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
This Office Action is sent in response to Applicant’s Communication received
on 10/13/2025 for application number 17742966.
Claims 1, 11, 15, and 18-20 have been amended.
Claims 1-20 remain pending in the application.
The amendment filed 10/13/2025 is sufficient to overcome the prior art rejections to claims 1-6 and 11-13 over Zhang in view of Tai, Sumida, and Beard, the prior art rejections of claims 7, 8, and 10 over Zhang in view of Tai, Sumida, Beard, and Dev, the prior art rejections of claims 14 and 15 over Zhang in view of Tai, Sumida, and Beard and further in view of Xaio, the prior art rejections of claims 16, 17, 19, and 20 over Zhang in view of Tai, and the prior art rejection of claim 18 over Zhang in view of Tai and further in view of Sumida and Beard. The previous rejections have been withdrawn.
Argument 1, regarding the 101 rejections, applicant argues that the claims do not recite any judicial exceptions. Examiner respectfully disagrees because the claims recite:
a) generating, by a computer system, an initial source window size and an initial target window size (Mental Process);…c) generating, by the computer system, a state value using the one or more initial source time series data values and the one or more initial target time series data values (Mathematical Concept); and d) for each time value up to a training epoch value, performing at least the following steps: (i) generating, by the computer system, using a context sampler, an action comprising a source window size and a target window size based on the state value (Mathematical Concept); …(iv) updating, by the computer system, the state value to an updated state value using the one or more source time series data values and the one or more target time series data values (Mathematical Concept); (v) computing, by the computer system, a reward value using the one or more source time series data values, and the one or more target time series data values (Mathematical Concept).
In view of P0072-P0082 of the specification of the instant application, generating and updating state values includes mathematical concepts. In view of P0013 of the specification of the instant application, determining an initial window size may be practically performed by a human domain expert, which means the limitation includes a mental process. Accordingly, the claims recite an abstract idea.
Applicant also argues that the claims reflect an improvement to domain adaptation by removing the need for a human expert as well as improve classification accuracy, and cites figures 12-13 and paragraphs 14 and 214 for how the invention provides these improvements. Examiner notes that these improvements are not reflected in the claims because claims do not reflect improving classification accuracy or reducing the need for a human expert even if the specification may explicitly recite these improvements. One way to determine integration into a practical application is when the claimed invention improves the functioning of a computer or improves another technology or technical field. To evaluate an improvement to a computer or technical field, the specification must set for than improvement in technology and the claim itself must reflect the disclosed improvement. See MPEP 2106.04(d)(1) and 2106.05(a).
Applicant also argues that the context sampler improves anomaly detection by producing window sizes and target window sizes that lead to better anomaly detection performance. Applicant also argues that the claims recite the use of target window size, source window size, computer system, context sampler, state value, action vector, and reward value, among other claimed elements, when taken in combination as a whole are a non-generic and non-conventional arrangement that provide an inventive concept of using different context window settings for domain adaptation for anomaly detection in time-series data. Examiner respectfully disagrees because the training steps are interpreted as generally linking the abstract ideas to the technological environment of machine learning, as discussed in MPEP 2106.05(h). There does not appear to be any training steps that integrate the abstract ideas into improved anomaly detection.
The full 101 rejections are outlined below.
Argument 2, regarding the prior art rejections, applicant argues that none of the cited prior art teaches a two-dimensional action vector. Applicant’s arguments regarding the prior art rejections have been considered but they are moot in view of the new grounds of rejection necessitated by applicant’s amendment.
US 12493796 B2 (Nia et al) teaches a two-dimensional action vector (a two-dimensional vector being processed by model 110 which is used for anomaly detection, P0068, P0037).
Accordingly, it would have been obvious to a person having ordinary skill in the
art before the effective filing date of the claimed invention, having the teachings of
Zhang, Tai, Sumida, Beard, and Nia before them, to include Nia’s specific teachings of a two-dimensional vector being processed by a model used for anomaly detection in Zhang’s system of anomaly detection. One would have been motivated to make such a combination of a two-dimensional vector being processed by a model used for anomaly detection (see Nia P0068, P0037), and analyzing training data stored in memory to detect anomalies (see Zhang P0013).
The full prior art rejections are outlined below.
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 1-20 are rejected under 35 U.S.C. 101 because the claimed
invention is directed to an abstract idea without significantly more.
Step 1: The claims recite a method and system, each being one of the four categories of eligible subject matter.
Claims 1 and 11
Step 2A Prong 1: The claims recite the following limitations:
a) generating, by a computer system, an initial source window size and an initial target window size (Mental Process);…c) generating, by the computer system, a state value using the one or more initial source time series data values and the one or more initial target time series data values (Mathematical Concept); and d) for each time value up to a training epoch value, performing at least the following steps: (i) generating, by the computer system, using a context sampler, a two-dimensional action vector comprising a source window size and a target window size based on the state value (Mathematical Concept); …(iv) updating, by the computer system, the state value to an updated state value using the one or more source time series data values and the one or more target time series data values (Mathematical Concept); (v) computing, by the computer system, a reward value using the one or more source time series data values, and the one or more target time series data values (Mathematical Concept).
In view of P0072-P0082 of the specification of the instant application, generating and updating state values includes mathematical concepts. In view of P0013 of the specification of the instant application, determining an initial window size may be practically performed by a human domain expert, which means the limitation includes a mental process. Accordingly, the claims recite an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical
application. The claims recite the following additional elements:
b) sampling, by the computer system, one or more initial source time series data values and one or more initial target time series data values using the initial source window size and the initial target window size;… (ii) sampling, by the computer system, one or more source time series data values based on the source window size and the time value; (iii) sampling, by the computer system, one or more target time series data values based on the target window size and the time value;… (vi) storing, by the computer system, a tuple including the state value, the two-dimensional action vector comprising the source window size and the target window size, the updated state value, and the reward value in a memory buffer of the context sampler; and (vii) training, by the computer system, the context sampler in an iterative process using sampled data comprising at least the tuple from the memory buffer of the context sampler.
Collecting source time series data values and storing data in memory is mere data gathering, which is an insignificant extra-solution activity as discussed in MPEP 2106.05(g). The memory, processor, and non-transitory computer-readable medium are
generic computing components recited at a high level as a means to apply the judicial
exception, as discussed in MPEP 2106.05(f). Performing iterative learning is generally linking the abstract ideas to the technological environment of machine learning, as discussed in MPEP 2106.05(h). The claims are directed to an abstract idea.
Step 2B: The claims do not include additional elements that are sufficient to
amount to significantly more than the judicial exception. Collecting source time series data values and storing data in memory is mere data gathering, which is an insignificant extra-solution activity as discussed in MPEP 2106.05(g). The memory, processor, and non-transitory computer-readable medium are generic computing components recited at a high level as a means to apply the judicial exception, as discussed in MPEP 2106.05(f). Performing iterative learning is generally linking the abstract ideas to the technological environment of machine learning, as discussed in MPEP 2106.05(h). The claims are not patent eligible.
Claim 16
Step 2A Prong 1: The claim recites the following limitations:
setting, by the computer system, a source window size and a target window size using a trained context sampler (Mental Process);… and determining, by the computer system, using the anomaly detector, whether a target data value in the one or more target time series data values comprises an anomalous data value (Mental Process).
In view of P0013 of the specification of the instant application, determining an initial window size may be practically performed by a human domain expert, which means the limitation includes a mental process. Under the broadest reasonable interpretation of the claim language, determining whether or not a data value is anomalous is a mental process. Accordingly, the claim recites an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical
application. The claim recites the following additional elements:
obtaining, by a computer system, a source data set and a target data set, wherein the source data set comprises the plurality of source time series data values, and wherein the target data set comprises a plurality of target time series data values, wherein the plurality of source time series data values are labeled and the plurality of target time series data values are unlabeled;… sampling, by the computer system, one or more source time series data values from the source data set using the source window size and based on a time value; sampling, by the computer system, one or more target time series data values from the target data set using the target window size and based on the time value; providing, by the computer system, the one or more source time series data values and the one or more target time series data values to an anomaly detector;
Collecting a source and target data set, sampling one or more source time series data values, and delivering one or more source time series data values to an anomaly detector are all mere data gathering, which is an insignificant extra-solution activity as discussed in MPEP 2106.05(g). The claim is directed to an abstract idea.
Step 2B: The claim does not include additional elements that are sufficient to
amount to significantly more than the judicial exception. Collecting a source and target data set, sampling one or more source time series data values, and delivering one or more source time series data values to an anomaly detector are all mere data gathering, which is an insignificant extra-solution activity as discussed in MPEP 2106.05(g). The claim is not patent eligible.
Dependent Claims
Claim 2
Step 2A Prong 1: The judicial exceptions of claim 1 are incorporated. The claim recites the following limitations:
and after step d), e) determining, by the computer system, a frequently selected source window size based on the plurality of source window sizes, wherein the frequently selected source window size is used during an anomaly detection process (Mental Process).
Determining a window size is a mental process under the broadest reasonable interpretation of the claim language and in view of P0013 of the specification of the instant application. Accordingly, the claim recites an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical
application. The claim recites the following additional elements:
after step (vii), (viii) recording, by the computer system, each source window size, thereby recording a plurality of source window sizes proportional to the training epoch value.
Recording window sizes proportional to each training epoch is mere data gathering, which is an insignificant extra-solution activity as discussed in MPEP 2106.05(g). The claim is directed to an abstract idea.
Step 2B: The claim does not include additional elements that are sufficient to
amount to significantly more than the judicial exception. Recording window sizes proportional to each training epoch is mere data gathering, which is an insignificant extra-solution activity as discussed in MPEP 2106.05(g). The claim is not patent eligible.
Claim 3
Step 2A Prong 1: The judicial exceptions of claim 1 are incorporated. The claim recites the following limitations:
… which determines one or more anomalies in the target time series data values (Mental Process).
Determining one or more anomalies in data values is a mental process under the broadest reasonable interpretation of the claim language. Accordingly, the claim recites an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical
application. The claim recites the following additional elements:
inputting source data values and target data values into the trained context sampler, which outputs source time series data values and target time series data values to an anomaly detector.
Inputting data into a trained context sampler which then outputs data into an anomaly detector is mere data gathering, which is an insignificant extra-solution activity as discussed in MPEP 2106.05(g). The claim is directed to an abstract idea.
Step 2B: The claim does not include additional elements that are sufficient to
amount to significantly more than the judicial exception. Inputting data into a trained context sampler which then outputs data into an anomaly detector is mere data gathering, which is an insignificant extra-solution activity as discussed in MPEP 2106.05(g). The claim is not patent eligible.
Claim 4
Step 2A Prong 1: The judicial exceptions of claim 3 are incorporated. The claim recites the following limitations:
wherein the reward value is computed using one or more loss values comprising a classification loss value, a reconstruction loss value, an alignment loss value, and a domain discrimination loss value, and wherein the method further comprises computing the one or more loss values (Mathematical Concept).
Computing a reward value is a mathematical concept under the broadest reasonable interpretation of the claim language. Accordingly, the claim recites an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical application and the claim does not recite additional elements. The claim is directed to an abstract idea.
Step 2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Claim 5
Step 2A Prong 1: The judicial exceptions of claim 4 are incorporated. The claim recites the following limitations:
wherein the reward value is computed using a classification hyper-parameter, a reconstruction hyper-parameter, an alignment hyper- parameter, and a domain discrimination hyper-parameter (Mathematical Concept).
Computing a reward value with various hyper-parameters is a mathematical concept in view of P0092 of the specification of the instant application.
Accordingly, the claim recites an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical application and the claim does not recite additional elements. The claim is directed to an abstract idea.
Step 2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Claim 6
Step 2A Prong 1: The judicial exceptions of claim 4 are incorporated. The claim recites the following limitations:
wherein the anomaly detector comprises an encoder, a decoder, a classifier, and a domain classifier, and wherein: the classification loss value is computed using the classifier, the encoder, the one or more source time series data values, one or more labels corresponding to the one or more source time series data values, and one or more weights corresponding to the one or more source time series data values (Mathematical Concept); the reconstruction loss value is calculated using the encoder, the decoder, the one or more source time series data values, and the one or more target time series data values; the alignment loss value is calculated using the encoder, the one or more source time series data values, and the one or more target time series data values (Mathematical Concept); and the domain discrimination loss value is calculated using the encoder, the domain classifier, the one or more source time series data values, the one or more target time series data values, and one or more domain labels (Mathematical Concept).
Classification loss is calculated using the formula described in P0125 of the specification of the instant application. Reconstruction loss is calculated using the formula described in P0128 of the specification of the instant application. Domain discrimination loss is calculated using the formula described in P0126 of the specification of the instant application. Each of these calculations are mathematical concepts.
Accordingly, the claim recites an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical application and the claim does not recite additional elements. The claim is directed to an abstract idea.
Step 2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Claim 7
Step 2A Prong 1: The judicial exceptions of claim 6 are incorporated. Accordingly, the claim recites an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical
application. The claim recites the following additional elements:
training, by the computer system, the anomaly detector by training the encoder, the decoder, the classifier, and the domain classifier.
Training components of a machine learning models is generally linking the abstract ideas to the technological environment of machine learning, as discussed in MPEP 2106.05(h). The claim is directed to an abstract idea.
Step 2B: The claim does not include additional elements that are sufficient to
amount to significantly more than the judicial exception. Training components of a machine learning models is generally linking the abstract ideas to the technological environment of machine learning, as discussed in MPEP 2106.05(h). The claim is not patent eligible.
Claim 8
Step 2A Prong 1: The judicial exceptions of claim 6 are incorporated. Accordingly, the claim recites an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical
application. The claim recites the following additional elements:
the classifier comprises a multi-layer perceptron classifier with a sigmoid activation function; the domain classifier comprises another multi-layer perceptron classifier with a sigmoid activation function; and the encoder and the decoder comprise two parts of an LSTM autoencoder.
The machine learning model including components such as multi-layer perceptron classifiers with a sigmoid activation function and parts of an LSTM autoencoder is generally linking the abstract ideas to the technological environment of machine learning, as discussed in MPEP 2106.05(h). The claim is directed to an abstract idea.
Step 2B: The claim does not include additional elements that are sufficient to
amount to significantly more than the judicial exception. The machine learning model including components such as multi-layer perceptron classifiers with a sigmoid activation function and parts of an LSTM autoencoder is generally linking the abstract ideas to the technological environment of machine learning, as discussed in MPEP 2106.05(h). The claim is not patent eligible.
Claim 9
Step 2A Prong 1: The judicial exceptions of claim 1 are incorporated. Accordingly, the claim recites an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical
application. The claim recites the following additional elements:
the one or more initial source time series data values comprise a subsequence of a source data set, wherein a number of initial source time series data values in the subsequence of the source data set is proportional to the initial source window size; the one or more initial target time series data values comprise a subsequence of a target data set, wherein a number of initial source time series data values in the subsequence of the target data set is proportional to the initial target window size; the one or more source time series data values comprise a subsequence of the source data set, wherein a number of source time series data values in the subsequence of the source data set is proportional to the source window size; and the one or more target time series data values comprise a subsequence of the target data set, wherein a number of target time series data values in the subsequence of the target data set is proportional to the target window size.
Sampling initial source, initial target, source, and target time series data which each comprising target or source data proportional to the initial window size is mere data gathering, which is an insignificant extra-solution activity as discussed in MPEP 2106.05(g). The claim is directed to an abstract idea.
Step 2B: The claim does not include additional elements that are sufficient to
amount to significantly more than the judicial exception. Sampling initial source, initial target, source, and target time series data which each comprising target or source data proportional to the initial window size is mere data gathering, which is an insignificant extra-solution activity as discussed in MPEP 2106.05(g). The claim is not patent eligible.
Claim 10
Step 2A Prong 1: The judicial exceptions of claim 9 are incorporated. Accordingly, the claim recites an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical
application. The claim recites the following additional elements:
wherein a cyclical sampling process is used to sample the one or more initial source time series data values, the one or more initial target time series data values, the one or more source time series data values, and the one or more target time series data values.
Cyclical sampling being used to sample one or more initial source time series data values, the one or more initial target time series data values, the one or more source time series data values, and the one or more target time series data values is mere data gathering, which is an insignificant extra-solution activity as discussed in MPEP 2106.05(g). The claim is directed to an abstract idea.
Step 2B: The claim does not include additional elements that are sufficient to
amount to significantly more than the judicial exception. Cyclical sampling being used to sample one or more initial source time series data values, the one or more initial target time series data values, the one or more source time series data values, and the one or more target time series data values is mere data gathering, which is an insignificant extra-solution activity as discussed in MPEP 2106.05(g). The claim is not patent eligible.
Claim 12
Step 2A Prong 1: The judicial exceptions of claim 11 are incorporated. The claim recites the following limitations:
wherein the state value comprises combination of a source encoding generated using the encoder and the one or more source time series data values and a target encoding generated using the encoder and the one or more target time series data values (Mathematical Concept).
In view of P0072-P0082 of the specification of the instant application, generating and updating state values, with state values comprising a combination of source and target encoding, includes mathematical concepts. Accordingly, the claim recites an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical application and the claim does not recite additional elements. The claim is directed to an abstract idea.
Step 2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Claim 13
Step 2A Prong 1: The judicial exceptions of claim 11 are incorporated. Accordingly, the claim recites an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical
application. The claim recites the following additional elements:
wherein the sampled data used to train the context sampler comprises the tuple and one or more previous tuples stored in the memory buffer.
Using tuples to train a context sampler is generally linking the abstract ideas to the technological environment of machine learning, as discussed in MPEP 2106.05(h). The memory is a generic computing component recited at a high level as a means to apply the judicial exception, as discussed in MPEP 2106.05(f). The claim is directed to an abstract idea.
Step 2B: The claim does not include additional elements that are sufficient to
amount to significantly more than the judicial exception. Using tuples to train a context sampler is generally linking the abstract ideas to the technological environment of machine learning, as discussed in MPEP 2106.05(h). The memory is a generic computing component recited at a high level as a means to apply the judicial exception, as discussed in MPEP 2106.05(f). The claim is not patent eligible.
Claim 14
Step 2A Prong 1: The judicial exceptions of claim 11 are incorporated. The claim recites the following limitations:
wherein training the context sampler comprises optimizing a Q-function associated with the context sampler (Mathematical Concept).
Optimizing a Q-function is a mathematical concept as described in P0163-P0166 of the specification of the instant application. Accordingly, the claim recites an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical application and the claim does not recite additional elements. The claim is directed to an abstract idea.
Step 2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Claim 15
Step 2A Prong 1: The judicial exceptions of claim 14 are incorporated. The claim recites the following limitations:
optimizing the Q-function improves a quality of one or more two-dimensional action vectors comprising one or more source window sizes and one or more target window sizes generated by the context sampler, which in turn improves a quality of one or more encodings, one or more reconstructions, one or more classifications, and one or more domain classifications generated by the anomaly detector, thereby reducing one or more loss values computed by the anomaly detector, thereby increasing the reward value (Mathematical Concept).
Optimizing a Q-function is a mathematical concept as described in P0163-P0166 of the specification of the instant application. Accordingly, the claim recites an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical application and the claim does not recite additional elements. The claim is directed to an abstract idea.
Step 2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Claim 17
Step 2A Prong 1: The judicial exceptions of claim 16 are incorporated. The claim recites the following limitations:
wherein determining, using the anomaly detector, whether the target data value comprises the anomalous data value comprises: generating, by the computer system, a target encoding using an encoder and the target data value (Mathematical Concept); generating, by the computer system, a target classification using a classifier and the target encoding (Mathematical Concept); generating, by the computer system, a target reconstruction using a decoder and the target encoding (Mathematical Concept); generating, by the computer system, a reconstruction loss value using the target data value and the target reconstruction (Mathematical Concept); generating, by the computer system, an anomaly score using the target classification and the reconstruction loss value (Mathematical Concept); and comparing, by the computer system, the anomaly score to an anomaly score threshold, wherein if the anomaly score is greater than the anomaly score threshold, the computer system determines that the target data value comprises the anomalous data value (Mental Process).
Generating encoding, classification, reconstruction, reconstruction loss, and an anomaly score is a mathematical concept under the broadest reasonable interpretation of the claim language. Comparing an anomaly value to a threshold and determining that a target value is anomalous depending on the comparison is a mental process under the broadest reasonable interpretation. Accordingly, the claim recites an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical application and the claim does not recite additional elements. The claim is directed to an abstract idea.
Step 2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Claim 18
Step 2A Prong 1: The judicial exceptions of claim 16 are incorporated. The claim recites the following limitations:
a) generating, by the computer system, an initial source window size and an initial target window size (Mental Process); … c) generating, by the computer system, a state value using the one or more initial source time series data values and the one or more initial target time series data values (Mathematical Concept); and d) for each time value up to a training epoch value, performing at least the following steps: (i) generating, by the computer system, using a context sampler, a two-dimensional action vector comprising another source window size and another target window size based on the state value (Mental Process); …(iv) updating, by the computer system, the state value to an updated state value using the one or more additional source time series data values and the one or more additional target time series data values (Mathematical Concept); (v) computing, by the computer system, a reward value using the anomaly detector, the one or more additional source time series data values, and the one or more additional target time series data values (Mathematical Concept).
In view of P0072-P0082 of the specification of the instant application, generating and updating state values includes mathematical concepts. In view of P0013 of the specification of the instant application, determining an initial window size may be practically performed by a human domain expert, which means the limitation includes a mental process. Accordingly, the claims recite an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical
application. The claims recite the following additional elements:
b) sampling, by the computer system, one or more initial source time series data values and one or more initial target time series data values using the initial source window size and the initial target window size;… (ii) sampling, by the computer system, one or more additional source time series data values based on the another source window size and the time value; (iii) sampling, by the computer system, one or more additional target time series data values based on the another target window size and the time value;… (vi) storing, by the computer system, a tuple including the state value, the two-dimensional action vector comprising the another source window size and the another target window size, the updated state value, and the reward value in a memory buffer of the context sampler; and (vii) training, by the computer system, the context sampler in an iterative process using sampled data comprising at least the tuple from the memory buffer of the context sampler.
Collecting source time series data values and storing data in memory is mere data gathering, which is an insignificant extra-solution activity as discussed in MPEP 2106.05(g). The memory, processor, and non-transitory computer-readable medium are
generic computing components recited at a high level as a means to apply the judicial
exception, as discussed in MPEP 2106.05(f). Performing iterative learning is generally linking the abstract ideas to the technological environment of machine learning, as discussed in MPEP 2106.05(h). The claims are directed to an abstract idea.
Step 2B: The claims do not include additional elements that are sufficient to
amount to significantly more than the judicial exception. Collecting source time series data values and storing data in memory is mere data gathering, which is an insignificant extra-solution activity as discussed in MPEP 2106.05(g). The memory, processor, and non-transitory computer-readable medium are generic computing components recited at a high level as a means to apply the judicial exception, as discussed in MPEP 2106.05(f). Performing iterative learning is generally linking the abstract ideas to the technological environment of machine learning, as discussed in MPEP 2106.05(h). The claims are not patent eligible.
Claim 19
Step 2A Prong 1: The judicial exceptions of claim 16 are incorporated. The claim recites the following limitations:
generating, by the computer system, a state value using an encoder, the one or more source time series data values and the one or more target time series data values (Mathematical Concept); and generating, by the computer system, a two-dimensional action vector comprising a subsequent source window size and a subsequent target window size using a context sampler and the state value, wherein the subsequent source window size and subsequent target window size are used to determine whether a subsequent target time series data value corresponding to a subsequent time value comprises a subsequent anomalous data value (Mathematical Concept).
In view of P0072-P0082 of the specification of the instant application, generating and updating state values includes mathematical concepts. Accordingly, the claim recites an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical application and the claim does not recite additional elements. The claim is directed to an abstract idea.
Step 2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Claim 20
Step 2A Prong 1: The judicial exceptions of claim 19 are incorporated. The claim recites the following limitations:
wherein the subsequent source window size is equal to the initial source window size (Mental Process), and wherein the two-dimensional action vector comprising the subsequent source window size and the subsequent target window size are generated by inputting the state value into a trained policy function associated with the context sampler (Mathematical Concept).
In view of P0013 of the specification of the instant application, determining an initial window size may be practically performed by a human domain expert. Thus, setting the subsequent source window size to be equal to the initial source window size includes a mental process. In view of P0072-P0082 of the specification of the instant application, generating and updating state values includes mathematical concepts. Accordingly, the claims recite an abstract idea.
Step 2A Prong 2: The judicial exceptions are not integrated into a practical application and the claim does not recite additional elements. The claim is directed to an abstract idea.
Step 2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Claim Rejections - 35 USC § 103
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 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 1-6 and 11-13 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang et al (Pub. No.: US 20220292074), hereafter Zhang, in view of Tai et al (Pub. No.: US 20250022164 A1), hereafter Tai, Sumida et al (Pub. No.: US 20240411283 A1), hereafter Sumida, Beard et al (Pub. No.: US 20240205260 A1), hereafter Beard, and Nia et al (Pub. No.: US 12493796 B2), hereafter Nia.
Regarding claims 1 and 11, Zhang teaches a) generating, by a computer system, an initial source window size and an initial target window size (Initial window size is set to two weeks of collected data, this window may be adjusted, P0079. “Such a default amount may be customized, for example, by users based on the nature of the data (e.g., 14 day time frame associated with throughput data collected every 5 minutes). As another example, a target number of data points may be obtained. A target number of data points N may be determined based on a minimal time coverage and data frequency. ”, P0039. Different sources/domains of data may have a different associated window size, as different sources of data may be sampled with data collection center 102, P0039, P0023); b) sampling, by the computer system, one or more initial … time series data values and one or more initial … time series data values using the initial source window size and the initial target window size (Data collection center 102 is used to sample time series data, as well as collect a target number of data points P0024, P0039. Time series data is collected as shown in figures 3A-3F, with an initial source window size as described in P0079. Data may be collected from different sources or domains, P0023); c) generating, by the computer system, a state value using the one or more initial source time series data values and the one or more initial target time series data values (Time series data may be used by a prediction model to perform predictions. State of the model is dependent upon sampled time series data, P0012, P0017, P0032); and d) for each time value up to a training epoch value, performing at least the following steps: (i) generating, by the computer system, using a context sampler, …a source window size and a target window size based on the state value (“a rolling window of up to a particular number of days D of latest data may be stored. The specific number of days used as a basis for the rolling window may depend on the granularity of the data… Generally, the number of days D may be reduced for data with higher frequencies as they already provide a larger amount of data points within a shorter period of time.”, P0076. Initial window size is proportional to the target number of data points to be received, with a default time frame being set which is dependent on N number of data points, P0039); (ii) sampling, by the computer system, one or more source time series data values based on the source window size and the time value (source time series data is collected based on the rolling window and time, P0074-P0076); (iii) sampling, by the computer system, one or more target time series data values based on the target window size and the time value (A target number of time series data points may be obtained based on a rolling window and time, P0039, P0074-P0076); (iv) updating, by the computer system, the state value to an updated state value using the one or more source time series data values and the one or more target time series data values (Collected time series data is used to update the state of the prediction model, P0032);… and (vii) training, by the computer system, the context sampler in an iterative process using sampled data (Data store 118, which captures sample data, may be updated in an iterative process, P0074, P0078).
Zhang does not appear to explicitly teach sampling… one or more initial source values…target values
Tai teaches sampling… one or more initial source values…target values (Source domain feature data is collected, P0069. Target domain feature data is collected, P0076).
Accordingly, it would have been obvious to a person having ordinary skill in the
art before the effective filing date of the claimed invention, having the teachings of
Zhang and Tai before them, to include Tai’s specific teachings of collecting data from a source and target domain in Zhang’s system of anomaly detection. One would have been motivated to make such a combination of collecting data from a source and target domain (see Tai P0069, P0076), and collecting data from different sources or domains to detect anomalies (see Zhang P0019, P0023).
Zhang in view of Tai does not appear to explicitly teach (v) computing, by the computer system, a reward value using the one or more source time series data values, and the one or more target time series data values.
Sumida teaches (v) computing, by the computer system, a reward value using the one or more source time series data values, and the one or more target time series data values (Reinforcement training and computing a reward value may be used in conjunction with time series data, P0045).
Accordingly, it would have been obvious to a person having ordinary skill in the
art before the effective filing date of the claimed invention, having the teachings of
Zhang, Tai, and Sumida before them, to include Sumida’s specific teachings of computing a reward value based on times series data in Zhang’s system of anomaly detection. One would have been motivated to make such a combination of computing a reward value based on times series data (see Sumida P0045), and collecting time series data used to detect anomalies (see Zhang P0019, P0023-P0024).
Zhang in view of Tai and Sumida does not appear to explicitly teach (vi) storing, by the computer system, a tuple including the state value, the action comprising the source window size and the target window size, the updated state value, and the reward value in a memory buffer of the context sampler;… comprising at least the tuple from the memory buffer of the context sampler.
Beard teaches (vi) storing, by the computer system, a tuple including the state value, … the source window size and the target window size, the updated state value, and the reward value in a memory buffer of the context sampler (Current and updated state values, reward values, and window sizes, and experience memory, which is a window of data, are stored in a tuple used to train an agent, P0057);… comprising at least the tuple from the memory buffer of the context sampler (Tuple stored in memory is used during training, P0057).
Accordingly, it would have been obvious to a person having ordinary skill in the
art before the effective filing date of the claimed invention, having the teachings of
Zhang, Tai, Sumida, and Beard before them, to include Beard’s specific teachings of storing values in a tuple to be used for training in Zhang’s system of anomaly detection. One would have been motivated to make such a combination of storing values in a tuple to be used for training (see Beard P0057), and analyzing training data stored in memory to detect anomalies (see Zhang P0013).
Zhang in view of Tai, Sumida, and Beard does not appear to explicitly teach a two-dimensional vector.
Nia teaches a two-dimensional action vector (a two-dimensional vector being processed by model 110 which is used for anomaly detection, P0068, P0037).
Accordingly, it would have been obvious to a person having ordinary skill in the
art before the effective filing date of the claimed invention, having the teachings of
Zhang, Tai, Sumida, Beard, and Nia before them, to include Nia’s specific teachings of a two-dimensional vector being processed by a model used for anomaly detection in Zhang’s system of anomaly detection. One would have been motivated to make such a combination of a two-dimensional vector being processed by a model used for anomaly detection (see Nia P0068, P0037), and analyzing training data stored in memory to detect anomalies (see Zhang P0013).
Regarding claim 2, Zhang in view of Tai, Sumida, Beard, and Nia teaches the limitations of claim 1 as outlined above. Zhang further teaches after step (vii), (viii) recording, by the computer system, each source window size, thereby recording a plurality of source window sizes proportional to the training epoch value (The length of the rolling window is tracked, with a different number of days D being used as the length of a window for different granularities of data, P0076); and after step d), e) determining, by the computer system, a frequently selected source window size based on the plurality of source window sizes, wherein the frequently selected source window size is used during an anomaly detection process (Different types of data have different frequently used window sizes of days D corresponding to the frequency of a type of data being received, P0076. These window sizes are used during an anomaly detection process, P0077, P0079).
Regarding claim 3, Zhang in view of Tai, Sumida, Beard, and Nia teaches the limitations of claim 1 as outlined above. Zhang further teaches inputting source data values and target data values into the trained context sampler, which outputs source time series data values and target time series data values to an anomaly detector, which determines one or more anomalies in the target time series data values (Time series data may be examined by anomaly detector 114 which determines anomalies being present in the time series data, P0017, P0069, P0074).
Regarding claim 4, Zhang in view of Tai, Sumida, Beard, and Nia teaches the limitations of claim 3 as outlined above. Tai further teaches wherein the reward value is computed using one or more loss values comprising a classification loss value, a reconstruction loss value, an alignment loss value, and a domain discrimination loss value, and wherein the method further comprises computing the one or more loss values (Model training includes classification loss (P0139), reconstruction loss (P0220), alignment loss (P0174), and domain discrimination loss (P0174)).
Regarding claim 5, Zhang in view of Tai, Sumida, and Beard teaches the limitations of claim 4 as outlined above. Tai further teaches wherein the reward value is computed using a classification hyper-parameter, a reconstruction hyper-parameter, an alignment hyper- parameter, and a domain discrimination hyper-parameter (Model training includes classification loss (P0139), reconstruction loss (P0220), alignment loss (P0174), and domain discrimination loss (P0174)).
Regarding claim 6, Zhang in view of Tai, Sumida, Beard, and Nia teaches the limitations of claim 4 as outlined above. Tai further teaches wherein the anomaly detector comprises an encoder, a decoder, a classifier, and a domain classifier, and wherein: the classification loss value is computed using the classifier, the encoder, the one or more source time series data values, one or more labels corresponding to the one or more source time series data values, and one or more weights corresponding to the one or more source time series data values (Computing classification loss includes a classifier (P0054), encoder (feature converter, P0046, P0049), labeled data (P0043-P0044), and weights (P0139)); the reconstruction loss value is calculated using the encoder, the decoder, the one or more source time series data values, and the one or more target time series data values (Computing reconstruction loss includes an encoder (feature conversion, P0195), decoder (decoding neural network, P0190), and source and target data (P0196)); the alignment loss value is calculated using the encoder, the one or more source time series data values, and the one or more target time series data values (Computing alignment loss includes an encoder (feature conversion, P0185, P0233), and source and target data (P0178)); and the domain discrimination loss value is calculated using the encoder, the domain classifier, the one or more source time series data values, the one or more target time series data values, and one or more domain labels (Computing domain discrimination includes an encoder (feature conversion, P0176), domain classifier (“target domains” are domains that a classifier targets, P0003, P0176), source and target data (P0176), and respective labels (P0176)).
Regarding claim 9, Zhang in view of Tai, Sumida, Beard, and Nia teaches the limitations of claim 1 as outlined above. Zhang further teaches the one or more initial source time series data values comprise a subsequence of a source data set, wherein a number of initial source time series data values in the subsequence of the source data set is proportional to the initial source window size (Initial window size is proportional to the type of data being received, with initial D days scaling with the source data, P0076); the one or more initial target time series data values comprise a subsequence of a target data set, wherein a number of initial source time series data values in the subsequence of the target data set is proportional to the initial target window size (Initial window size is proportional to the target number of data points to be received, with a default time frame being set which is dependent on N number of data points, P0039); the one or more source time series data values comprise a subsequence of the source data set, wherein a number of source time series data values in the subsequence of the source data set is proportional to the source window size (Window size may be adjusted to be proportional to the type of data being received, with D days scaling with the source data, P0076); and the one or more target time series data values comprise a subsequence of the target data set, wherein a number of target time series data values in the subsequence of the target data set is proportional to the target window size (Window size may be adjusted to be proportional to the target number of data points to be received, with a time frame being set which is dependent on N number of data points, P0039).
Regarding claim 12, Zhang in view of Tai, Sumida, Beard, and Nia teaches the limitations of claim 11 as outlined above. Zhang further teaches wherein the state value comprises combination of a source encoding generated using the encoder and the one or more source time series data values and a target encoding generated using the encoder and the one or more target time series data values (Data collection center 102 retrieves data from a data source periodically, the data being time series data, P0024. This source data may be analyzed by the predictive model, P0025. Predictive model may analyze a target number of data points consisting of time series data, P0052-P0053. Collected time series data is used to update the state of the prediction model, P0032).
Regarding claim 13, Zhang in view of Tai, Sumida, Beard, and Nia teaches the limitations of claim 11 as outlined above. Beard further teaches wherein the sampled data used to train the context sampler comprises the tuple and one or more previous tuples stored in the memory buffer (Current and updated state values, reward values, and window sizes, and experience memory, which is a window of data, are stored in a tuple used to train an agent. Tuple stored in memory is used during training. Previous data may also be stored, P0057).
Claims 7, 8, and 10 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang in view of Tai, Sumida, Beard, and Nia and further in view of Dev (Pub. No.: US 20240184282 A1), hereafter Dev.
Regarding claim 7, Zhang in view of Tai, Sumida, Beard, and Nia teaches the limitations of claim 6 as outlined above. Zhang does not appear to explicitly teach training, by the computer system, the anomaly detector by training the encoder, the decoder, the classifier, and the domain classifier.
Dev teaches training, by the computer system, the anomaly detector by training the encoder, the decoder, the classifier, and the domain classifier (Autoencoder includes an encoder and decoder, P0186, and is trained as described in P0189. Machine learning based classifier is trained as described in P0165. Machine learning model 300 is used to classify domains, P0088, and is trained as described in P0097).
Accordingly, it would have been obvious to a person having ordinary skill in the
art before the effective filing date of the claimed invention, having the teachings of
Zhang, Tai, Sumida, Beard, Nia, and Dev before them, to include Dev’s specific teachings of an autoencoder, classifier, and domain classifier in Zhang’s system of anomaly detection. One would have been motivated to make such a combination of an autoencoder, classifier, and domain classifier used for anomaly detection (see Dev P0186, P0189, P0088, P0097), and collecting data from different sources or domains to detect anomalies (see Zhang P0019, P0023).
Regarding claim 8, Zhang in view of Tai, Sumida, Beard, and Nia teaches the limitations of claim 6 as outlined above. Zhang does not appear to explicitly teach the classifier comprises a multi-layer perceptron classifier with a sigmoid activation function; the domain classifier comprises another multi-layer perceptron classifier with a sigmoid activation function; and the encoder and the decoder comprise two parts of an LSTM autoencoder.
Dev teaches the classifier comprises a multi-layer perceptron classifier with a sigmoid activation function (Machine learning based classifier uses a sigmoid activation function, P0164, P0184); the domain classifier comprises another multi-layer perceptron classifier with a sigmoid activation function (Machine learning model includes a sigmoid activation function, and is used to classify domains, P0081, P0088); and the encoder and the decoder comprise two parts of an LSTM autoencoder (Encoder and decoder are each parts of an LSTM autoencoder, P0186).
Accordingly, it would have been obvious to a person having ordinary skill in the
art before the effective filing date of the claimed invention, having the teachings of
Zhang, Tai, Sumida, Beard, Nia and Dev before them, to include Dev’s specific teachings of an autoencoder and sigmoid activation functions in Zhang’s system of anomaly detection. One would have been motivated to make such a combination of an autoencoder and sigmoid activation (see Dev P0164, P0184, P0186, P0088, P0081), and collecting data from different sources or domains to detect anomalies (see Zhang P0019, P0023).
Regarding claim 10, Zhang in view of Tai, Sumida, Beard, and Nia teaches the limitations of claim 9 as outlined above. Zhang does not appear to explicitly teach wherein a cyclical sampling process is used to sample the one or more initial source time series data values, the one or more initial target time series data values, the one or more source time series data values, and the one or more target time series data values.
Dev teaches wherein a cyclical sampling process is used to sample the one or more initial source time series data values, the one or more initial target time series data values, the one or more source time series data values, and the one or more target time series data values (Data may be sampled in cycles, with the purpose of identifying anomalies in cycles of sensor data, P0123, P0135).
Accordingly, it would have been obvious to a person having ordinary skill in the
art before the effective filing date of the claimed invention, having the teachings of
Zhang, Tai, Sumida, Beard, Nia and Dev before them, to include Dev’s specific teachings of sampling data in cycles in Zhang’s system of anomaly detection. One would have been motivated to make such a combination of sampling data in cycles (see Dev P0123, P0135), and collecting data from different sources or domains to detect anomalies (see Zhang P0019, P0023).
Claims 14 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang in view of Tai, Sumida, Beard, and Nia and further in view of Xaio et al (Pub. No.: CA 3153903 A1), hereafter Xaio.
Regarding claim 14, Zhang in view of Tai, Sumida, Beard, and Nia teaches the limitations of claim 1 as outlined above. Zhang does not appear to explicitly teach wherein training the context sampler comprises optimizing a Q-function associated with the context sampler.
Xaio teaches wherein training the context sampler comprises optimizing a Q-function associated with the context sampler (Q function may be used in a reinforcement learning algorithm associated with a data stream page 29 lines 5-17, page 30 lines 4-30).
Accordingly, it would have been obvious to a person having ordinary skill in the
art before the effective filing date of the claimed invention, having the teachings of
Zhang, Tai, Sumida, Beard, Nia and Xaio before them, to include Xaio’s specific teachings of implementing a Q function in an algorithm related to a data stream. One would have been motivated to make such a combination of implementing a Q function in an algorithm related to a data stream (see Xaio page 29 lines 5-17, page 30 lines 4-30), and collecting data from different sources or domains to detect anomalies (see Zhang P0019, P0023).
Regarding claim 15, Zhang in view of Tai, Sumida, Beard, and Nia and further in view of Xaio teaches the limitations of claim 14 as outlined above. Nia further teaches one or more two-dimensional action vectors (a two-dimensional vector being processed by model 110 which is used for anomaly detection, P0068, P0037). Xaio further teaches optimizing the Q-function improves a quality of ... comprising one or more source window sizes and one or more target window sizes generated by the context sampler, which in turn improves a quality of one or more encodings, one or more reconstructions, one or more classifications, and one or more domain classifications generated by the anomaly detector, thereby reducing one or more loss values computed by the anomaly detector, thereby increasing the reward value (Q-function is designed to progressively increase the reward and decrease error with each iteration of training, page 30 lines 4-30).
Claims 16 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang in view of Tai.
Regarding claim 16, Zhang teaches obtaining, by a computer system, a source data set and a target data set, wherein the source data set comprises the plurality of source time series data values, and wherein the target data set comprises a plurality of target time series data values (Data collection center 102 retrieves data from a data source periodically, the data being time series data, P0024. Predictive model may analyze a target number of data points consisting of time series data, P0052-P0053),… setting, by the computer system, a source window size and a target window size using a trained context sampler (initial window size is proportional to the type of data being received, with initial D days scaling with the source data, P0076. Initial window size is proportional to the target number of data points to be received, with a default time frame being set which is dependent on N number of data points, P0039); sampling, by the computer system, one or more source time series data values from the source data set using the source window size and based on a time value (data collection center 102 may obtain data in real time as the data is generated by a data source, P0024. source time series data is collected based on the rolling window and time, P0074-P0076); sampling, by the computer system, one or more target time series data values from the target data set using the target window size and based on the time value (Data collection center 102 may obtain a target number of data points, P0039. A target number of time series data points may be obtained based on a rolling window and time, P0039, P0074-P0076); providing, by the computer system, the one or more source time series data values and the one or more target time series data values to an anomaly detector (Anomaly detector 114 receives datasets from training data manager 110 P0065); and determining, by the computer system, using the anomaly detector, whether a target data value in the one or more target time series data values comprises an anomalous data value (If an observed value is above an anomaly threshold, anomaly detector 114 may determine the observed value to be an anomalous data value, P0060).
Zhang does not appear to explicitly teach wherein the plurality of source time series data values are labeled and the plurality of target time series data values are unlabeled.
Tai teaches wherein the plurality of source time series data values are labeled and the plurality of target time series data values are unlabeled (Source domain feature data is collected, P0069. Target domain feature data is collected, P0076. Data from source domain is labeled compared to data from target domain which has data without labels, P0054).
Accordingly, it would have been obvious to a person having ordinary skill in the
art before the effective filing date of the claimed invention, having the teachings of
Zhang and Tai before them, to include Tai’s specific teachings of collecting data from a source and target domain in Zhang’s system of anomaly detection. One would have been motivated to make such a combination of collecting data from a source and target domain (see Tai P0069, P0076), and collecting data from different sources or domains to detect anomalies (see Zhang P0019, P0023).
Regarding claim 17, Zhang in view of Tai teaches the limitations of claim 16 as outlined above. Tai further teaches wherein determining, using the anomaly detector, whether the target data value comprises the anomalous data value comprises: generating, by the computer system, a target encoding using an encoder and the target data value (Encoder (feature converter, P0046, P0049) may be used to encode data from the target domain, P0053); generating, by the computer system, a target classification using a classifier and the target encoding (Computing classification loss includes a classifier (P0054), encoder (feature converter, P0046, P0049), and labeled data (P0043-P0044). Classification of the target domain may be performed using a classifier and the encoded data from the target domain, P0053-P0054); generating, by the computer system, a target reconstruction using a decoder and the target encoding (decoding neural network may be used to reconstruct images based on converted features from the target domain P0190, P0194); generating, by the computer system, a reconstruction loss value using the target data value and the target reconstruction (“The auxiliary loss computation section 322 calculates differences between the target domain reconstructed images and the auxiliary label data to obtain a target domain reconstruction loss”, P0196);.
Zhang further teaches generating, by the computer system, an anomaly score using the target classification and the reconstruction loss value (An observed value is compared to an expected value to determine if an anomaly is detected. The difference between an observed value and an expected value is a loss value. The result is a value compared to a threshold to determine if an anomaly is present in the data P0059-P0060.); and comparing, by the computer system, the anomaly score to an anomaly score threshold, wherein if the anomaly score is greater than the anomaly score threshold, the computer system determines that the target data value comprises the anomalous data value (If an observed value is above an anomaly threshold, anomaly detector 114 may determine the observed value to be an anomalous data value, P0060).
Claim 18 is rejected under 35 U.S.C. 103 as being unpatentable over Zhang in view of Tai and further in view of Sumida, Beard, and Nia.
Regarding claim 18, Zhang in view of Tai teaches the limitations of claim 16 as outlined above. Zhang further teaches a) generating, by a computer system, an initial source window size and an initial target window size (Initial window size is set to two weeks of collected data, this window may be adjusted, P0079. “Such a default amount may be customized, for example, by users based on the nature of the data (e.g., 14 day time frame associated with throughput data collected every 5 minutes). As another example, a target number of data points may be obtained. A target number of data points N may be determined based on a minimal time coverage and data frequency. ”, P0039. Different sources/domains of data may have a different associated window size, as different sources of data may be sampled with data collection center 102, P0039, P0023); b) sampling, by the computer system, one or more initial … time series data values and one or more initial … time series data values using the initial source window size and the initial target window size (Data collection center 102 is used to sample time series data, as well as collect a target number of data points P0024, P0039. Time series data is collected as shown in figures 3A-3F, with an initial source window size as described in P0079. Data may be collected from different sources or domains, P0023); c) generating, by the computer system, a state value using the one or more initial source time series data values and the one or more initial target time series data values (Time series data may be used by a prediction model to perform predictions. State of the model is dependent upon sampled time series data, P0012, P0017, P0032); and d) for each time value up to a training epoch value, performing at least the following steps: (i) generating, by the computer system, using a context sampler,…comprising a source window size and a target window size based on the state value (“a rolling window of up to a particular number of days D of latest data may be stored. The specific number of days used as a basis for the rolling window may depend on the granularity of the data… Generally, the number of days D may be reduced for data with higher frequencies as they already provide a larger amount of data points within a shorter period of time.”, P0076. Initial window size is proportional to the target number of data points to be received, with a default time frame being set which is dependent on N number of data points, P0039); (ii) sampling, by the computer system, one or more source time series data values based on the source window size and the time value (source time series data is collected based on the rolling window and time, P0074-P0076); (iii) sampling, by the computer system, one or more target time series data values based on the target window size and the time value (A target number of time series data points may be obtained based on a rolling window and time, P0039, P0074-P0076); (iv) updating, by the computer system, the state value to an updated state value using the one or more source time series data values and the one or more target time series data values (Collected time series data is used to update the state of the prediction model, P0032);… and (vii) training, by the computer system, the context sampler in an iterative process using sampled data (Data store 118, which captures sample data, may be updated in an iterative process, P0074, P0078).
Tai further teaches sampling… one or more initial source values…target values (Source domain feature data is collected, P0069. Target domain feature data is collected, P0076).
Zhang in view of Tai does not appear to explicitly teach (v) computing, by the computer system, a reward value using the one or more source time series data values, and the one or more target time series data values.
Sumida teaches (v) computing, by the computer system, a reward value using the one or more source time series data values, and the one or more target time series data values (Reinforcement training and computing a reward value may be used in conjunction with time series data, P0045).
Accordingly, it would have been obvious to a person having ordinary skill in the
art before the effective filing date of the claimed invention, having the teachings of
Zhang, Tai, and Sumida before them, to include Sumida’s specific teachings of computing a reward value based on times series data in Zhang’s system of anomaly detection. One would have been motivated to make such a combination of computing a reward value based on times series data (see Sumida P0045), and collecting time series data used to detect anomalies (see Zhang P0019, P0023-P0024).
Zhang in view of Tai and Sumida does not appear to explicitly teach (vi) storing, by the computer system, a tuple including the state value, the action comprising the source window size and the target window size, the updated state value, and the reward value in a memory buffer of the context sampler;… comprising at least the tuple from the memory buffer of the context sampler.
Beard teaches (vi) storing, by the computer system, a tuple including the state value, …comprising the source window size and the target window size, the updated state value, and the reward value in a memory buffer of the context sampler (Current and updated state values, reward values, and window sizes, and experience memory, which is a window of data, are stored in a tuple used to train an agent, P0057);… comprising at least the tuple from the memory buffer of the context sampler (Tuple stored in memory is used during training, P0057).
Accordingly, it would have been obvious to a person having ordinary skill in the
art before the effective filing date of the claimed invention, having the teachings of
Zhang, Tai, Sumida, and Beard before them, to include Beard’s specific teachings of storing values in a tuple to be used for training in Zhang’s system of anomaly detection. One would have been motivated to make such a combination of storing values in a tuple to be used for training (see Beard P0057), and analyzing training data stored in memory to detect anomalies (see Zhang P0013).
Zhang in view of Tai, Sumida, and Beard does not appear to explicitly teach a two-dimensional vector.
Nia teaches a two-dimensional action vector (a two-dimensional vector being processed by model 110 which is used for anomaly detection, P0068, P0037).
Accordingly, it would have been obvious to a person having ordinary skill in the
art before the effective filing date of the claimed invention, having the teachings of
Zhang, Tai, Sumida, Beard, and Nia before them, to include Nia’s specific teachings of a two-dimensional vector being processed by a model used for anomaly detection in Zhang’s system of anomaly detection. One would have been motivated to make such a combination of a two-dimensional vector being processed by a model used for anomaly detection (see Nia P0068, P0037), and analyzing training data stored in memory to detect anomalies (see Zhang P0013).
Claims 19 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang in view of Tai and further in view of Nia.
Regarding claim 19, Zhang in view of Tai teaches the limitations of claim 16 as outlined above. Zhang further teaches generating, by the computer system, a state value using an encoder, the one or more source time series data values and the one or more target time series data values (State of the prediction model is updated based on sampled time series data, P0032); and generating, by the computer system, … comprising a subsequent source window size and a subsequent target window size using a context sampler and the state value, wherein the subsequent source window size and subsequent target window size are used to determine whether a subsequent target time series data value corresponding to a subsequent time value comprises a subsequent anomalous data value (Different sources/domains of data may have a different associated window size, as different sources of data may be sampled with data collection center 102, P0039, P0023. Adjustable window size is used in conjunction with anomaly detector 114 to analyze time series data from different domains, P0030, P0032, P0039).
Zhang in view of Tai, Sumida, and Beard does not appear to explicitly teach a two-dimensional vector.
Nia teaches a two-dimensional action vector (a two-dimensional vector being processed by model 110 which is used for anomaly detection, P0068, P0037).
Accordingly, it would have been obvious to a person having ordinary skill in the
art before the effective filing date of the claimed invention, having the teachings of
Zhang, Tai, Sumida, Beard, and Nia before them, to include Nia’s specific teachings of a two-dimensional vector being processed by a model used for anomaly detection in Zhang’s system of anomaly detection. One would have been motivated to make such a combination of a two-dimensional vector being processed by a model used for anomaly detection (see Nia P0068, P0037), and analyzing training data stored in memory to detect anomalies (see Zhang P0013).
Regarding claim 20, Zhang in view of Tai and further in view of Nia teaches the limitations of claim 19 as outlined above. Zhang further teaches wherein the subsequent source window size is equal to the initial source window size, and wherein the two-dimensional action vector comprising the subsequent source window size and the subsequent target window size are generated by inputting the state value into a trained policy function associated with the context sampler (Different types of data have different frequently used window sizes of days D corresponding to the frequency of a type of data being received. The different types of data correspond to their domain/source, meaning same source will have the same window, P0076. Control parameters may be used to determine window size based upon identified outliers, P0041, P0042, P0077, P0079).
Conclusion
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 nonprovisional extension fee (37 CFR 1.17(a)) 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 mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ISHAN MOUNDI whose telephone number is (703)756-1547. The examiner can normally be reached 8:30 A.M. - 5 P.M..
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, Matthew Ell can be reached at (571) 270-3264. 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.
/I.M./Examiner, Art Unit 2141
/MATTHEW ELL/Supervisory Patent Examiner, Art Unit 2141