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 .
Claim Rejections - 35 USC § 102
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
Claims 1-14 are rejected under 35 U.S.C. 102(a)(1) as being taught by Yamashita et al. (JP7059162B2, herein after referred to as Yamashita).
Regarding Claim 1, Yamashita Teaches An information processing apparatus comprising a processor including a hardware ("The analyzer disclosed in the present application is an analyzer that has a processor that executes a program"; See Para. 21), configured to: calculate a latent representation (the feature amount space; See Para. 21) representing a feature amount (feature amount: See Para. 21) regarding a prediction target event (prediction target; See Para. 21) from processing target data including the feature amount ("the first feature amount data including the feature amount"; See Para. 21); inputting the latent representation (the feature amount space; See Para. 21) to monotonic neural networks ("neural network"; See Para. 23; The neural networks count as monotonic as the functions are ran over an increasing time period.) that are modeled to output a scalar value (“Even in the prediction of the time when a failure or cancellation occurs (hereinafter, all concepts similar to survival time like these are called survival time), if the factors leading to the failure or cancellation can be output, the prevention of machine failure and service Appropriate marketing can be done. The most important quantity in predicting survival time, including censoring, is a function of features called the hazard function. The hazard function is defined at each time, and the hazard function at time T represents the probability of death (or failure, cancellation) at time T. The integration of the hazard functions up to time T (cumulative hazard function) is the probability of dying by time T, and the point where the cumulative hazard function exceeds a certain threshold is considered to be the death time. Also, if the cumulative hazard function does not exceed the threshold at any time, it is considered alive (censored). Therefore, the prediction of the hazard function is equivalent to the prediction of the survival time, and in the prediction of the survival time, most of the analytical models that predict the hazard function that is easy to interpret are the targets of the prediction.”; See Para. 10-11; The machine learning model outputs the data’s timeline (Just called time). The neural network outputs “factors” of failure including time, which time and survival time are scalar values as they don’t have a specified direction. The time is then used in a monotonically increasing function which increases the time.) in accordance with a monotonically increasing function (“The hazard function is defined at each time…”; See Para. 11) defined by the latent representation (Feature Space) and a clock time (“Training data including the first feature amount data including the feature amount, the objective variable indicating the analysis time from the start of the analysis to the end of the analysis of the learning target”; See Para. 21) and obtaining the scalar value (Time) from the monotonic neural networks; and estimate at least one of a hazard function (hazard function; See Para. 21) and a survival function ("The most important quantity in predicting survival time, including censoring, is a function of features called the hazard function."; See Para. 11; The cumulative Hazard function comprised of the hazard functions output is put into the Survival function which compares the value against a "Censor threshold" that determines if they are alive or not) on the basis of the scalar value (Time) output obtained from the monotonic neural networks.
Regarding Claim 2, most of the limitations of this claim have been noted in the rejection of claim 1 (and thus the rejections of claim 1 are incorporated). Yamashita teaches wherein the processor (Processor) is further configured to learn parameters (The learning parameter; See Para. 39) of a calculation of the latent representation (Feature space) and the monotonic neural networks, through meta learning (Machine Learning; See Para. 5) and output learned parameters ("The learning parameter 265 is output data from the learning unit 261 and includes the learning parameters {Wh, W, w} described later."; See Para. 39).
Regarding Claim 3, most of the limitations of this claim have been noted in the rejection of claims 1 and 2 (and thus the rejections of claims 1 and 2 are incorporated). Yamashita teaches wherein the processor (Processor) is further configured to update the learned parameters on the basis of a plurality of pieces of prediction data including the feature amount regarding the prediction target event. ("Machine learning is one of the techniques for realizing AI (Artificial Intelligence). In machine learning, learning is called learning so that the error between the predicted value obtained from the feature vector and the actual value (true value) is minimized (weight vector in Perceptron, etc.). When the learning process is completed, new predicted values are calculated from the data not used for training (hereinafter referred to as test data). In the perceptron, the magnitude of each element value of the weight vector is used as the importance of the factors that contributed to the prediction."; See Para. 5)
Regarding claim 4, most of the limitations of this claim have been noted in the rejection of claims 1, 2 and 3 (and thus the rejections of claims 1, 2, and 3 are incorporated). Yamashita teaches wherein the function estimation unit (Predicting unit; See Fig. 2b; Yamashita teaches units are described as functions or physical pieces of hardware related to those functions, and each function can be a unit or multiple. (“the analyzer 220 may be configured by a plurality of units for each function”; See Para. 40)) includes a function calculation unit that calculates a cumulative hazard function (“The hazard function is defined at each time, and the hazard function at time T represents the probability of death (or failure, cancellation) at time T. The integration of the hazard functions up to time T (cumulative hazard function) is the probability of dying by time T, and the point where the cumulative hazard function exceeds a certain threshold is considered to be the death time. Also, if the cumulative hazard function does not exceed the threshold at any time, it is considered alive (censored). Therefore, the prediction of the hazard function is equivalent to the prediction of the survival time, and in the prediction of the survival time, most of the analytical models that predict the hazard function that is easy to interpret are the targets of the prediction.”; See Para. 11) on the basis of the scalar value (Time; See Para. 10-11) output from the monotonic neural networks, and an automatic differentiation unit that calculates the hazard function (Hazard Function; See Para. 11; All Prediction functions can be or are units (“the analyzer 220 may be configured by a plurality of units for each function”; See Para. 40)) by automatically differentiating the cumulative hazard function calculated by the function calculation unit (Cumulative Hazard function; See Para. 11; All predication functions can be or are units (“the analyzer 220 may be configured by a plurality of units for each function”; See Para. 40), They are using the hazard function outputs to calculate the Cumulative Hazard).
Regarding Claim 5, most of the limitations of this claim have been noted in the rejection of claims 1-4 (and thus the rejections of claims 1-4 are incorporated). Yamashita teaches wherein the processor (Processor) is further configured to convert the cumulative hazard function (Cumulative Hazard Function) into the survival function ("The most important quantity in predicting survival time, including censoring, is a function of features called the hazard function."; See Para. 11; The Hazard functions output is put into the Survival function which compares the value against a "Censor threshold" that determines if they are alive or not).
Regarding claim 6, most of the limitations of this claim have been noted in the rejection of claims 1, 2 and 3 (and thus the rejections of claims 1, 2, and 3 are incorporated). Yamashita teaches wherein, for estimating the hazard function (Hazard Function) and the survival function (Survival function), the processor (Processor) is configured to: calculate the survival function on the basis of the scalar value (Time) output obtained from the monotonic neural networks; and calculate the hazard function by automatically differentiating the survival function ("Also, if the cumulative hazard function does not exceed the threshold at any time, it is considered alive (censored). Therefore, the prediction of the hazard function is equivalent to the prediction of the survival time, and in the prediction of the survival time, most of the analytical models that predict the hazard function that is easy to interpret are the targets of the prediction."; See Para. 11).
Regarding Claim 7, most of the limitations of this claim have been noted in the rejection of claim 1 (and thus the rejections of claim 1 are incorporated). Yamashita teaches an information processing method executed by a processor (Processor; See Para. 21) of an information processing apparatus, comprising: calculating a latent representation (Feature space) representing a feature amount (Feature amount) regarding a prediction target event (Prediction Target (event or function); see Para. 20) from processing target data (Test Data; See Para. 5) including the feature amount (Feature Amount); inputting the latent representation (Feature space) to monotonic neural networks that is modeled to output a scalar value (Survival Time) in accordance with a monotonically increasing function (Increasing time) defined by the latent representation (Feature space) and a clock time (“Training data including the first feature amount data including the feature amount, the objective variable indicating the analysis time from the start of the analysis to the end of the analysis of the learning target”; See Para. 21) and obtaining the scalar value (time) from the monotonic neural networks; and estimating at least one of a hazard function and a survival function ("The most important quantity in predicting survival time, including censoring, is a function of features called the hazard function."; See Para. 11; The cumulative Hazard function comprised of the hazard functions output is put into the Survival function which compares the value against a "Censor threshold" that determines if they are alive or not) on the basis of the scalar value (time) obtained from the monotonic neural networks.
Regarding claim 8, most of the limitations of this claim have been noted in the rejection of claim 1 (and thus the rejections of claim 1 are incorporated). Yamashita teaches A non-transitory tangible computer-readable storage medium storing a program (“The analyzer disclosed in the present application is an analyzer that has a processor that executes a program, a storage device that stores the program, and can access a database, and the database is a plurality of data that the learning target has.”; See Para. 21) that causes a processor (Processor) including a hardware in an[[the]] information processing apparatus to execute: calculate a latent representation (Feature Space) representing a feature amount (Feature Amount) regarding a prediction target event (Prediction Target; See Para. 20) from processing target data (Test Data; See Para. 5) including the feature amount (Feature Amount); input the latent representation (Feature Space) to monotonic neural networks that are modeled to output a scalar value (Time) in accordance with a monotonically increasing function (Increasing time) defined by the latent representation (feature Space) and a clock time (“Training data including the first feature amount data including the feature amount, the objective variable indicating the analysis time from the start of the analysis to the end of the analysis of the learning target”; See Para. 21) and obtain the scalar value (Time) from the monotonic neural networks; and estimate at least one of a hazard function (Hazard Function) and a survival function ("The most important quantity in predicting survival time, including censoring, is a function of features called the hazard function."; See Para. 11; The cumulative Hazard function comprised of the hazard functions output is put into the Survival function which compares the value against a "Censor threshold" that determines if they are alive or not) on the basis of the scalar value (Time) obtained from the monotonic neural networks.
Regarding Claim 9, most of the limitations of this claim have been noted in the rejection of claims 1 and 2 (and thus the rejections of claims 1 and 2 are incorporated). Yamashita teaches wherein, for estimating the hazard function (Hazard Function), the processor (Processor) is configured to: calculate a cumulative hazard function (Cumulative Hazard Function) on the basis of the scalar value (Time) obtained from the monotonic neural networks, and calculate the hazard function (Hazard Function) by automatically differentiating the cumulative hazard function ("The hazard function is defined at each time, and the hazard function at time T represents the probability of death (or failure, cancellation) at time T. The integration of the hazard functions up to time T (cumulative hazard function) is the probability of dying by time T, and the point where the cumulative hazard function exceeds a certain threshold is considered to be the death time"; See Para. 11).
Regarding Claim 10, most of the limitations of this claim have been noted in the rejection of claims 1,2, and 9 (and thus the rejections of claims 1,2 and 9 are incorporated). Yamashita teaches wherein the processor (Processor) is further configured to: convert the cumulative hazard function (Cumulative Hazard Function) into the survival function ("Also, if the cumulative hazard function does not exceed the threshold at any time, it is considered alive (censored). Therefore, the prediction of the hazard function is equivalent to the prediction of the survival time, and in the prediction of the survival time, most of the analytical models that predict the hazard function that is easy to interpret are the targets of the prediction."; See Para. 11; The survival function becomes equivalent to the Cumulative hazard function prediction.).
Regarding claim 11, most of the limitations of this claim have been noted in the rejection of claims 1 and 2 (and thus the rejections of claims 1 and 2 are incorporated). Yamashita teaches wherein, for estimating the hazard function (Hazard Function) and the survival function (Survival Function), the processor (Processor) is configured to: calculate the survival function (Survival Function) on the basis of the scalar value (Time) obtained from the monotonic neural networks; and calculate the hazard function by automatically differentiating the survival function ("Also, if the cumulative hazard function does not exceed the threshold at any time, it is considered alive (censored). Therefore, the prediction of the hazard function is equivalent to the prediction of the survival time, and in the prediction of the survival time, most of the analytical models that predict the hazard function that is easy to interpret are the targets of the prediction."; See Para. 11).
Regarding Claim 12, most of the limitations of this claim have been noted in the rejection of claim 1 (and thus the rejections of claim 1 are incorporated). Yamashita teaches wherein, for estimating the hazard function (Hazard Function), the processor (Processor) is configured to: calculate a cumulative hazard function on the basis of the scalar value ("The integration of the hazard functions up to time T (cumulative hazard function)"; See Para. 11) obtained from the monotonic neural networks, and calculate the hazard function by automatically differentiating the cumulative hazard ("The integration of the hazard functions up to time T (cumulative hazard function)"; See Para. 11).
Regarding Claim 13, most of the limitations of this claim have been noted in the rejection of claims 1 and 12 (and thus the rejections of claims 1 and 12 are incorporated). Yamashita teaches, wherein the processor (Processor) is further configured to: convert the cumulative hazard function ("The integration of the hazard functions up to time T (cumulative hazard function)"; See Para. 11) into the survival function ("Also, if the cumulative hazard function does not exceed the threshold at any time, it is considered alive (censored). Therefore, the prediction of the hazard function is equivalent to the prediction of the survival time, and in the prediction of the survival time, most of the analytical models that predict the hazard function that is easy to interpret are the targets of the prediction."; See Para. 11).
Regarding Claim 14, most of the limitations of this claim have been noted in the rejection of claim 1 (and thus the rejections of claim 1 are incorporated). Yamashita teaches wherein, for estimating the hazard function (Hazard Function) and the survival function (Survival Function), the processor (Processor) is configured to: calculate the survival function (Survival Function) on the basis of the scalar value (Time) obtained from the monotonic neural networks; and calculate the hazard function (Hazard Function) by automatically differentiating the survival function ("Also, if the cumulative hazard function does not exceed the threshold at any time, it is considered alive (censored). Therefore, the prediction of the hazard function is equivalent to the prediction of the survival time, and in the prediction of the survival time, most of the analytical models that predict the hazard function that is easy to interpret are the targets of the prediction."; See Para. 11).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to BRETT DAVID KOLB JR whose telephone number is (571)270-0751. The examiner can normally be reached Monday-Friday.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Cesar Paula can be reached at (571) 272-4128. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/BRETT DAVID KOLB JR/ Examiner, Art Unit 2145
/CESAR B PAULA/ Supervisory Patent Examiner, Art Unit 2145
Prosecution Insights