5959 eq.5, col2 ¶4: the LSTM simultaneously provides imputation and forecasting, and furthermore provides imputed values during the forecasting phase; see also §Capturing Local Statistics, fig. 3 contemplating various simpler forms of imputation including Empirical Mean, Last Observation performed during forecasting phase ).
For claim 5, Tang modified by Lee discloses the method of claim 1, as described above. Tang further discloses: wherein training the first function comprises:
using a loss function to minimize a difference between an actual output of the first function and an expected output of the first function (eq.6-8, p.5959 col.2 contemplates generating correlations between various features based on trainable parameters Wq, Bq, see also p.5960 col.2 ¶1: using stochastic gradient descent to update model parameters),
wherein the expected output includes the second set of features (As the memory module is trained jointly with the overall LSTM (see p.5959: §Modeling Global Dynamics ¶2), the loss function includes the true values of the MTS data, which includes the second set of features).
For claim 7, Tang modified by Lee discloses the method of claim 1, as described above. Tang further discloses: wherein the deep learning model is a first deep learning model (fig.2 shows overall deep learning model),
wherein the first function is a second deep learning model trained to determine missing values from multivariate time series data provided to the second deep learning model (fig.2: memory module with trainable parameters, lookup vectors, etc. for determining MTS data via global patterns), and
wherein the second function is a third deep learning model trained to forecast multivariate time series data from the multivariate time series data provided to the second deep learning model (fig.2: LSTM module for forecasting based on the MTS data from the memory module).
For claim 8, Tang discloses: program instructions to train a machine learning model using training data (fig.2 gives training process overview),
wherein the training data includes a first set of features and a second set of features (§Introduction ¶1-2 contemplates application to multivariate time series (MTS) data, with §Problem Formulation ¶1 contemplating d-dimensional data, hence, the data having various feature sets), and
program instructions to train a first function to determine a relationship between the first set of features and the second set of features (§Modeling Global Dynamics (p.5959) eq.6-8 contemplates weighted correlations of input variables (eq.6: x, z, z’) with each other in order to access global patterns, hence, the memory module keyed by local statistics and local variables constitutes a set of learned cross-feature relations, see also p.5959 col.2 ¶4: “Besides, since variables at the same time interval interact with each other in Equation 6, ai also preserves inner correlations of variables at the same time interval), and
program instructions to train a second function to determine a relationship between missing data of a first period of time and complete data of a second period of time that follows the first period of time (fig.2: local statistics in a both forward and backward temporal direction and global dynamics are fed into an LSTM in order to generate predictions, hence, the LSTM constitutes a trained function for determining relationship between missing data of a first period of time (such as denoted via mask vectors, see §Problem Formulation ¶1) and complete data of a second period of time (such as the non-missing data); furthermore, memory modules for recognizing global dynamics constitutes learning a relationship between a first period of time (possibly including missing data) and a second period of time, such as learned via §Modeling Global Dynamics eq.6);
program instructions to generate imputation time series data and forecasted time series data using the trained machine learning model (fig.2: imputation data is generated by the LSTM, see p.5959 eq.5, ¶3: providing estimates for missing variables; forecast data is generated, see p.5959 col.2 ¶4: “The forecasting results … are generated after the n-th iteration”),
wherein the imputation time series data is generated based on the trained deep learning model performing an imputation task on input data (ibid: p.5959 eq.5 shows learned parameters U, b for performing imputation task), and
wherein the forecasted time series data is generated based on the trained deep learning model performing a forecasting task on the input data (ibid: likewise, the forecasting data is generated via a forecasting task).
Tang does not disclose: a computer program product comprising: one or more computer readable storage media, and program instructions collectively stored on the one or more computer readable storage media, the program instructions comprising the above steps; wherein the program instructions to train the machine learning model comprise the above steps; program instructions to control using the imputation time series data and the forecasted time series data an operation of one or more devices to control manufacture of one or more products.
Lee: a computer program product comprising: one or more computer readable storage media, and program instructions collectively stored on the one or more computer readable storage media, the program instructions comprising the above steps (fig.13, in particular 0067-69); wherein the program instructions to train the machine learning model comprise the above steps (ibid); program instructions to control using the imputation time series data and the forecasted time series data an operation of one or more devices to control manufacture of one or more products (Abstract, fig.2: imputing and forecasting hot metal temperature (HMT) data for triggering control actions in a manufacturing process).
It would have been obvious before the effective filing date to a person of ordinary skill in the art to modify the method of Tang by incorporating the industrial control integration of Lee. Both concern the art of imputation and prediction systems for time series data via machine learning, and the incorporation would have, according to Lee, allowed improved quality and better control of manufacturing processes where data availability is irregular (0002-4).
For claim 9, Tang modified by Lee discloses the product of claim 8, as described above. Tang further discloses: wherein the machine learning model is a first deep learning model (fig.2 shows overall deep learning model),
wherein the first function is a second deep learning model trained to determine missing values from multivariate time series data provided to the second deep learning model (fig.2: memory module with trainable parameters, lookup vectors, etc. for determining MTS data via global patterns), and
wherein the second function is a third deep learning model trained to forecast multivariate time series data from the multivariate time series data provided to the second deep learning model (fig.2: LSTM module for forecasting based on the MTS data from the memory module).
For claim 10, Tang modified by Lee discloses the product of claim 9, as described above. Tang further discloses: wherein the program instructions to train the first function comprise:
program instructions to use a loss function to minimize a difference between an actual output of the first function and an expected output of the first function (eq.6-8, p.5959 col.2 contemplates generating correlations between various features based on trainable parameters Wq, Bq, see also p.5960 col.2 ¶1: using stochastic gradient descent to update model parameters),
wherein the actual output includes the second set of features determined using the machine learning model (As the memory module is trained jointly with the overall LSTM (see p.5959: §Modeling Global Dynamics ¶2), the loss function includes the predicted values of the MTS data, which includes the second set of features), and
wherein the expected output includes the second set of features (As the memory module is trained jointly with the overall LSTM (see p.5959: §Modeling Global Dynamics ¶2), the loss function includes the true values of the MTS data, which includes the second set of features).
For claim 11, Tang modified by Lee discloses the product of claim 8, as described above. Tang further discloses: wherein the program instructions to train the second function comprise:
program instructions to provide masks, in the training data, to indicate time series data that is missing (§Problem Formulation ¶1 (p.5957)); and
program instructions to train the second function to determine the time series data based on the masks indicating that the time series data that is missing (fig.3 (p.5959) various data is concatenated (x, z, z’) to feed into LSTM predictor for determining the time series data, z being local statistic data (forward and backwards) calculated based on the masked values, see §Capturing Local Statistics, p.5958 last ¶).
For claim 12, Tang modified Lee discloses the product of claim 11, as described above. Tang further discloses: wherein the program instructions to train the second function comprise:
program instructions to use a loss function to minimize a difference between an actual output of the second function and an expected output of the second function (p.5959 eq.9 gives loss function including difference between actual and expected output component),
wherein the actual output includes the determined time series data (ibid), and wherein the expected output includes the time series data that is missing (ibid: dot product with the masked component).
For claim 13, Tang modified by Lee discloses the product of claim 8, as described above. Tang modified by Lee further discloses: wherein the first set of features include quantitative features of multivariate time series data (Tang §Experiment: Datasets (p.5960) contemplates quantitative features; Lee fig.2, 0032 HMT, temperature senso readings);
wherein the second set of features includes qualitative features (Lee 0026, 0028, 0039 contemplates incorporation of product quality measurements or features, including temperature and pressure profile, material speed, gas permeability, mineral content, hence raw material charge rate, descent speed, etc. readings comprise material or blast furnace state quality features).
For claim 14, Tang modified by Lee discloses the product of claim 8, as described above. Tang further discloses: wherein the first function is a first deep learning model (fig.2, §Memory Module contemplates an attention-based memory module with trainable parameters, hence, a deep learning model),
wherein the second function is a second deep learning model (fig.2: LSTM is a deep learning model §MTS forecasting with LSTM (p.5958)), and
wherein the program instructions to train the second function comprises:
program instructions to train the second deep learning model to simultaneously perform the imputation tasks and the forecasting tasks (p.2: shows simultaneous performing of imputation tasks and forecasting tasks).
Claim 15 recites a system analogous to the media of claim 8 and is hence rejected for the same reasons.
For claim 16, Tang modified by Lee discloses the system of claim 15, as described above. Tang further discloses: wherein the input data is provided with missing values (§Problem Formulation ¶1 (p.5957), and
wherein, to perform the imputation task on the input data to obtain the imputation time series data, the one or more devices are configured to:
perform the imputation task on the input data to obtain the missing values (missing values are imputed in various ways, including via local statistics (§Capturing local statistics, p.5958), via the LSTM (p.5959 col.1 ¶2), and via the global memory module (p.5959 col.2 ¶4)).
For claim 18, Tang modified by Lee discloses the system of claim 15, as described above. Tang further discloses: wherein the first function is a first deep learning model trained to determine missing values from input provided to the first deep learning model (p.5959 col.2 ¶4 contemplates determining missing values via correlation of variables based on input, the memory module being a attention-based module with tunable parameters, hence, a deep learning model), and
wherein the second function is a second deep learning model trained to forecast time series data from the input provided to the first deep learning model (fig.2: The LSTM further transforms the provided input to forecast time series data, see §MTS forecasting with LSTM (p.5958)).
For claim 19, Tang modified by Lee discloses the system of claim 15, as described above. Tang further discloses: wherein, to train the first function, the one or more devices are configured to:
provide a parameter indicating an initial relationship between the first features and the second features (p.5959 col.2 eq.6: Memory module with tunable parameters is initialized);
train the first function to determine the second features based on the parameter (ibid: the features are determined via correlation and memory; the parameters being tuned); and
use a loss function to minimize a difference between an actual output of the first function and an expected output of the first function (p.5959 col.2 eq.9) ,
wherein the actual output includes the determined second features (ibid), and
wherein the expected output includes the second features (ibid).
For claim 20, Tang modified by Lee discloses the system of claim 15, as described above. Tang further discloses: wherein, to train the first function, the one or more devices are configured to:
use a loss function to minimize a difference between an actual output of the second function and an expected output of the second function (p.5959 eq.19 contemplates loss function).
Claim(s) 6 are rejected under 35 U.S.C. 103 as being unpatentable over Tang.joint ("Joint modeling of local and global temporal dynamics for multivariate time series forecasting with missing values", published 2020) in view of Lee (US 20200172989 A1) in view of Razavi-Far ("An integrated imputation-prediction scheme for prognostics of battery data with missing observations", published 2019).
For claim 6, Tang discloses the method of claim 1, as described above. Tang further discloses: wherein training the second function comprises:
using a loss function to minimize a difference between an actual output of the second function and an expected output of the second function (p.5959 col. eq.5 contemplates the LSTM output with trainable parameters U, b, trained via stochastic gradient descent, see p.5960 col.2 ¶1, with eq.9 disclosing the output function);
Tang does not disclose: wherein the expected output includes outputs of the imputation tasks and the forecasting tasks, because eq.9 (p.5959) multiplies the difference term by the masking function therefore ignoring values that were absent in the original data in the loss function (even though imputation has filled in these values).
Razavi-Far discloses: wherein the expected output includes outputs of the imputation tasks and the forecasting tasks (Razavi-Far discloses that it would be a good idea to use imputed values during training (p.710 col.2 last ¶-p.711 col.1), such as the various strategies outlined in fig.1:preprocessing module, hence, combination with Tang yielding a technique where imputed values would be used as part of the loss function during training).
It would have been obvious before the effective filing date to a person of ordinary skill in the art to modify the method of Tang by incorporating the training technique of Razavi-Far. Both concern the art of imputation and prediction for time series data, and the incorporation would have, according to Razavi-Far, allow the use of multiple prediction techniques and used for training, such as to improve performance (p.710 col.2 last ¶-p.711 col.1).
Claim(s) 17 are rejected under 35 U.S.C. 103 as being unpatentable over Tang.joint ("Joint modeling of local and global temporal dynamics for multivariate time series forecasting with missing values", published 2020) in view of Razavi-Far ("An integrated imputation-prediction scheme for prognostics of battery data with missing observations", published 2019) in view of Li.874 (US 20240069874 A1).
For claim 17, Tang modified by Lee discloses the system of claim 16, as described above. Tang modified by Lee further discloses: wherein the one or more devices are further configured to:
obtain the input data from a data server system associated with the system (Tang §Datasets (p.5960) contemplates data obtained from web servers in footnote).
Tang modified by Lee does not disclose: provide the missing values to the data server system.
Li discloses: provide the missing values to the data server system (Li 0039, 0053 contemplates incorporating a remote-job for an imputor, hence, Li contemplates running imputor services remotely to providing imputing results to a data server system).
It would have been obvious before the effective filing date to a person of ordinary skill in the art to modify the product of Tang modified by Razavi-Far by incorporating the remote imputor technique of Li. Both concern the art of imputing for machine learning, and the incorporation would have, according to Li, be better able to handle large imputing tasks (0053).
Response to Arguments
Applicant’s arguments have been fully considered. The 101 rejections are withdrawn. However, the 103 rejections are maintained for the reason given in claim 1.
In the interests of compact prosecution, an alternative mapping of claim 1 via Liu ("Naomi: Non-autoregressive multiresolution sequence imputation", published 2019) is provided below:
For claim 1, Liu discloses: a computer-implemented method (§3 ¶1-2 gives overview of the imputation and forecasting process including contrast with auto-regressive approaches) comprising:
identifying a first set of features of training data and a second set of features of the training data (§3 ¶1 gives overview of multivariate time series input (D-dimension vector time slices x) features alongside masked values, with §4.1 contemplating application to multivariate traffic data);
training a deep learning model using the training data (§3.1, fig.2 gives overview of deep learning architecture, p.4 Alg. 1 gives training overview),
wherein training the deep learning model comprises training a first function to determine a relationship between the first set of features and the second set of features (§3.1, fig.2 contemplates the generation of hidden state which generate output variables, these hidden states encoding the relationship between the various dimensions of x ), and training a second function to determine a relationship between missing data of a first period of time and complete data of a second period of time that follows the first period of time (ibid: the hidden states, such as when applied to forward prediction (§4.4) constitutes relation between first period including missing or masked values and complete, predicted second future data);
generating imputation time series data and forecasted time series data using the trained deep learning model (§4.4 ¶1 contemplates extension to forward prediction by including future masked values alongside missing values for imputation),
wherein the imputation time series data is generated based on the trained deep learning model performing an imputation task on input data (§4.4 generating inferences based on training of p.4 Alg.1),
wherein the forecasted time series data is generated based on the trained deep learning model performing a forecasting task on the input data (ibid), and
wherein the trained deep learning model is a multi-task model that simultaneously performs the imputation task and the forecasting task on the input data (§3 ¶1-2 discloses simultaneous or non-autoregressive imputation and forecasting, such as via algorithm of §4.4, Alg.1).
Liu does not disclose: controlling an operation of a system to control one or more products manufactured by the system;
wherein the operation is controlled based on the imputation time series data and the forecasted time series data.
Lee discloses: controlling an operation of a system to control one or more products manufactured by the system; wherein the operation is controlled based on the imputation time series data and the forecasted time series data (Abstract, fig.2: imputing and forecasting hot metal temperature (HMT) data for triggering control actions in a manufacturing process).
It would have been obvious before the effective filing date to a person of ordinary skill in the art to modify the method of Liu by incorporating the industrial control integration of Lee. Both concern the art of imputation and prediction systems for time series data via machine learning, and the incorporation would have, according to Lee, allowed improved quality and better control of manufacturing processes where data availability is irregular (0002-4).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. Liu ("Naomi: Non-autoregressive multiresolution sequence imputation", published 2019), see “Response to Arguments” above.
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 extension fee 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 date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to LIANG LI whose telephone number is (303)297-4263. The examiner can normally be reached Mon-Fri 9-12p, 3-11p MT (11-2p, 5-1a 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. The examiner is available for interviews Mon-Fri 6-11a, 2-7p MT (8-1p, 4-9p ET).
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor Jennifer Welch can be reached on (571)272-7212. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from Patent Center and the Private Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from Patent Center or Private PAIR. Status information for unpublished applications is available through Patent Center or Private PAIR to authorized users only. Should you have questions about access to Patent Center or the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free).
/LIANG LI/
Primary examiner AU 2143
Prosecution Insights