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
Status of Claims
Claims 1-12 are rejected under 35 USC §101 Rejection.
Claims 1-12 are rejected under 35 USC §103 Rejection.
Claim Rejections - 35 USC §101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-12 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more as addressed below.
The new 2019 Revised Patent Subject Matter Eligibility Guidance published in the Federal Register (Vol. 84 No. 4, Jan 7, 2019 pp 50-57) has been applied and the claims are deemed as being patent ineligible.
The current 35 USC 101 analysis is based on the current guidance (Federal Register vol. 79, No. 241. pp. 74618-74633). The analysis follows several steps. Step 1 determines whether the claim belongs to a valid statutory class. Step 2A prong 1 identifies whether an abstract idea is claimed. Step 2A prong 2 determines whether an abstract idea is integrated into a practical application. If the abstract idea is integrated into a practical application the claim is patent eligible under 35 USC 101. Last, step 2B determines whether the claims contain something significantly more than the abstract idea. In most cases the existence of a practical application predicates the existence of an additional element that is significantly more.
Under Step 1 of the eligibility analysis, we determine whether the claims are to a statutory category by considering whether the claimed subject matter falls within the four statutory categories of patentable subject matter identified by 35 U.S.C.
101: Process, machine, manufacture, or composition of matter. The below claim is considered to be in a statutory category (process).
Under Step 1 of the analysis, claim 1 does belong to a statutory category, namely it is a process claim. Claim 6 is a machine claim, and claim 12 is a manufacture claim.
Under Step 2A Prong 1, the independent claim 1 includes abstract ideas as highlighted (using a bold font) below.
“1. A method for detecting an anomaly in sensor time series data of a sensing arrangement, via a processing unit, the method comprising:
- implementing a neural network trained on training data comprising prior sensor time series data;
- re-constructing a sample sensor time series data for a target time period using the trained neural network;
- determining an anomaly score variable based on a re-construction error in the re-constructed sample sensor time series data, corresponding to each of a plurality of time instants in the target time period;
- determining a confidence interval for the target time period based on a distribution of the determined anomaly score variable;
- mapping a target sensor time series data, generated by the sensing arrangement corresponding to the target time period, to the determined confidence interval; and
- indicating an anomaly in the target sensor time series data if the target sensor time series data is not substantially within the determined confidence interval,
wherein the anomaly score variable is determined based on a maximum absolute reconstruction error in the re-constructed sample sensor time series data.”
“6. A system comprising:
- a sensing arrangement integrated with a statistical process control of a semiconductor manufacturing process, the sensing arrangement configured to generate sensor time series data for the process;
- a neural network trained on training data comprising prior sensor time series data, the neural network configured to re-construct a sample sensor time series data for a target time period; and
- a processing unit configured to:
- determine an anomaly score variable based on a re-construction error in the re- constructed sample sensor time series data, corresponding to each of a plurality of time instants in the target time period;
- determine a confidence interval for the target time period based on a distribution of the determined anomaly score variable;
- map a target sensor time series data, generated by the sensing arrangement corresponding to the target time period, to the determined confidence interval; and
- indicate an anomaly in the target sensor time series data if the target sensor time series data is not substantially within the determined confidence interval,
wherein the processing unit is configured to determine the anomaly score variable based on a maximum absolute reconstruction error in the re-constructed sample sensor time series data.”
The highlighted steps indicated as abstract ideas are considered to be equivalent to mathematical steps and fundamental aspects of mathematics or directed to mental processes performed in the human mind (including observation, evaluation and opinion).
The steps of reconstructing, determining, determining, mapping, and indicating are limitations that are directed mathematical calculations on a data set.
In summary, the highlighted steps in the claims above therefore recite an abstract idea at Prong 1 of the 101 analysis.
Under step 2A prong 2,
Claim 1 does not comprise any particular field of use and claims are not directed to any particular practical application.
The limitations not in bold, namely “implementing a neural network training on training data comprising prior sensor time series data” and “using the trained neural network”, are additional elements. The steps of implementing and using a neural network are just using a generic computer processing component as a tool to implement the mathematical calculations, so the claim is not integrated into a particular practical application.
In claim 6 the limitation “a sensing arrangement integrated with a statistical process control of a semiconductor manufacturing process,” is just a general sensor arrangement recited at a broad level of generality, which is an insignificant extra-solution limitation related to the data gathering necessary to obtain the data for the mathematical calculations. It sets a broad field of use (semiconductor manufacturing), but does not identify a particular practical application.
The claims taken as a whole do not integrate the abstract idea into a particular practical application.
Under step 2B
Following the same analysis as given above at prong 2, the claims taken as a whole are not significantly more than the abstract idea.
Claims 1 and 6 are therefore rejected under 35 U.S.C. 101 as being patent ineligible.
Dependent claims 2, 3, 7, 8, and 11 merely extend the details of the abstract idea of mathematical concepts, adding more mathematical calculations or mental steps.
Claims 4 and 9 are directed to an autoencoder with encoder and decoder, which are just generic processing components of the neural network, and hence generic processing components which do not add significantly more to the abstract idea.
Claims 5 and 10 are just related to the data formatting steps, and are insignificant additional steps.
Claim 12 comprises a computer program product comprising computer executable program code stored on a non-statutory computer readable medium, which includes software running on the computer. The software running on the computer does not make the claims significantly more than the abstract idea. As recited in the MPEP, 2106.07(b), merely adding a generic computer components (processor and memory), or a programmed computer to perform generic computer functions does not automatically overcome an eligibility rejection. Alice Corp. Pty. Ltd. v. CLS Bank Int'l, 134S. Ct. 2347, 2359-60, 110 USPQ2d 1976, 1984 (2014). See also OIP Techs, v. Amazon.com, 788 F.3d 1359, 1364, 115 USPQ2d 1090, 1093-94. See MPEP 2106.05(f).
Thus, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception, and the claims are ineligible under 35 USC 101.
Claims 2-5 and 7-12 are therefore similarly rejected under 35 U.S.C. 101.
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.
Claims 1, 2, 4-7 and 9-12 are rejected under 35 U.S.C. 103 as being unpatentable over Tajima (EP3379360 A2), hereinafter Tajima in view of Fan “Defective wafer detection using a denoising autoencoder for semiconductor manufacturing processes”, hereinafter Fan.
Regarding Claim 1, Tajima discloses a method for detecting an anomaly (para 101, concerning an anomaly detection system for industrial systems) in sensor time series data of a sensing arrangement (para [103], [116], where anomaly detection system of the present embodiments may be such that the arithmetic device uses the predictive model and past operational data to perform structured prediction of future time-series data for a predetermined coming time period or an occurrence probability of the time-series data, and calculates the anomaly score based on an accumulated deviation), via a processing unit, the method comprising:
implementing a neural network (“autoencoder”, para [0019]) trained on training data comprising prior sensor time series data (Papara [104], where terminal 13 reads the detection result data 1D4 from the local data management unit 113 and presents detection results to the operator);
re-constructing a sample sensor time series data for a target time period using the trained neural network (Para [0104] and [126], where device uses reconstruction error for prediction error of the predictive model);
determining an anomaly score variable based on a re-construction error in the re- constructed sample sensor time series data (Para [0104] and [126], where anomaly detection system may be such that as the anomaly score, the arithmetic device uses reconstruction error for prediction error of the predictive model), corresponding to each of a plurality of time instants in the target time period (Para [0112], the predictive model and past operational data to perform structured prediction of future time-series data for a predetermined coming time period or an occurrence probability of the time-series data, and calculates the anomaly score based on an accumulated deviation of the operational data acquired from the device from results of the structural prediction);
determining a confidence interval (para [103], where threshold g is set to “h+2s” where h and s respectfully a mean and standard deviation of anomaly score) for the target time period based on a distribution of the determined anomaly score variable (para [126] and [129]);
mapping a target sensor time series data, generated by the sensing arrangement corresponding to the target time period, to the determined confidence interval (para [126], [129] and graphic 1G103 in Fig. 12, para [99], the encoder-decoder recurrent neural network, the detection unit 112 of the controller 11 consecutively inputs the operational data 1D1 approximately several tens to several hundreds of time units before the time point t to update the internal state of the recurrent neural network. Further, the detection unit 112 calculates a prediction error sequence by calculating a predicted value sequence within a window size (30) from the time point t and computing the absolute values of the differences between the predicted value sequence and the operational data 1D1); and
indicating an anomaly in the target sensor time series data if the target sensor time series data is not substantially within the determined confidence interval (para [20] When this anomaly score exceeds a predetermined threshold, para [0069], where (detection unit 112 of the controller 11 checks whether the anomaly score is below a threshold γ), e.g., anomaly score below or exceeds the threshold same as indicating an anomaly is not in confidence interval).
Further, Tajima discloses “calculates an anomaly score based on the sum of the absolute values of the … reconstruction errors” (see para [0100]).
Tajima does not disclose wherein the anomaly score variable is determined based on a maximum absolute reconstruction error in the re-constructed sample sensor time series data.
Fan discloses the anomaly score variable is determined based on a maximum absolute reconstruction error (Fig. 8, (a), (b) (para 4.4. where Maximum reconstruction error (MaxRE) is usually used as a threshold for anomaly detection. For the online monitoring scheme during production, if the result of the reconstruction error is greater than the threshold, then the wafer will be judged to be defective … called MaxREwoo, which reduces the variance of reconstruction errors by removing outliers from the training dataset).
Therefore, it would have been obvious to one of ordinary skill in the art at the time the applicants' invention was made to determine an anomaly score based on maximum reconstruction error, as taught by Fan into Tajima in order to provide higher sensitivity to outliers, better contrast between normal and anomalous samples, and improved robustness in semi‑supervised settings.
Regarding Claim 2, Tajima and Fan disclose the method according to claim 1, further Tajima discloses wherein re-constructing the sample sensor time series data using the neural network comprises predicting at least one variable and a corresponding timestamp for each of the plurality of time instants in the target time period (Fig. 5, Fig. 6, where 1D401, detection date and time and window size 1D404 para [0036], where predictive model parameters 1D302 indicate parameters of the predictive model that predicts the time-series behavior of the monitored facility 10. The window size estimation model parameters 1D303 indicate parameters of a window size estimation model that dynamically changes the window size for calculation of an anomaly score so that anomaly scores of normal operation data may stay approximately the same.).
Regarding Claim 4, Tajima and Fan disclose the method according to claim 1, further Tajima disclose wherein the neural network is an autoencoder comprising an encoder and a decoder (para [0092], where assumed that the above predictive model uses an encoder-decoder recurrent neural network), and
wherein the encoder is trained on the training data, and the decoder is implemented to re-construct the sample sensor time series data (para [0046], where predictive model is able to constitute an encoder-decoder recurrent neural network using long short-term memory (LSTM), like a predictive model 1N101 illustrated in Fig. 9. para [0092], [0093], where predictive model uses an encoder-decoder recurrent neural network … a simpler autoencoder may be used to predict (reconstruct) data in the same section).
Regarding Claim 5, Tajima and Fan disclose the method according to claim 1, further Tajima discloses wherein the sensing arrangement comprises a plurality of sensor devices, and wherein the sensor time series data comprises sensor parameters with timestamps for each of the plurality of sensor devices (Fig. 3, where date and time, item name, para [026], where As shown by Fig. 3, a next description relates to operational data 1D1 collected by each controller 11 from each facility 10 or the controller 11 itself and managed by the local data management unit 113, e.g., 1D101 is the timestamp. 1D102 is plurality of devices, 1D103 is sensor parameter (value) from a sensor attached in the facility 10 (see Fig. 1)).
Regarding Claim 6, Tajima discloses a system comprising:
a sensing arrangement integrated with a statistical process control of a semiconductor manufacturing process, the sensing arrangement (para [103], and [116], where anomaly detection system of the present embodiments may be such that the arithmetic device uses the predictive model and past operational data to perform structured prediction of future time-series data for a predetermined coming time period or an occurrence probability of the time-series data, and calculates the anomaly score based on an accumulated deviation), configured to generate sensor time series data for the process;
a neural network trained on training data (“autoencoder”, para [0019]) comprising prior sensor time series data, the neural network configured to re-construct a sample sensor time series data for a target time period (Para [0104] and [126], where device uses reconstruction error for prediction error of the predictive model); and
a processing unit (Fig.1 and 2, para [0023], where a collection unit 111, a detection unit 112, and a local data management unit 113) configured to:
determine an anomaly score variable based on a re-construction error in the re- constructed sample sensor time series data (Para [0104] and [126], where anomaly detection system may be such that as the anomaly score, the arithmetic device uses reconstruction error for prediction error of the predictive model), corresponding to each of a plurality of time instants in the target time period (Para [0112], the predictive model and past operational data to perform structured prediction of future time-series data for a predetermined coming time period or an occurrence probability of the time-series data, and calculates the anomaly score based on an accumulated deviation of the operational data acquired from the device from results of the structural prediction);
determine a confidence interval for the target time period (para [103], where threshold g is set to “m+ 2s” where s and m respectfully a mean and standard deviation of anomaly score) based on a distribution of the determined anomaly score variable (para [126] and [129]);
map a target sensor time series data, generated by the sensing arrangement corresponding to the target time period, to the determined confidence interval (para [126], [129] and graphic 1G103 in Fig. 12, para [99], the encoder-decoder recurrent neural network, the detection unit 112 of the controller 11 consecutively inputs the operational data 1D1 approximately several tens to several hundreds of time units before the time point t to update the internal state of the recurrent neural network. Further, the detection unit 112 calculates a prediction error sequence by calculating a predicted value sequence within a window size (30) from the time point t and computing the absolute values of the differences between the predicted value sequence and the operational data 1D1); and
indicate an anomaly in the target sensor time series data if the target sensor time series data is not substantially within the determined confidence interval (para [20] When this anomaly score exceeds a predetermined threshold, para [0069], where (detection unit 112 of the controller 11 checks whether the anomaly score is below a threshold γ), e.g., anomaly score below or exceeds the threshold same as indicating an anomaly is not in confidence interval).
Further, Tajima discloses “calculates an anomaly score based on the sum of the absolute values of the … reconstruction errors” (see para [0100]).
Tajima does not disclose wherein the anomaly score variable is determined based on a maximum absolute reconstruction error in the re-constructed sample sensor time series data.
Fan discloses the processing unit is configured to determine the anomaly score variable based on a maximum absolute reconstruction error in the re-constructed sample sensor time series data (Fig. 8, (a), (b) (para 4.4. where Maximum reconstruction error (MaxRE) is usually used as a threshold for anomaly detection. For the online monitoring scheme during production, if the result of the reconstruction error is greater than the threshold, then the wafer will be judged to be defective … called MaxREwoo, which reduces the variance of reconstruction errors by removing outliers from the training dataset).
Therefore, it would have been obvious to one of ordinary skill in the art at the time the applicants' invention was made to determine anomaly score based on maximum reconstruction error, as taught by Fan into Tajima in order to provide higher sensitivity to outliers, better contrast between normal and anomalous samples, and improved robustness in semi supervised settings.
Regarding Claim 7, Tajima and Fan disclose the system according to claim 6, wherein the neural network is configured to predict at least one variable and a corresponding timestamp for each of the plurality of time instants in the target time period, to re-construct the sample sensor time series data (Fig. 5, Fig. 6, where 1D401, detection date and time and window size 1D404 para [0036], where predictive model parameters 1D302 indicate parameters of the predictive model that predicts (reconstructs) the time-series behavior of the monitored facility 10. The window size estimation model parameters 1D303 indicate parameters of a window size estimation model that dynamically changes the window size for calculation of an anomaly score so that anomaly scores of normal operation data may stay approximately the same.).
Regarding Claim 9, Tajima and Fan disclose the system according to claim 6, further Tajima discloses wherein the neural network is an autoencoder comprising an encoder and a decoder (para [0092], where assumed that the above predictive model uses an encoder-decoder recurrent neural network), and wherein the encoder is trained on the training data and the decoder is implemented to re-construct the sample sensor time series data (para [0046], where predictive model is able to constitute an encoder-decoder recurrent neural network using long short-term memory (LSTM), like a predictive model 1N101 illustrated in Fig. 9. para [0092], [0093], where predictive model uses an encoder-decoder recurrent neural network … a simpler autoencoder may be used to predict (reconstruct) data in the same section).
Regarding Claim 10, Tajima and Fan disclose the system according to claim 6, further Tajima discloses wherein the sensing arrangement comprises a plurality of sensor devices, and wherein the sensor time series data comprises sensor parameters with timestamps for each of the plurality of sensor devices (Fig. 3, where date and time, item name, para [026], where As shown by Fig. 3, a next description relates to operational data 1D1 collected by each controller 11 from each facility 10 or the controller 11 itself and managed by the local data management unit 113, e.g., 1D101 is the timestamp. 1D102 is plurality of devices, 1D103 is sensor parameter (value) from a sensor attached in the facility 10 (see Fig. 1)).
Regarding Claim 11, Tajima and Fan disclose the system according to claim 6, further Tajima discloses wherein the indicated anomaly in the target sensor time series data is used to control the process (para [21], where anomaly detection system 1 of the present embodiment is assumed to include facilities 10 (monitored systems) having sensors and actuators, controllers 11 that control the facilities 10, a server 12 that performs learning of the above-described predictive model and management of data, a terminal 13 that presents information indicating an anomaly or a sign thereof to an operator).
Regarding Claim 12, Tajima and Fan disclose a computer program product, further Tajima discloses comprising computer executable program code stored on a non-transitory computer readable medium, which when executed by a processing unit causes a system to perform the method of any one of claims 1 (Fig. 1 and 2, para [0023], where a central processing unit (CPU) 1H101 loads a program stored in a read-only memory (ROM) 1H102 or an external storage device 1H104 into a read-access memory (RAM) 1H103 and executes the program to control a communication interface (IF) 1H105).
Claims 3 and 8 are rejected under 35 U.S.C. 103 as being unpatentable over Tajima in view of Fan, as applied above and further in view of Anaissi “Multi-Objective Variational Autoencoder: an Application for Smart Infrastructure Maintenance”, hereinafter Anaissi.
Regarding Claim 3, Tajima and Fan disclose the method according to claim 1, but do not disclose wherein the confidence interval for the trained neural network is determined based on a mean of the determined anomaly score variable and a standard deviation of the determined anomaly score variable.
Anaissi discloses a confidence interval for a trained neural network is determined based on a mean of the determined anomaly score variable and a standard deviation of the determined anomaly score variable (see pages 39:7 – 39:8, and recall Tajima para [103], where threshold g is set to “h+2s” where h and s respectfully a mean and standard deviation of anomaly score).
Therefore, it would have been obvious to one of ordinary skill in the art at the time the applicants' invention was made to confidence interval for the trained neural network is determined based on a mean of the determined anomaly score variable and a standard deviation, as taught by Anaissi in combination of Tajima and Fan in order to provide higher prediction uncertainty, improving interpretability, reliability, and decision-making.
Regarding Claim 8, Tajima and Fan disclose the system according to claim 6, but do not disclose wherein the processing unit is configured to determine the confidence interval for the trained neural network based on a mean of the determined anomaly score variable and a standard deviation of the determined anomaly score variable.
Anaissi discloses the confidence interval for the trained neural network is determined based on a mean of the determined anomaly score variable and a standard deviation of the determined anomaly score variable (see pages 39:7 – 39:8, and recall Tajima para [103], where threshold g is set to “h+2s” where h and s respectfully a mean and standard deviation of anomaly score).
Therefore, it would have been obvious to one of ordinary skill in the art at the time the applicants' invention was made to confidence interval for the trained neural network is determined based on a mean of the determined anomaly score variable and a standard deviation, as taught by Anaissi in combination of Tajima and Fan in order to provide higher prediction uncertainty, improving interpretability, adaptability, reliability, and decision-making.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to KALERIA KNOX whose telephone number is (571)270-5971. The examiner can normally be reached M-F 8am-5pm.
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, Andrew Schechter can be reached at (571)2722302. 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.
/KALERIA KNOX/
Examiner, Art Unit 2857
/ANDREW SCHECHTER/Supervisory Patent Examiner, Art Unit 2857
Prosecution Insights