DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
This action is in response to amendments filed February 9th, 2026. The status of the claims is as follows. Claims 1 and 9 are amended. Claims 1-16 are currently pending.
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 1-16 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding Claim 1,
(Step 1): Claim 1 recites A system for constructing a probability model and automatically responding to process anomalies identified by the probability model, thus a machine, one of the four statutory categories of patentable subject matter.
(Step 2A Prong 1): However, Claim 1 further recites to
calculate a Cartesian product of all probability vectors that were output to obtain a tensor of the two or more dimensions which falls in the mathematical concept and mental process grouping of abstract ideas,
determine whether a probability in the tensor associated with the current state is less than a predetermined threshold which constitutes the judgement of a tensor probability against a predetermined threshold value, thus corresponding to a mental process which can be done mentally or by pen and paper
in response to determining that the probability is less than the predetermined threshold … generate a … communication which constitutes the evaluation of the probability comparison result to determine a communication configuration to generate, thus corresponding to a mental process which can be done mentally or by pen and paper
Thus, Claim 1 recites an abstract idea.
(Step 2A Prong 2): The claim does not recite any additional elements which integrate the abstract idea into a practical application because the additional elements consist of:
a central server in communication with one or more sensor devices; a client computing device communicatively coupled to the central server; and non-transitory memory storing instructions that, when executed by one or more processors of the central server or of the client computing device, cause the one or more processors to, which is implementing an abstract idea on generic computer components (MPEP 2106.05(f))
receive data from the one or more sensor devices comprising variables in at least two dimensions which is insignificant extra-solution activity of data gathering (MPEP 2106.05(g))
the variables representing a current state of a running process, and the at least two dimensions being not independently and identically distributed which merely recites the particular technological environment or field of use in which the abstract idea is to be performed (MPEP 2106.05(h))
receive or retrieve a set of variables representing states of the running process, previous to the current state which is insignificant extra-solution activity of data gathering (MPEP 2106.05(g))
select a segment of a fixed number of prior states from the set of variables which is insignificant extra-solution activity of data gathering (MPEP 2106.05(g))
feed the segment to a neural network which is insignificant extra-solution activity of data inputting (MPEP 2106.05(g))
to output a probability vector for each of the two or more dimensions which is insignificant extra-solution activity of data outputting (MPEP 2106.05(g))
wherein each value in the tensor represents a probability that the prior states would be followed by a state associated with that value, which merely recites the particular technological environment or field of use in which the abstract idea is to be performed (MPEP 2106.05(h))
automatically generate an electronic communication, which is implementing an abstract idea on generic computer components (MPEP 2106.05(f))
transmit it to the client computing device, which is insignificant extra-solution activity of data outputting (MPEP 2106.05(g))
to cause the client computing device to take prompt mitigating action, which is implementing an abstract idea on generic computer components (MPEP 2106.05(f))
during runtime:, which merely recites the particular technological environment or field of use in which the abstract idea is to be performed (MPEP 2106.05(h))
and thus, the claim is directed to the abstract idea of determining a tensor by calculating the Cartesian product of probability vectors and generating a communication if the tensor is determined to exceed a threshold.
(Step 2B) The additional elements, taken alone or in combination, cannot provide significantly
more than the abstract idea itself because elements a), i) and k) ((via MPEP 2106.05(f), “apply it on a computer”) cannot provide an inventive concept, elements b), d), e), f), g), j) are further well-understood, routine, and conventional activity of “receiving or transmitting data over a network” by MPEP 2106.05(d), which cannot provide significantly more than the abstract idea itself, and elements c) and h), l) (via MPEP 2106.05(h)) cannot integrate the abstract idea into a practical application or provide significantly more than the abstract idea itself. Thus, Claim 1 is subject-matter ineligible.
Claim 2, dependent upon Claim 1 recites the additional elements:
a) wherein the at least two dimensions comprise a first dimension related to an operation and a second dimension related to state within the operation which merely recites the particular technological environment or field of use in which the abstract idea is to be performed (MPEP 2106.05(h))
The additional elements, taken alone or in combination, cannot provide significantly
more than the abstract idea itself because element a) (via MPEP 2106.05(h)) cannot integrate the abstract idea into a practical application or provide significantly more than the abstract idea itself. Thus, Claim 2 is subject-matter ineligible.
Claim 3, dependent upon Claim 1 recites the additional element:
a) wherein the at least two dimensions comprise three or more dimensions that are not hierarchically related which merely recites the particular technological environment or field of use in which the abstract idea is to be performed (MPEP 2106.05(h))
The additional elements, taken alone or in combination, cannot provide significantly
more than the abstract idea itself because element a) (via MPEP 2106.05(h)) cannot integrate the abstract idea into a practical application or provide significantly more than the abstract idea itself. Thus, Claim 3 is subject-matter ineligible.
Claim 4, dependent upon Claim 1 recites the additional element:
a) wherein the one or more sensor devices generate sensor readings on a continuous scale, and the received data comprises a conversion of those sensor readings to one of a set of predetermined discrete values which merely recites the particular technological environment or field of use in which the abstract idea is to be performed (MPEP 2106.05(h))
The additional elements, taken alone or in combination, cannot provide significantly
more than the abstract idea itself because element a) (via MPEP 2106.05(h)) cannot integrate the abstract idea into a practical application or provide significantly more than the abstract idea itself. Thus, Claim 4 is subject-matter ineligible.
Claim 5, dependent upon Claim 1 recites the additional element:
a) wherein multiple neural networks … are each used to generate output probability tensors which is implementing an abstract idea on generic computer components (MPEP 2106.05(f))
b) each trained on segments of a fixed length different from a fixed length on which each other neural network was trained which merely recites the particular technological environment or field of use in which the abstract idea is to be performed (MPEP 2106.05(h))
The additional elements, taken alone or in combination, cannot provide significantly
more than the abstract idea itself because element a) ((via MPEP 2106.05(f), “apply it on a computer”) cannot provide an inventive concept and element b) (via MPEP 2106.05(h)) cannot integrate the abstract idea into a practical application or provide significantly more than the abstract idea itself. Thus, Claim 5 is subject-matter ineligible.
Claim 6, dependent upon Claim 1 recites additional mathematical concept steps of the abstract idea (Claim 6: probability of anomaly is computed as a function of each of the output probability tensors' values for the current state, which falls in the mathematical concept and mental process grouping of abstract ideas). The claim does not recite any additional elements which integrate the abstract idea into a practical application nor provide significantly more than the abstract idea itself because the additional elements consist of:
wherein multiple neural networks are each used to generate output probability tensors which is implementing an abstract idea on generic computer components (MPEP 2106.05(f))
The additional elements, taken alone or in combination, cannot provide significantly more than the abstract idea itself, because element a) ((via MPEP 2106.05(f), “apply it on a computer”) cannot provide an inventive concept. Thus, Claim 6 is subject-matter ineligible.
Claim 7, dependent upon Claim 1 recites the additional elements:
a) wherein the client computing device … automatically activates which is implementing an abstract idea on generic computer components (MPEP 2106.05(f))
b) in response to receiving the electronic communication which merely recites the particular technological environment or field of use in which the abstract idea is to be performed (MPEP 2106.05(h))
c) activates a functionality of the client computing device which is insignificant extra-solution activity of data outputting (MPEP 2106.05(g))
d) a functionality of the client computing device to mitigate an expected harm to the client computing device or to a human user of the client computing device which merely recites the particular technological environment or field of use in which the abstract idea is to be performed (MPEP 2106.05(h))
The additional elements, taken alone or in combination, cannot provide significantly
more than the abstract idea itself because element a) ((via MPEP 2106.05(f), “apply it on a computer”) cannot provide an inventive concept, elements b) and d) (via MPEP 2106.05(h)) cannot integrate the abstract idea into a practical application or provide significantly more than the abstract idea itself, and element c) is further well-understood, routine, and conventional activity of “receiving or transmitting data over a network” by MPEP 2106.05(d), which cannot provide significantly more than the abstract idea itself. Thus, Claim 7 is subject-matter ineligible.
Claim 8, dependent upon Claim 1 recites the additional elements:
a) wherein the client computing device … automatically deactivates which is implementing an abstract idea on generic computer components (MPEP 2106.05(f))
b) in response to receiving the electronic communication which merely recites the particular technological environment or field of use in which the abstract idea is to be performed (MPEP 2106.05(h))
c) deactivates a functionality of the client computing device which is insignificant extra-solution activity of data outputting (MPEP 2106.05(g))
d) a functionality of the client computing device to mitigate an expected harm to the client computing device or to a human user of the client computing device which merely recites the particular technological environment or field of use in which the abstract idea is to be performed (MPEP 2106.05(h))
The additional elements, taken alone or in combination, cannot provide significantly
more than the abstract idea itself because element a) ((via MPEP 2106.05(f), “apply it on a computer”) cannot provide an inventive concept, elements b) and d) (via MPEP 2106.05(h)) cannot integrate the abstract idea into a practical application or provide significantly more than the abstract idea itself, and element c) is further well-understood, routine, and conventional activity of “receiving or transmitting data over a network” by MPEP 2106.05(d), which cannot provide significantly more than the abstract idea itself. Thus, Claim 8 is subject-matter ineligible.
Claims 9-16 recites the method performed by the system of Claims 1-8. As performance of an abstract idea on generic computing components cannot integrate an abstract idea into a practical application nor provide significantly more than the abstract idea itself (see MPEP 2106.05(f)), Claims 9-16 are rejected for reasons set forth in the rejection of Claims 1-8.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1-4, 7-8; 9-12, 15-16 are rejected under 35 U.S.C. 103 as being unpatentable over Sha et al. (“A Multi-Order Markov Chain Based Scheme for Anomaly Detection” [2013], hereinafter “Sha”) in view of Bremaud (“Non-Homogenous Markov Chains” [2020]) in view of Kou (TWI737816B) further in view of Bai et al. (“Modeling Neural Networks Training Process with Markov Decision Process” [2021], hereinafter “Bai”).
Regarding Claim 1,
Sha discloses A system for constructing a probability model and automatically responding to process anomalies identified by the probability model (Sha [Section II Subsection A]; “Therefore, to compensate the uncertainty in anomaly detection, results from multiple indicators or even multiple models are desired to issue a real-time warning for detrimental operations. In this paper, we address the above question by deploying the multi-order Markov chain based model”
Sha [Section II Subsection B2];
PNG
media_image1.png
173
339
media_image1.png
Greyscale
Wherein the Markov chain based model is a probability model)
during runtime: (Sha [Section II Subsection A]; “Therefore, to compensate the uncertainty in anomaly detection, results from multiple indicators or even multiple models are desired to issue a real-time warning for detrimental operations. In this paper, we address the above question by deploying the multi-order Markov chain based model.” wherein the multi-order Markov chain model is deployed in a real-time scenario for anomaly detection, thus reading on such a model and its operations occurring during runtime)
receive data … comprising variables in at least two dimensions (Sha [Table V];
PNG
media_image2.png
98
236
media_image2.png
Greyscale
)
the variables representing a current state of a running process, and the at least two dimensions being not independently and identically distributed (Sha [Section II Subsection C];
PNG
media_image3.png
267
332
media_image3.png
Greyscale
)
receive or retrieve a set of variables representing states of the running process, previous to the current state (Sha [Section II Subsection C];
PNG
media_image3.png
267
332
media_image3.png
Greyscale
)
select a segment of a fixed number of prior states from the set of variables (Sha [Section II Subsection C1];
PNG
media_image4.png
191
333
media_image4.png
Greyscale
)
feed the segment … to output a probability vector for each of the two or more dimensions (Sha [Section II Subsection C1];
PNG
media_image5.png
404
346
media_image5.png
Greyscale
Wherein the transmission matrix calculated for the inputted high-order sequence Xn reads on feeding the segment to output a probability vector
Sha [Section II Subsection C2];
PNG
media_image6.png
374
340
media_image6.png
Greyscale
Wherein the Markov chain model is applicable for multi-dimensional input)
calculate a Cartesian product of all probability vectors that were output to obtain a tensor of the two or more dimensions, wherein each value in the tensor represents a probability that the prior states would be followed by a state associated with that value (Sha [Section III];
PNG
media_image7.png
446
367
media_image7.png
Greyscale
PNG
media_image8.png
814
354
media_image8.png
Greyscale
Wherein the state spaces associated with their individual dimensions used to obtain a transition matrix representing a sequence of state transitions is performed, thus reading on a calculated cartesian product in derivation of the transition matrix)
Sha fails to explicitly disclose but Bremaud discloses during runtime: … calculate a Cartesian product of all probability vectors that were output to obtain a tensor of the two or more dimensions, wherein each value in the tensor represents a probability that the prior states would be followed by a state associated with the graph. (Bremaud [Section 12.1.1];
PNG
media_image9.png
310
480
media_image9.png
Greyscale
wherein the non-homogenous transition matrix at real-time variable n thus reads on a dynamic transition matrix calculating cartesian products at runtime)
It would have been obvious to replace the Cartesian product calculated transition matrix calculated statically during Sha’s training to instead be Bremaud’s dynamically updated non-homogenous transition matrix. One would have been motivated to do so because “For non-homogeneous Markov chains (nhmc), the Markov property is retained but the transition probabilities may depend on time.” (Bremaud [Section 12.1]) thus allowing Sha’s determination of transition matrices to be time-considerate.
Sha/Bremaud fails to disclose but Kou discloses a central server in communication with one or more sensor devices (Kou [Page 4 Line 18]; “The abnormality detection device 1 and the remote server 3 are connected in a manner capable of communicating via the network 2. The type of the connected network 2 is not particularly limited, and it can be any network such as the Internet, wide area network, and local area network. In addition, it can be any one of a wireless network and a wired network, or a combination thereof. The abnormality detection device 1 is connected to a remote server 3 that always collects observation values observed in the semiconductor manufacturing device 4 via a network 2 to realize online monitoring of the semiconductor manufacturing device 3 on-line”
Kou [Page 4 Line 11]; “The remote server 3 is connected to a semiconductor manufacturing device 4 that is a monitoring target device that is a target of abnormality detection. The semiconductor manufacturing apparatus 4 is provided with any number of sensors, and each time the manufacturing process in the semiconductor manufacturing apparatus 4 is executed, specific parameters are measured.”)
a client computing device communicatively coupled to the central server (Kou [Page 4 Line 25]; “In addition, the communication unit 10 sends the information generated in the abnormality detection device 1 to the remote server 3 under the control of the control unit 20. The control unit 20 controls the operation and function of the abnormality detection device 1 … The control unit 20 has an observation value acquisition unit 201, a summary value generation unit 202, a selection unit 203, a first predictive value generation unit 204, a second predictive value generation unit 205, an abnormal score calculation unit 206, a change score calculation unit 207, and a detection unit 208. The warning unit 209, and the abnormality report creation unit 210.”)
non-transitory memory storing instructions that, when executed by one or more processors of the central server or of the client computing device, cause the one or more processors to (Kou [Page 4 Line 29]; “For the memory portion 30, any semiconductor memory device or the like can be used. For example, RAM (Random Access Memory), ROM (Read Only Memory), etc. can be used as the memory unit 30. In addition, hard disks, optical disks, etc. can also be used as the memory unit 30.”)
determine whether a probability in the tensor associated with the current state is less than a predetermined threshold (Kou [Page 5 Line 62]; “For example, the detection unit 208 determines whether the abnormality score calculated by the abnormality score calculation unit 206 exceeds a threshold value. In addition, the detection unit 208 determines whether the change score calculated by the change score calculation unit 207 exceeds a threshold value. Then, the detection unit 208 notifies the warning unit 209 when it is determined that any one of the abnormality score and the change score exceeds the threshold value.” wherein the determination of whether the abnormality score is above a threshold implicitly reads on determination of whether the same abnormality score is below said threshold since such a determination is binary)
and in response to determining that the probability is less than the predetermined threshold, automatically generate an electronic communication and transmit it to the client computing device to take prompt mitigating action (Kou [Page 5 Line 62]; “For example, the detection unit 208 determines whether the abnormality score calculated by the abnormality score calculation unit 206 exceeds a threshold value. In addition, the detection unit 208 determines whether the change score calculated by the change score calculation unit 207 exceeds a threshold value. Then, the detection unit 208 notifies the warning unit 209 when it is determined that any one of the abnormality score and the change score exceeds the threshold value.”)
Sha/Bremaud discloses to receive data … comprising variables in at least two dimensions. Sha/Bremaud does not disclose to receive data from the one or more sensor devices comprising variables in at least two dimensions. Kou, however, discloses one or more sensor devices. By using the sensor devices of Kou to transmit data for the method of Sha/Bremaud, the combination reads to receive data from the one or more sensor devices comprising variables in at least two dimensions.
It would have been obvious to modify Sha/Bremaud’s models that determine probability vectors for present and past state spaces for to be performed in Kou’s system comparing sensor-data anomaly probabilities to a threshold and issuing a communication depending on the threshold being met. One would have been motivated to do so because “if the abnormal score and the change score are used together to prompt the user to pay attention when any one of them deviates, and a warning is issued when the two deviate, then "Attention" can be issued” (Kou [Page 7 Line 42]) and user awareness of the issue will heighten accordingly.
The combination of Sha/Bremaud/Kou fails to disclose but Bai discloses to Feed the segment to a neural network to output a probability (Bai [Section III];
PNG
media_image10.png
369
355
media_image10.png
Greyscale
PNG
media_image11.png
51
331
media_image11.png
Greyscale
PNG
media_image12.png
697
333
media_image12.png
Greyscale
Wherein inputting sample set sequences into the neural network to compute a transition matrix representative of state probability distributions in a R2 two-dimensional space is performed)
Sha/Bremaud/Kou discloses to feed the segment … to output a probability vector for each of the two or more dimensions. Sha/Bremaud/Kou does not disclose to feed the segment to a neural network to output a probability vector for each of the two or more dimensions. Bai, however, discloses a neural network by which to feed the segment for probability vector output. By using the neural network of Bai in the method of Sha/Bremaud/Kou, the combination of Sha/Bremaud/Kou/Bai reads to feed the segment to a neural network to output a probability vector for each of the two or more dimensions.
It would have been obvious to modify Sha/Bremaud/Kou’s method of determining probability vectors present and past state spaces and issuing a communication depending on the probability meeting a threshold by using Bai’s neural networks implementing a Markov Decision Process instead of Sha/Bremaud/Kou’s Hidden Markov Models for obtaining the output probability distributions. One would have been motivated to do so because “we can optimize the training process by judging the current parameter state of the neural networks. And based on it, we can reduce the number of training steps.“ (Bai [Section III Subsection F Paragraph 4]).
Regarding Claim 2,
The combination of Sha/Bremaud/Kou/Bai teaches the method of Claim 1 (and thus the rejection of Claim 1 is incorporated). The combination already discloses wherein the at least two dimensions comprise a first dimension related to an operation and a second dimension related to state within the operation (Sha [Section II Subsection C1];
PNG
media_image13.png
108
353
media_image13.png
Greyscale
Wherein each of the dimensions comprise probabilities associated with the state values, thus reading on a first dimension related to an operation as well as a second dimension related to state within the operation)
Regarding Claim 3,
The combination of Sha/Bremaud/Kou/Bai teaches the method of Claim 1 (and thus the rejection of Claim 1 is incorporated). The combination already discloses wherein the at least two dimensions comprise three or more dimensions that are not hierarchically related (Sha [Section III];
PNG
media_image14.png
176
341
media_image14.png
Greyscale
Wherein both dimensions A and B are comprised of three or more non-hierarchical dimensions including time, state, and state probability)
Regarding Claim 4,
The combination of Sha/Bremaud/Kou/Bai teaches the method of Claim 1 (and thus the rejection of Claim 1 is incorporated). The combination already discloses wherein the one or more sensor devices generate sensor readings on a continuous scale, and the received data comprises a conversion of those sensor readings to one of a set of predetermined discrete values (Kou [Page 4 Line 22]; “The abnormality detection device 1 is connected to a remote server 3 that always collects observation values observed in the semiconductor manufacturing device 4 via a network 2 to realize online monitoring of the semiconductor manufacturing device 3 on-line. Therefore, the abnormality detection device 1 can immediately detect the abnormality of the semiconductor manufacturing device 3 and notify the user” wherein the observation values are collected continuously
Kou [Page 3 Line 2]; “predictive value generation unit applies statistical modeling to the summary value obtained by summarizing the observation values obtained at a specific time sequence in the processing that will be repeatedly executed in the monitoring target device and becoming an indicator of the operating state of the monitoring target device” wherein the continuously collected observation values are summarized at specific time sequences, reading on predetermined discrete values)
Regarding Claim 7,
The combination of Sha/Bremaud/Kou/Bai teaches the method of Claim 1 (and thus the rejection of Claim 1 is incorporated). The combination does not explicitly disclose but Kou further discloses wherein the client computing device, in response to receiving the electronic communication, automatically activates a functionality of the client computing device to mitigate an expected harm to the client computing device or to a human user of the client computing device (Kou [Page 5 Line 73]; “For example, the detection unit 208 notifies the warning unit 209 of the first-level abnormality when any one of the two abnormal scores or the change score exceeds the threshold … The warning unit 209 transmits, for example, a warning that can recognize each situation in which the detection unit 208 notifies the first-level abnormality, the second-level abnormality, and the third-level abnormality. Based on the information stored in the memory unit 30, the abnormality report preparation unit 210 generates an abnormality report that collects the results of the abnormality detection processing in the abnormality detection device 1. The anomaly report created by the anomaly report creating unit 210 is sent to the remote server 3 via the communication unit 10. In addition, the anomaly report created by the anomaly report creating unit 210 is output from the output unit 40.” wherein the warning unit, upon receiving notification communication from the detection unit, activates the abnormality report preparation unit in the client control unit reading on mitigation of expected harm to the client computing device)
It would have been obvious to modify the combination of Sha/Bremaud/Kou/Bai’s method of determining probability vectors for present and past state spaces and issuing a communication depending on the probability meeting a threshold by incorporating Kou’s method of, upon receival of an electronic communication, activating a functionality of the control unit to automatically mitigate harm to the client device. One would have been motivated to do so because “when a large-scale abnormality detection service is to be provided to a plurality of semiconductor manufacturing devices, for example, using cloud computing, etc., manually adjusting the threshold for each device requires a lot of labor, which is not realistic” (Kou [Page 2 “Problem to be Solved by the Invention” Section, Line 13])
Regarding Claim 8,
The combination of Sha/Bremaud/Kou/Bai teaches the method of Claim 1 (and thus the rejection of Claim 1 is incorporated). The combination does not explicitly disclose but Kou further discloses wherein the client computing device, in response to receiving the electronic communication, automatically deactivates a functionality of the client computing device to mitigate an expected harm to the client computing device or to a human user of the client computing device (Kou [Page 7 Line 60]; “When the detection unit 208 determines that the score exceeds the threshold, that is, an abnormality is detected (step S9, Yes), the warning unit 209 is notified, and the warning unit 209 sends a warning to the remote server 3. In addition, the abnormality report creation unit 210 outputs an abnormality report (step S10). In addition, when the detection unit 208 determines that the score is less than or equal to the threshold, that is, when no abnormality is detected (step S9, No), the process returns to step S1. In this way, the abnormality detection processing ends.”)
It would have been obvious to modify the combination of Sha/Bremaud/Kou/Bai’s method of determining probability vectors for present and past state spaces and issuing a communication depending on the probability meeting a threshold by incorporating Kou’s method of, upon receival of an electronic communication, deactivating a functionality of the control unit to automatically mitigate harm to the client device. One would have been motivated to do so because “when a large-scale abnormality detection service is to be provided to a plurality of semiconductor manufacturing devices, for example, using cloud computing, etc., manually adjusting the threshold for each device requires a lot of labor, which is not realistic” (Kou [Page 2 “Problem to be Solved by the Invention” Section, Line 13])
Claims 9-12, 15 and 16 recite the method performed by the system of Claims 1-4, 7 and 8. Thus, Claims 9-12, 15 and 16 are rejected for reasons set forth in the rejection of Claims 1-4, 7 and 8.
Claims 5 and 13 are rejected under 35 U.S.C. 103 as being unpatentable over Sha et al. (“A Multi-Order Markov Chain Based Scheme for Anomaly Detection” [2013], hereinafter “Sha”) in view of Bremaud (“Non-Homogenous Markov Chains” [2020]) in view of Kou (TWI737816B) further in view of Bai et al. (“Modeling Neural Networks Training Process with Markov Decision Process” [2021], hereinafter “Bai”) further in view of García et al. (EP3982298A1, hereinafter “García”) further in view of Kolychev et al. (US20190377880A1, hereinafter “Kolychev”).
Regarding Claim 5,
The combination of Sha/Bremaud/Kou/Bai teaches the method of Claim 1 (and thus the rejection of Claim 1 is incorporated). The combination fails to explicitly disclose but García discloses wherein multiple neural networks, each trained on segments …, are each used to generate output probability tensors (García [0135]; “At operation 8.2, the set of telemetry data is processed through a plurality of neural network layers of the one or more neural networks. The one or more neural networks maybe any of the (discriminator) neural networks described in relation to FIG.s 1-5. At operation 8.3, a classification for each of one or more of the channels of telemetry data in the set of telemetry data is output by the neural network. Each classification indicative of whether a corresponding channel of telemetry data (or ensemble of telemetry data/group of telemetry data) input to the neural network contains an anomaly. Each output classification may be indicative of a probability that the corresponding channel/ensemble/group of telemetry data input to the neural network contains an anomaly.” wherein multiple neural networks trained on telemetry data channels generate output classification
García [0062]; “The discriminator output 312 for such a set of input telemetry data 308 is an N-dimensional vector, and is denoted D(x).” wherein the neural networks’ output is a probability tensor)
It would have been obvious to modify Sha/Bremaud/Kou/Bai’s method of determining probability vectors for present and past state spaces and issuing a communication depending on the probability meeting a threshold by using multiple neural networks for probability tensor generation. One would have been motivated to do so because “Using a plurality of neural networks can increase the anomaly/ failure isolation capabilities of the system.” (García [0104]).
The combination of Sha/Bremaud/Kou/Bai/García does not explicitly disclose but Kolychev discloses segments of a fixed length different from a fixed length on which each other neural network was trained (Kolychev [0054]; “The tokenized potential vulnerabilities (e.g., string of characters) can be transmitted to one or more neural networks, trained, for example at different lengths of input array of numbers (e.g., excerpts). In some examples, a plurality of neural networks can be trained to each receive different lengths (of numbers in an array) of the tokenized potential vulnerabilities)
Sha/Bremaud/Kou/Bai/García discloses wherein multiple neural networks, each trained on segments …, are each used to generate output probability tensors. Sha/Bremaud/Kou/Bai/García does not disclose wherein multiple neural networks, each trained on segments of a fixed length different from a fixed length on which each other neural network was trained, are each used to generate output probability tensors. Kolychev, however, discloses segments of a fixed length different from a fixed length on which each other neural network was trained. By using the segments of differing fixed lengths for training input in the multiple neural networks of Sha/Bremaud/Kou/Bai/García, the combination of Sha/Bremaud/Kou/Bai/García/Kolychev reads wherein multiple neural networks, each trained on segments of a fixed length different from a fixed length on which each other neural network was trained, are each used to generate output probability tensors.
It would have been obvious to modify Sha/Bremaud/Kou/Bai/García’s method of determining probability vectors for present and past state spaces through multiple neural networks and issuing a communication depending on the probability meeting a threshold by using multiple neural networks of different segment input lengths for probability tensor generation. One would have been motivated to do so because “In other examples, combinations of neural networks (at different lengths of input array of numbers) can be provide a higher confidence of accuracy for verifying certain type of vulnerabilities.” (Kolychev [0054]).
Claim 13 recites the method performed by the system of Claim 5. Thus, Claim 13 is rejected for reasons set forth in the rejection of Claim 5.
Claims 6 and 14 are rejected under 35 U.S.C. 103 as being unpatentable over Sha et al. (“A Multi-Order Markov Chain Based Scheme for Anomaly Detection” [2013], hereinafter “Sha”) in view of Bremaud (“Non-Homogenous Markov Chains” [2020]) in view of Kou (TWI737816B) further in view of Bai et al. (“Modeling Neural Networks Training Process with Markov Decision Process” [2021], hereinafter “Bai”) further in view of García et al. (EP3982298A1, hereinafter “García”).
Regarding Claim 6,
The combination of Sha/Bremaud/Kou/Bai teaches the method of Claim 1 (and thus the rejection of Claim 1 is incorporated). The combination fails to explicitly disclose but García discloses wherein multiple neural networks are each used to generate output probability tensors, and a probability of anomaly is computed as a function of each of the output probability tensors' values for the current state (García [0135]; “At operation 8.2, the set of telemetry data is processed through a plurality of neural network layers of the one or more neural networks. The one or more neural networks maybe any of the (discriminator) neural networks described in relation to FIG.s 1-5. At operation 8.3, a classification for each of one or more of the channels of telemetry data in the set of telemetry data is output by the neural network. Each classification indicative of whether a corresponding channel of telemetry data (or ensemble of telemetry data/group of telemetry data) input to the neural network contains an anomaly. Each output classification may be indicative of a probability that the corresponding channel/ensemble/group of telemetry data input to the neural network contains an anomaly.” wherein the classifications derived from tensor neural network output of each neural network and its respective telemetry data channel reads on multiple neural networks generating output probability tensors to compute a probability of anomaly
García [0062]; “The discriminator output 312 for such a set of input telemetry data 308 is an N-dimensional vector, and is denoted D(x).” wherein the neural networks’ output is a probability tensor)
It would have been obvious to modify Sha/Bremaud/Kou/Bai’s method of determining probability vectors for present and past state spaces and issuing a communication depending on the probability meeting a threshold by using multiple neural networks to generate anomaly detections based on probability tensors. One would have been motivated to do so because “Using a plurality of neural networks can increase the anomaly/ failure isolation capabilities of the system” (García [0104]).
Claim 14 recites the method performed by the system of Claim 6. Thus, Claim 14 is rejected for reasons set forth in the rejection of Claim 6.
Response to Arguments
The Examiner acknowledges the Applicant’s amendments to Claims 2, 4, 13 and 15.
Applicant’s arguments filed February 9th, 2026, traversing the rejection of claims 1-16 under 35 U.S.C. § 101, but are not fully persuasive.
Applicant argues that the prior Office Action fails to explain how a human could mentally maintain a real-time, high-dimensional tensor probability map of a continuous process, such as, for example, the medical or meteorological processes described in the specification.
Examiner respectfully disagrees. The claim language at present does not recite the narrow scope of maintaining a real-time, high-dimensional tensor probability map of a continuous process. Instead, the claim language at present merely recites the abstract idea of determining a tensor by calculating the Cartesian product of probability vectors and generating a communication if the determined tensor exceeds a threshold. Such an idea is merely an evaluation mental process step (calculating Cartesian product of probability vectors, performable by pen and paper) and judgement (generating a communication if the determined tensor exceeds a threshold, thus a judgement of the tensor against a threshold).
Applicant argues that the claim language recites an improvement in technology or technical field, specifically enabling the system to detect anomalies in non-IID data with a precision that generic computers, which rely on standard Markov or IID-based models, cannot achieve.
Examiner respectfully disagrees. The claim language at present does not recite such improvements; although the specification may disclose such improvements in technology, they are not reflected in the claim language and thus cannot incorporate the abstract idea into practical application.
Applicant argues that Amended Claims 1 and 9 are integrated into practical application, stating that “automatically generate an electronic communication … to cause the client device to take prompt mitigating action” is a specific affirmative step that solves a technical problem.
Examiner respectfully disagrees. Examiner interprets the limitation as merely applying the abstract idea on generic computing components; in this instance, generation of an electronic communication is merely data outputting, and to cause the client device to take prompt mitigating action is applying the abstract idea of “determining a tensor … and generating a communication if the determined tensor exceeds a threshold” on a client device.
Applicant argues that activating and deactivating functionalities to mitigate an expected harm is not merely reciting the technological environment or field of use in which the abstract idea is to be performed. Therefore, it is a practical application of technological system to solve a technical problem.
Examiner respectfully disagrees. Examiner notes to applicant that “activating and deactivating functionalities” was presented in the original office action, not as a technological environment or field of use additional element, but rather mere data outputting. The field of use lies in the activating and deactivating of functionalities to mitigate an expected harm.
The remainder of the claim language is directed to additional elements that do not incorporate the abstract idea into practical application. As such, Claim 1 is subject-matter ineligible.
The rejection of Claim 1 under 35 U.S.C. § 101 has been maintained. Similarly, the rejection of Claim 9 under 35 U.S.C. § 103 has been maintained.
The rejection of Claims 2-8 under 35 U.S.C. § 101, which depend directly or indirectly from Claim 1, have been maintained.
The rejection of Claims 10-16 under 35 U.S.C. § 101, which depend directly or indirectly from Claim 9, have been maintained.
Applicant’s arguments filed February 9th, 2026, traversing the rejection of claims 1-16 under 35 U.S.C. § 103 have been fully considered, but are not fully persuasive.
Applicant recites, on Pages 9-11 of the Remarks, that Sha fails to suggest the specific limitation of calculating a Cartesian product during runtime on live output vectors to obtain a probability tensor, nor the specific limitation of using that specific tensor for anomaly detection on a running process.
Examiner respectfully disagrees. Examiner points to Section II Subsection A of Sha, which discloses “Therefore, to compensate the uncertainty in anomaly detection, results from multiple indicators or even multiple models are desired to issue a real-time warning for detrimental operations. In this paper, we address the above question by deploying the multi-order Markov chain based model.” wherein the multi-order Markov chain model is deployed in a real-time scenario for anomaly detection, thus reading on such a model and its inference operations occurring during runtime. As such, the combination of Sha/Ko/Bai all interpreted to be for use in a dynamic, real-time model. Additionally, examiner notes that applicant argues that Sha’s Cartesian product step is statically calculated in its transition matrix. However, introduced reference Bremaud introduces the application of non-homogenous transition matrices which comprise the determination of transition matrices dependent on real time n, thus disclosing an obvious replacement of Sha’s homogenous matrices which do not perform such Cartesian product determination in runtime environments.
Applicant argues, on Pages 10-11 of the Remarks, that Kou fails to suggest the claimed threshold determination, specifically because the nature of Kou’s data revolving around a scalar value being compared versus the claimed tensor language.
Examiner respectfully disagrees. Examiner notes that a scalar is interpretable and usable in the art as a rank-0 tensor, and thus suffices under the generic “tensor” language claimed.
The rejection of Claim 1 under 35 U.S.C. § 103 has been maintained. Similarly, the rejection of Claim 9 under 35 U.S.C. § 103 has been maintained.
The rejection of Claims 2-8 under 35 U.S.C. § 103, which depend directly or indirectly from Claim 1, have been maintained.
The rejection of Claims 10-16 under 35 U.S.C. § 103, which depend directly or indirectly from Claim 9, have been maintained.
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 JONATHAN J KIM whose telephone number is (571)272-0523. The examiner can normally be reached 8-6.
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, Matt Ell can be reached on (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.
/JONATHAN J KIM/Examiner, Art Unit 2141 /MATTHEW ELL/Supervisory Patent Examiner, Art Unit 2141
Prosecution Insights