CTNF 18/723,909 CTNF 101555 DETAILED ACTION Notice of Pre-AIA or AIA Status 07-03-aia AIA 15-10-aia The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA. Specification 06-11 AIA The title of the invention is not descriptive. A new title is required that is clearly indicative of the invention to which the claims are directed. 06-11-01 AIA The following title is suggested: “Systems and Methods for Generating Pseudo Labels for Time-Series Data Based on Inference Agreement” Claim Rejections - 35 USC § 101 07-04-01 AIA 07-04 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 1-11 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The apparatus of claim 1 is directed to a machine, which is one of the statutory categories of invention, and passes Step 1: Statutory Category- MPEP § 2106.03. However, all of the limitations of Claim 1 constitute mental processes because they describe acts of observation, evaluation, and judgement that can practically be performed in the human mind, or by a human using pen and paper as a physical aid, therefore failing Step 2A Prong One. These acts are mental processes because a human can review each data piece in a sequence over time and decide what class/category it belongs to, look at several classification results for neighboring time points and determine whether they are mostly the same, partly the same, or inconsistent, and review the inference results, consider the degree of agreement, and then decide what label should be assigned to one or more data pieces. Claim 1 fails Step 2A Prong Two because the additional elements beyond the judicial exception, including a processor, do not integrate the judicial exception into a practical application. There are no improvements to the functioning of a computer or any other technology or technical field (MPEP § 2106.05(a)) as the processor merely apply the abstract idea on a computer (MPEP § 2106.05(f)). Furthermore, the claim does not impose meaningful limits on the computer components such that they are tied to a particular machine; the additional elements are described at a high level of generality and can be implemented on any generic computing system (MPEP § 2106.05(b)). Claim 1 also fails Step 2B, as these additional elements are well-understood, routine, and conventional (WURC), adding nothing significantly more than the abstract idea itself (MPEP § 2106.07(a)((III)). A processor is WURC (see MPEP § 2106.05(d)). Claims 10 and 11 contain this identical ineligible subject matter, with the only additional elements beyond the judicial exception being a non-transitory computer-readable medium, which also does not integrate the judicial exception into a practical application (see claim 1 analysis above) and is WURC (see MPEP § 2106.05(d)). Therefore, they are rejected. Claims 2-9 recite limitations that constitute mental processes because they describe acts of observation, evaluation, and judgement that can practically be performed in the human mind, or by a human using pen and paper as a physical aid, therefore failing Step 2A Prong One. In Claim 2, these acts are mental processes because a human can determine confidence in classifications (“this looks more like class A than B”) and observe classifications and decide whether enough are the same (“most of them agree”). In Claims 3 and 4, these acts are mental processes because a human can apply rule-based decisions to assign labels to all data pieces when certain sameness conditions are met. In Claim 5, these acts are mental processes because a human can perform quantitative comparison and evaluation to compare differences between confidence values and apply a constraint. In Claim 6, these acts are mental processes because a human can visually inspect a sequence and judge whether values are clustered or dispersed. In Claim 7, these acts are mental processes because a human can compare confidence values across classes, evaluate variation, and decide a label. In Claim 8, these acts are mental processes because a human can compare a set of values, evaluate similarity, and apply a threshold. Lastly, in Claim 9, these acts are mental processes because a human learn from labeled examples (e.g., learning patterns from past labeled data). These claims also fail Step 2A Prong Two and Step 2B because the additional elements beyond the judicial exception, including a processor, do not integrate the judicial exception into a practical application and are WURC (see claim 1 analysis above); therefore, they are rejected. Claim Rejections - 35 USC § 102 07-07-aia AIA 07-07 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 – 07-08-aia AIA (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. 07-12-aia AIA (a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention. 07-15 AIA Claim s 1-11 are rejected under 35 U.S.C. 102 ( a)(1 ) as being anticipated by Losing (EP4195109A1) . Regarding Claim 1, Losing teaches an information processing apparatus comprising at least one processor, the at least one processor carrying out: Paragraph [0023]: “The system and/or any of the functions described herein may be implemented using individual hardware circuitry, using software functioning in conjunction with at least one of a programmed microprocessor, a general purpose computer, using an application specific integrated circuit (ASIC) and using one or more digital signal processors (DSPs) .” an inferring process of inferring a class regarding each of data pieces which constitute time-series data; Paragraph [0006]: “According to the present invention, a computer-implemented method for classifying time-series data performed by an apparatus comprises a step of classifying the time-series data by a real-time classification using a real-time classification model pre-trained based on labeled data …” Paragraph [0041]: “As t is the current time, the goal is to assign the correct class for each xt .” Explanation: “Inferring a class” = classifying time series data and “each data piece” = each time step xt a calculating process of calculating a degree of agreement among results of inference made by the inferring means process regarding a plurality of data pieces contained in a section which is temporally continuous; Paragraph [0043]: “The most common approach is to select feature vectors based on a sliding window of a predefined size w . The underlying assumption is that the most recent data is most relevant and that the last w feature vectors are sufficient for classification .” Paragraph [0051]: “Concretely, mretro is retrained in an iterative way, every time transferring the data whose classification confidence surpasses a predefined threshold confmin from the unlabeled to the labeled set …Note that for a set of inputs, the model prediction function returns the labels Ỹ as well as the corresponding confidences C (line 5 of the first algorithm).” Explanation: “Degree of agreement” corresponds to confidence/consistency of predictions, and “plurality of data pieces contained in a section which is temporally continuous” corresponds to sliding window/buffered time-series. and a pseudo label assigning process of assigning a pseudo label based on the results of inference in the section, to at least one of the plurality of data pieces in the section, according to the degree of agreement. Paragraph [0010]: “With the retrospective model, accurate pseudo-labels can be generated .” Paragraph [0018]: “wherein the optimizing means is configured to generate pseudo-labeled data by semi-supervised learning based on the labeled data and the buffered time-series data and to retrain the real-time classification model based on the labeled data and the pseudo-labeled data .” Paragraph [0051]: “Based on the self-training procedure given in the first algorithm shown in Fig. 3, mretro is used to generate the pseudo-labeled data X := {(x̃ 1 , ỹ 1 ), (x̃ 2 , ỹ 2 ), ..., (x̃k , ỹK )} for the unlabeled data U. Concretely, mretro is retrained in an iterative way, every time transferring the data whose classification confidence surpasses a predefined threshold confmin from the unlabeled to the labeled set.” Explanation: Pseudo label assignment is explicitly disclosed and based on confidence (e.g., agreement). Regarding Claim 2, Losing teaches the information processing apparatus according to claim 1, wherein: in the inferring process, a degree of confidence in each of a plurality of classes regarding each of the data pieces is calculated, Paragraph [0051]: “Note that for a set of inputs, the model prediction function returns the labels Ỹ as well as the corresponding confidences C (line 5 of the first algorithm).” Explanation: Direct teaching of per-class confidence. and in the pseudo label assigning process, in a case where a sameness of classes of highest degrees of confidence regarding the plurality of data pieces contained in the section is not smaller than a predetermined proportion, a pseudo label based on the results of inference in the section is assigned to at least one of the plurality of data pieces in the section. Paragraph [0051]: “Based on the self-training procedure given in the first algorithm shown in Fig. 3, mretro is used to generate the pseudo-labeled data X := {(x̃ 1 , ỹ 1 ), (x̃ 2 , ỹ 2 ), ..., (x̃k , ỹK )} for the unlabeled data U…Concretely, mretro is retrained in an iterative way, every time transferring the data whose classification confidence surpasses a predefined threshold confmin from the unlabeled to the labeled set.” Explanation: Threshold = predetermined proportion and “sameness” = consistent high-confidence predictions. Regarding Claim 3, Losing teaches the information processing apparatus according to claim 2, wherein in the pseudo label assigning process, in a case where the sameness of classes of highest degrees of confidence regarding the plurality of data pieces contained in the section is not smaller than a predetermined proportion, a pseudo label based on the results of inference in the section is assigned to all of the plurality of data pieces in the section. Paragraph [0051]: “Based on the self-training procedure given in the first algorithm shown in Fig. 3, mretro is used to generate the pseudo-labeled data X := {(x̃ 1 , ỹ 1 ), (x̃ 2 , ỹ 2 ), ..., (x̃k , ỹK )} for the unlabeled data U…Concretely, mretro is retrained in an iterative way, every time transferring the data whose classification confidence surpasses a predefined threshold confmin from the unlabeled to the labeled set.” Explanation: Labels are assigned across the dataset. Regarding Claim 4, Losing teaches the information processing apparatus according to claim 3, wherein in the pseudo label assigning process, in a case where all of classes of highest degrees of confidence regarding the plurality of data pieces contained in the section are the same, a pseudo label based on the results of inference in the section is assigned to all of the plurality of data pieces in the section. Paragraph [0051]: “Based on the self-training procedure given in the first algorithm shown in Fig. 3, mretro is used to generate the pseudo-labeled data X := {(x̃ 1 , ỹ 1 ), (x̃ 2 , ỹ 2 ), ..., (x̃k , ỹK )} for the unlabeled data U…Concretely, mretro is retrained in an iterative way, every time transferring the data whose classification confidence surpasses a predefined threshold confmin from the unlabeled to the labeled set.” Explanation: Only consistent high-confidence predictions are used which shows a same dominant class. Regarding Claim 5, Losing teaches the information processing apparatus according to claim 1, wherein in the pseudo label assigning process, in a case where classes of highest degrees of confidence regarding any pair of data pieces of the plurality of data pieces in the section are the same, and a difference between the highest degrees of confidence is such that a greater one of the highest degrees of confidence is not greater than any constant number of times the other, a pseudo label based on the results of inference in the section is assigned to at least one of the plurality of data pieces in the section. Paragraph [0051]: “Based on the self-training procedure given in the first algorithm shown in Fig. 3, mretro is used to generate the pseudo-labeled data X := {(x̃ 1 , ỹ 1 ), (x̃ 2 , ỹ 2 ), ..., (x̃k , ỹK )} for the unlabeled data U…Concretely, mretro is retrained in an iterative way, every time transferring the data whose classification confidence surpasses a predefined threshold confmin from the unlabeled to the labeled set.” Explanation: Threshold bounds differences between confidences. Regarding Claim 6, Losing teaches the information processing apparatus according to claim 1, wherein in the pseudo label assigning process, according to dispersion, along a time axis, in highest degrees of confidence in the plurality of respective data pieces contained in the section, a pseudo label based on the results of inference in the section is assigned to at least one of the plurality of data pieces in the section. Paragraph [0009]: “In the iterative self-training process, a retrospective model classifying a sample x̃t of the buffered time-series data based on the labeled data and samples x̃t-l , ... , x̃t , ... , x̃t+m of the buffered time-series data can be retrained , where m and l are parameters set based on, for example, the type of the time series data, desired classification accuracy and/or computational effort.” Paragraph [0051]: “Based on the self-training procedure given in the first algorithm shown in Fig. 3, mretro is used to generate the pseudo-labeled data X := {(x̃ 1 , ỹ 1 ), (x̃ 2 , ỹ 2 ), ..., (x̃k , ỹK )} for the unlabeled data U.” Paragraph [0054]: “xt represents the raw input vector at time t. The actual input vector of the online classification model monline is a feature map based on the sliding window of the recent w raw inputs Øt = Ø(x t-w +1 , ..., xt ) .” Explanation: Explicit temporal dispersion across time-series. Regarding Claim 7, Losing teaches the information processing apparatus according to claim 1, wherein in the pseudo label assigning process, according to dispersion in degrees of confidence of at least one of the plurality of data pieces contained in the section, the degrees of confidence being of respective classes in the at least one of the plurality of data pieces, a pseudo label based on the results of inference in the section is assigned to the at least one of the plurality of data pieces in the section. Paragraph [0051]: “Based on the self-training procedure given in the first algorithm shown in Fig. 3, mretro is used to generate the pseudo-labeled data X := {(x̃ 1 , ỹ 1 ), (x̃ 2 , ỹ 2 ), ..., (x̃k , ỹK )} for the unlabeled data U…Note that for a set of inputs, the model prediction function returns the labels Ỹ as well as the corresponding confidences C (line 5 of the first algorithm).” Explanation: Confidence exists per class, meaning that dispersion is present. Regarding Claim 8, Losing teaches the information processing apparatus according to claim 1, wherein: in the inferring process, a degree of confidence in each of a plurality of classes regarding each of the data pieces is calculated, Paragraph [0051]: “Note that for a set of inputs, the model prediction function returns the labels Ỹ as well as the corresponding confidences C (line 5 of the first algorithm).” and in the pseudo label assigning process, in a case where a distance between distributions of degrees of confidence regarding the plurality of data pieces contained in the section is not greater than a predetermined value, assign a pseudo label based on the results of inference in the section is assigned to at least one of the plurality of data pieces in the section. Paragraph [0051]: “Based on the self-training procedure given in the first algorithm shown in Fig. 3, mretro is used to generate the pseudo-labeled data X := {(x̃ 1 , ỹ 1 ), (x̃ 2 , ỹ 2 ), ..., (x̃k , ỹK )} for the unlabeled data U…Concretely, mretro is retrained in an iterative way, every time transferring the data whose classification confidence surpasses a predefined threshold confmin from the unlabeled to the labeled set.” Explanation: Threshold constrains distribution differences. Regarding Claim 9, Losing teaches the information processing apparatus according to claim 1, the at least one processor further carries out a training process for training an inferring means with use of time-series data which contains a pseudo label assigned by the pseudo label assigning process, the inferring means carrying out the inferring process. Paragraph [0006]: “According to the present invention, a computer-implemented method for classifying time-series data performed by an apparatus comprises a step of classifying the time-series data by a real-time classification using a real-time classification model pre-trained based on labeled data , a step of buffering the time-series data that has already been classified by the real-time classification and a step of optimizing the real-time classification model during the real-time classification by generating pseudo-labeled data by semi-supervised learning based on the labeled data and the buffered time-series data and retraining the real-time classification model based on the labeled data and the pseudo-labeled data .” Regarding Claim 10, Losing teaches all of the limitations of claim 1 above because claim 10 recites a method that performs substantially the same steps as claim 1. Regarding Claim 11, Losing teaches all of the limitations of claim 1 above because claim 11 recites a non-transitory storage medium storing a program for causing a computer to perform substantially the same steps as claim 1. Paragraph [0017]: “ According to the present invention, a program that, when running on a computer or loaded onto a computer, causes the computer to execute the steps of : classifying the time-series data by a real-time classification using a real-time classification model pre-trained based on labeled data, buffering the time-series data that have already been classified by the real-time classification and optimizing the real-time classification model during the real-time classification by generating pseudo-labeled data by semi-supervised learning based on the labeled data and the buffered time-series data and retraining the real-time classification model based on the labeled data and the pseudo-labeled data.” Conclusion 07-96 AIA The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. Fan et. al (“Semi-Supervised Time Series Classification by Temporal Relation Prediction”) teaches a simple and effective method of Semi-supervised Time series classification architecture (termed as SemiTime) by gaining from the structure of unlabeled data in a self-supervised manner. Any inquiry concerning this communication or earlier communications from the examiner should be directed to WILLIAM ADU-JAMFI whose telephone number is (571) 272-9298. The examiner can normally be reached M-T 8:00-6:00. Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice. If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Andrew Bee can be reached at (571) 270-5183. 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. /WILLIAM ADU-JAMFI/Examiner, Art Unit 2677 /ANDREW W BEE/Supervisory Patent Examiner, Art Unit 2677 Application/Control Number: 18/723,909 Page 2 Art Unit: 2677 Application/Control Number: 18/723,909 Page 3 Art Unit: 2677 Application/Control Number: 18/723,909 Page 4 Art Unit: 2677 Application/Control Number: 18/723,909 Page 5 Art Unit: 2677 Application/Control Number: 18/723,909 Page 6 Art Unit: 2677 Application/Control Number: 18/723,909 Page 7 Art Unit: 2677 Application/Control Number: 18/723,909 Page 8 Art Unit: 2677 Application/Control Number: 18/723,909 Page 9 Art Unit: 2677 Application/Control Number: 18/723,909 Page 10 Art Unit: 2677 Application/Control Number: 18/723,909 Page 11 Art Unit: 2677
Prosecution Insights