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 .
Status of Claims
The present application is being examined under the claims filed 01/02/2026. The status of the claims are as follows:
Claims 1-20 are pending.
Claims 1-20 are amended.
Response to Amendment
The Office Action is in response to Applicant’s communication filled 01/02/2026 in response to office action mailed 10/02/2025. The Applicant’s remarks and any amendments to the claims or specification have been considered with the results that follow.
Response to Arguments
Regarding 35 U.S.C. § 112(b) (p. 7)
Applicant argues that claims 8 and 20 have been amended to address the antecedent basis issues raised in the prior Office Action, and requests withdrawal of the indefiniteness rejection. Applicant further points to paragraph [0025] of the specification as support for the newly added recitations concerning determining that an article of manufacture is defective or anomalous and sending the article for inspection.
Examiner’s response:
These arguments are persuasive only with respect to the specific antecedent-basis and number-consistency issues previously identified in claims 8 and 20. In the prior non-final Office Action, claim 8 was rejected because it recited “the sensor measurement data” without proper antecedent basis, and claim 20 was rejected because it introduced “one or more non-transitory memories” and later recited “the memory”, creating a number/antecedent mismatch. Applicant’s amendments appear to address those previously identified issues.
However, the present amendment introduces new indefiniteness issues in independent claims 9 and 20. As amended, claim 9 recites a “method for training a classifier” including generating training data and iteratively adjusting classifier parameters, but then further recites “to produce a predicted class of the article of manufacture”, “determining that the article of manufacture is defective or anomalous in response to the predicted class of the article of manufacture;” and “controlling the manufacturing process by sending the defective or anomalous article of manufacture for inspection.” Amended claims 20 contains corresponding limitations in system form.
These newly added limitations render the scope of claims 9 and 20 unclear because it is not reasonably clear, from the claim language itself, which article of manufacture is being classified and acted upon in the context of the recited training method/system, whether that article is one of the training examples or a separate production article, and how the recited live manufacturing-control action is tied to the preceding training operations. Thus, although Applicant has overcome the particular § 112 issues previously applied to claims 8 and 20, amended claims 9 and 20 remains indefinite under 35 U.S.C. § 112(b). The cited disclosure in paragraph [0025] pertains to operational manufacturing control based on a predicted class, but does not resolve the internal ambiguity created by inserting those operational limitations into the recited classifier-training claims.
Regarding 35 U.S.C. § 101 (pp. 7-8)
Applicant argues that the pending claims are not directed to an abstract idea because the amendments now expressly recite determining that an article of manufacture is defective or anomalous and controlling the manufacturing process by sending the article for inspection. Applicant contents that the claims therefore recite a concrete industrial process and integrate any alleged abstract idea into a practical application.
Examiner’s response
These arguments are not persuasive. Although Applicant amended independent claims 1 and 7 to additionally recite “determining that the article of manufacture is defective or anomalous in response to the predicted class of the article of manufacture” and “controlling the manufacturing process by sending the defective or anomalous article of manufacture for inspection”, the added limitations doe not integrate the recited judicial exception into a practical application.
In particular, the step of determining that the article of manufacture is defective or anomalous in response to the predicted class merely evaluates the result of the recited classification and, as such, amounts to an observation, judgment, or mental evaluation based on the classification output. Further, the step of controlling the manufacturing process by sending the defective or anomalous article of manufacture of inspection constitutes insignificant extra-solution activity, i.e., post-solution activity occurring after the recited data analysis and classification have been performed. The additional language merely uses the result of the abstract data analysis to decide whether an article should be routed for inspection.
Although Applicant points to paragraph [0025] as support for these limitations, the claim remains at a high level of generality and does not set forth a specific asserted improvement in computer functionality, manufacturing technology, or inspection technology. Rather, the claims continue to recite collecting sensor measurement data, extracting features, aggregating data, and classifying the article, followed by determining a condition of the article and sending the article for inspection based on that classification result. These additional limitations do not impose a meaningful limit on the judicial exception and therefore do not integrate the exception into a practical application.
Regarding 35 U.S.C. § 103 (pp. 8-9)
Applicant argues that neither Ba nor Toroman disclose determining that an article of manufacture itself is defective based on a predicted class or controlling a manufacturing process by sending defective or anomalous articles for inspection, and therefore the pending claims are not obvious over the cited combination.
Examiner’s response
These arguments are not persuasive. The Applicant is correct that the prior rejection over Ba in view of Toroman alone do not expressly address the newly added limitations. However, the rejection has been updated to further rely on Bufi, which teaches or at least suggests these additional limitations. Applicant’s argument therefore does not address the rejection as presently set forth (see 103 rejection below).
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 9-20 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim 9 is indefinite because it recites, inter alia, “a method for training a classifier to classify articles of manufacture”, including “generating training data for the classifier”, and “iteratively adjusting parameters of the classifier by reducing an error in the outputs of the classifier generated when the classifier is applied to each of the inputs in the training data pairs”, but then further recites “to produce a predicted class of the article of manufacture”, “determining that the article of manufacture is defective or anomalous in response to the predicted class of the article of manufacture”, and “controlling the manufacturing process by sending the defective or anomalous article of manufacture for inspection.” This language is unclear and fails to distinctly define and scope of the claimed training method, at least because:
the claim is directed to training a classifier using a plurality of training data pairs, but the later-recited “article of manufacture” is not clearly identified as to whether it is one of the articles represented in the training data, a separate article evaluated after training or some other article altogether; and
the claim does not clearly define how the recited acts of “determining that the article of manufacture is defective or anomalous” and “controlling the manufacturing process by sending the defective or anomalous article of manufacture for inspection” are tied to the preceding training steps, such that it is not distinctly clear whether the claim is directed to a training process, an operational classification process, or an unclear combination of both.
Additionally, claim 9 introduces the training-data framework reciting “a plurality of training data pairs” and iterative parameter adjustment based on classifier outputs for those pairs, but later recites a single “predicted class of the article of manufacture” and a corresponding manufacturing-control action. The claim does not distinctly specific whether the “predicted class” is generated for each training input, for one selected article, or for an article outside the recited training data, thereby rendering the scope of the step indeterminate.
Claims 10-19 are indefinite for the same reasons as claim 9 because they depend therefrom and do not cure the ambiguity introduced by claim 9 regarding which “article of manufacture” is being classified and acted upon, and how the recited manufacturing-control actions relate to the recited classifier-training operations.
Claim 20 is indefinite because it recites, inter alia, “a system for training a classifier”, including instructions for “generating training data for the classifier” and “iteratively adjusting parameters of the classifier by reducing an error in the outputs of the classifier generated when the classifier is applied to each of the inputs in the training data pairs”, but then further recites instructions “to produce a predicted class of the article of manufacture”, “determining that the article of manufacture is defective or anomalous in response to the predicted class of the article of manufacture”, and “controlling the manufacturing process by sending the defective or anomalous article of manufacture for inspection”. This language is unclear and fails to distinctly define the scope of the claimed system, at least because:
although claim 20 is framed as a system for training a classifier using training data pairs, the later-recited “article of manufacture” is not distinctly identified, such that it is unclear whether the claimed system is producing a predicted class for an article represented in the training data, for a separate production article, or for the same other article; and
the claim does not distinctly specify how the recited live manufacturing-control action of sending a defective or anomalous article for inspection is related to the recited training functionality, thereby rendering unclear whether the claim is directed to a training system, an operational manufacturing-control system, or an unclear combination of both.
Although Applicant points to paragraph [0025] of the specification as support for determining that an article is defective or anomalous and sending the article for inspection, that disclosure pertains to operational control of a manufacturing process based on a predicted class, and does not resolve the ambiguity created by inserting those operational limitations into claims otherwise directed to training a classifier. Accordingly, the metes and bounds of claims 9 and 20 are not reasonably clear.
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 rejected under 35 U.S.C. 101 because the claims are directed to abstract ideas without significantly more.
Regarding Claim 1
Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process (method).
Claim 1 – Step 2A Prong One - Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
Claim 1 recites the following abstract ideas:
applying a feature extractor to the received measurements to generate a feature vector of the article; - this limitation expressly calls for generating a “feature vector” form numerical “measurements”. A vector is a numerical array; generating it form input values is a mathematical calculation/representation and a mathematical concept under MPEP § 2106.04(a)(2)(I). See Electric Power Group: collecting/processing/analyzing data via algorithms are abstract.
aggregating the feature vector of the article with encoded time series data - this limitation recites mathematical aggregation/combination of vectors/encodings, a mental concept and abstract idea under MPEP § 2106.04(a)(2).
applying a classifier to the input to produce a predicted class of the article of manufacture. – this limitation recites mathematical classification (a model applying calculations to produce an output). Classifying data and presenting a result (prediction) are steps that can be done in the human mind or on pen and paper, i.e., a mental process and mathematical concept under MPEP § 2106.04(a)(2).
determining that the article of manufacture is defective or anomalous in response to the predicted class of the article of manufacture; - this limitation merely evaluates or interprets the result of the recited classification. The step of determining whether the article is “defective” or “anomalous” amounts to a conclusion, observation, or judgment based on that classification output. See MPEP § 2106.04(a)(2)(III).
Claim 1 – Step 2A Prong Two – Do the additional elements integrate the exception into a
practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
receiving measurements related to an article of manufacture, the measurements being captured at a first station in a manufacturing process; - this limitation recites mere data gathering in a specified field-of-use (manufacturing). It is an insignificant extra-solution activity that does not integrate the abstract calculations into a practical application. MPEP § 2106.05(g) (insignificant pre/post-solution activity); MPEP § 2106.04(g) (integration test).
“… representing a history of measurements of articles of a same type as the article of manufacture captured at a sequence of stations of the manufacturing process prior to the first station to generate an input to a classifier;” – this merely describes the source and content of the data being processed. Constraining data to a particular environment or content is an extra-solution/field-of-use limitation and does not amount to integration. MPEP § 2106.04(d), §2106.05(g).
controlling the manufacturing process by sending the defective or anomalous article of manufacture for inspection. – this limitation merely recites generic post-classification handling of the article at a high level of generality. The claim does not recite any particular manner of inspection, any specific inspection machine, any particular control logic for manufacturing equipment, or any technological manner in which the manufacturing process itself is improved. Instead, the limitation merely uses the result of the abstract analysis to decide that an article should be routed for inspection. This amounts to insignificant extra-solution activity and/or a field-of-use application of the abstract idea in a manufacturing environment, and therefore does not integrate the exception into a practical application. See MPEP § 2106.05(g); MPEP § 2106.04(d).
Claim 1 – Step 2B- Do the additional elements amount to “significantly more” than the judicial
exception?
No. When considered individually and in combination, the additional elements are well-understood,
routine, and conventional (WURC) activities. The additional elements:
receiving measurements related to an article of manufacture, the measurements being captured at a first station in a manufacturing process; - the specification describes conventional sensors (e.g., cameras, acoustics, pressure, ultrasound, spectroscopy), routine measurement capture, and a generic arrangement that merely provides input data. This is well-understood, routine, and conventional data-gathering activity that does not amount to “significantly more” than the exception. (WURC) MPEP § 2106.05(d).
“… representing a history of measurements of articles of a same type as the article of manufacture captured at a sequence of stations of the manufacturing process prior to the first station to generate an input to a classifier;” – merely specifying the content/origin of the data (prior stations in a manufacturing line; same article type) is a conventional data context and does not add a non-conventional computer component or a specific technological improvement. (WURC) MPEP § 2106.05(d).
controlling the manufacturing process by sending the defective or anomalous article of manufacture for inspection. – this limitation is recited at a high level of generality and does not require any specific technical mechanism for implementing the routing or inspection. It therefore amounts to well-understood, routine, and conventional (WURC) activity appended to the abstract analysis and does not supply an inventive concept. See MPEP § 2106.05(d).
Regarding Claim 7
Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a machine (system).
Claim 7 – Step 2A Prong One- Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
Claim 7 recites the following abstract ideas:
“applying a feature extractor to the received measurements to generate a feature vector of the article;” - this limitation expressly calls for generating a “feature vector” form numerical “measurements”. A vector is a numerical array; generating it form input values is a mathematical calculation/representation and a mathematical concept under MPEP § 2106.04(a)(2)(I). See Electric Power Group: collecting/processing/analyzing data via algorithms are abstract.
“aggregating the feature vector of the article with encoded time series data representing a history of measurements of articles of a same type as the article of manufacture captured at a sequence of stations of the manufacturing process prior to the first station to generate an input to a classifier;” - this limitation recites mathematical aggregation/combination of vectors/encodings, a mental concept and abstract idea under MPEP § 2106.04(a)(2).
“and applying a classifier to the input to produce a predicted class of the article of manufacture.” – this limitation recites mathematical classification (a model applying calculations to produce an output). Classifying data and presenting a result (prediction) are steps that can be done in the human mind or on pen and paper, i.e., a mental process and mathematical concept under MPEP § 2106.04(a)(2).
determining that the article of manufacture is defective or anomalous in response to the predicted class of the article of manufacture; - this limitation merely evaluates or interprets the result of the recited classification. The step of determining whether the article is “defective” or “anomalous” amounts to a conclusion, observation, or judgment based on that classification output. See MPEP § 2106.04(a)(2)(III).
Claim 7 – Step 2A Prong Two – Do the additional elements integrate the exception into a
practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
“one or more processors;” and “one or more non-transitory memories communicatively connected to the one or more processors, the one or more memories including computer-executable instructions that when executed cause the system to perform the following functions:” – these two limitations are merely invoking a computer as a tool to execute the abstract calculations. There is no improvement to the computer itself, no particular machine beyond generic hardware. It does not integrate into a practical application. See MPEP § 2106.05(a) (improvement to computer technology), §2106.05(b) (particular machine), and §2106.05(f) (mere instructions to apply an exception).
“receiving measurements related to an article of manufacture, the measurements being captured at a first station in a manufacturing process;” – this limitation recites receiving measurements at a first station in a manufacturing flow, a pre-solution data gathering and field-of-use (manufacturing) limitation. This amounts to insignificant extra-solution activity and does not effect a transformation nor improve sensing/computing technology. No integration. See MPEP § 2106.05(g) and §2106.05(h).
controlling the manufacturing process by sending the defective or anomalous article of manufacture for inspection. – this limitation merely recites generic post-classification handling of the article at a high level of generality. The claim does not recite any particular manner of inspection, any specific inspection machine, any particular control logic for manufacturing equipment, or any technological manner in which the manufacturing process itself is improved. Instead, the limitation merely uses the result of the abstract analysis to decide that an article should be routed for inspection. This amounts to insignificant extra-solution activity and/or a field-of-use application of the abstract idea in a manufacturing environment, and therefore does not integrate the exception into a practical application. See MPEP § 2106.05(g); MPEP § 2106.04(d).
Claim 7 – Step 2B- Do the additional elements amount to “significantly more” than the judicial
exception?
No. When considered individually and in combination, the additional elements are well-understood,
routine, and conventional (WURC) activities. The additional elements:
“one or more processors;” and “one or more non-transitory memories communicatively connected to the one or more processors, the one or more memories including computer-executable instructions that when executed cause the system to perform the following functions:” - these two limitations recite generic processors and memories executing software. The specification itself depicts a standard computer environment and ordinary computer-readable media, that is well-understood, routine, and conventional (WURC):
“general computer system 200” can “include a set of instructions that can be executed” and used to “perform any one or more of the methods or computer-based functions” and may include “one or more processors 202… may include … one or more … CPUs … GPUs … [and] may also include a disk drive unit … signal generation device … network interface device…” (Spec. ¶[0026]-[0027]); (MPEP § 2106.05(d)).
“receiving measurements related to an article of manufacture, the measurements being captured at a first station in a manufacturing process;” – this limitation merely recites routine data capture in manufacturing lines, that, per the disclosure’s own background/detail, is well-understood, routine, and conventional activity. See MPEP § 2106.05(d), WURC; Spec. ¶[0005].
controlling the manufacturing process by sending the defective or anomalous article of manufacture for inspection. – this limitation is recited at a high level of generality and does not require any specific technical mechanism for implementing the routing or inspection. It therefore amounts to well-understood, routine, and conventional (WURC) activity appended to the abstract analysis and does not supply an inventive concept. See MPEP § 2106.05(d).
Regarding Claim 9
Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a method (process).
Claim 9 – Step 2A Prong One- Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
The additional elements:
“generating training data for the classifier, the training data including a plurality of training data pairs, wherein each of the plurality of training data pairs includes an input to the classifier and a predetermined output that the classifier is being trained to produce when the classifier is applied to the input” – this limitation recites creating/organizing information into labeled pairs by rule, which is a mathematical concept (information organization/creation for computation) and can also be performed as a mental process (defining inputs and expected outputs using pen and paper). See MPEP § 2106.04(a)(2)(I) (mathematical concepts) and (III)(C) (mental processes).
“and wherein each input in the plurality of inputs in the plurality of training data pairs includes an aggregation of:” - this limitation recites an aggregation operation (e.g., concatenating/fusing) to form a model input, a mathematical manipulation of data, a mathematical concept. MPEP § 2106.04(a)(2)(I).
a feature vector of an article of manufacture based on one or more measurements related to the article captured at a first station of a manufacturing process – this recites, “a feature vector of an article of manufacture”, i.e., a numerical array produced/used for computation; specifying it as part of the input recites a mathematical relationship. (Mathematical concept; MPEP § 2106.04(a)(2)(I).
“and iteratively adjusting parameters of the classifier by reducing an error in the outputs of the classifier generated when the classifier is applied to each of the inputs in the training data pairs to produce a predicted class of the article of manufacture” – this limitation recites error reduction/optimization during classifier training, i.e., mathematical processing, a mathematical concept under MPEP § 2106.04(a)(2)(I). Further, producing a “predicted class” is merely generating a classification result from the mathematical model, which is itself a mathematical classification and may also be performed as a mental evaluation or categorization, i.e., a mental process under MPEP § 2106.04(a)(2)(III)(C).
determining that the article of manufacture is defective or anomalous in response to the predicted class of the article of manufacture; - this limitation merely evaluates or interprets the result of the recited classification. The step of determining whether the article is “defective” or “anomalous” amounts to a conclusion, observation, or judgment based on that classification output. See MPEP § 2106.04(a)(2)(III).
Claim 9 – Step 2A Prong Two – Do the additional elements integrate the exception into a
practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
“and encoded time series data representing a history of measurements of articles of a same type as the article of manufacture captured at a sequence of stations of the manufacturing process prior to the first station;” – this merely identifies where the input information comes from and what it contains (manufacturing station context), a pre-solution data gathering/field-of-use limitation that does not improve computer functionality, tie the abstract processing to a particular machine in a meaningful way, or effect a particular transformation of an article. See MPEP § 2106.05(g) (insignificant extra-solution activity), §2106.05(h) (field-of-use), §2106.05(a)-(c) (no improvement to computer/tech; no particular machine; no transformation).
controlling the manufacturing process by sending the defective or anomalous article of manufacture for inspection. – this limitation merely recites generic post-classification handling of the article at a high level of generality. The claim does not recite any particular manner of inspection, any specific inspection machine, any particular control logic for manufacturing equipment, or any technological manner in which the manufacturing process itself is improved. Instead, the limitation merely uses the result of the abstract analysis to decide that an article should be routed for inspection. This amounts to insignificant extra-solution activity and/or a field-of-use application of the abstract idea in a manufacturing environment, and therefore does not integrate the exception into a practical application. See MPEP § 2106.05(g); MPEP § 2106.04(d).
Claim 9 – Step 2B- Do the additional elements amount to “significantly more” than the judicial
exception?
No. When considered individually and in combination, the additional elements are well-understood,
routine, and conventional (WURC) activities. The additional elements:
“and encoded time series data representing a history of measurements of articles of a same type as the article of manufacture captured at a sequence of stations of the manufacturing process prior to the first station;” – the station/history data-source/content constraints recited in this limitation are, individually and in combination, well-understood, routine, and conventional (WURC) activities in the field (routine multi-station measurement logging and using those logs to assemble training inputs). See MPEP § 2106.05(d) (Berkheimer). Appending such routine data-gathering/context to an otherwise abstract training pipeline (construct labeled pairs [Wingdings font/0xE0] mathematically aggregate representations [Wingdings font/0xE0] optimize by error reduction) does not supply an inventive concept. See also MPEP § 2106.05(f) (mere instructions to apply an exception on a computer).
controlling the manufacturing process by sending the defective or anomalous article of manufacture for inspection. – this limitation is recited at a high level of generality and does not require any specific technical mechanism for implementing the routing or inspection. It therefore amounts to well-understood, routine, and conventional (WURC) activity appended to the abstract analysis and does not supply an inventive concept. See MPEP § 2106.05(d).
Regarding Claim 20
Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a machine (system).
Claim 20 – Step 2A Prong One- Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
“generating training data for the classifier, the training data including a plurality of training data pairs, wherein each of the plurality of training data pairs includes an input to the classifier and a predetermined output that the classifier is being trained to produce when the classifier is applied to the input” – this limitation recites creating and organizing information (labeled input/output pairs) by rule, which can be done in the human mind or with pen and paper. See MPEP § 2106.04(a)(2)(I) and (III)(C) (mental processes).
“wherein each input in the plurality of inputs in the plurality of training data pairs includes an aggregation of: a feature vector of an article of manufacture based on one or more measurements related to the article captured at a first station of a manufacturing process;” – a feature vector is a numeric array/representation used for computation and a mathematical concept under MPEP § 2106.04(a)(2).
and iteratively adjusting parameters of the classifier by reducing an error in the outputs of the classifier generated when the classifier is applied to each of the inputs in the training data pairs to produce a predicted class of the article of manufacture;” – this limitation recites error reduction/optimization during classifier training, i.e., mathematical processing, a mathematical concept under MPEP § 2106.04(a)(2)(I). Further, producing a “predicted class” is merely generating a classification result from the mathematical model, which is itself a mathematical classification and may also be performed as a mental evaluation or categorization, i.e., a mental process under MPEP § 2106.04(a)(2)(III)(C).
determining that the article of manufacture is defective or anomalous in response to the predicted class of the article of manufacture; - this limitation merely evaluates or interprets the result of the recited classification. The step of determining whether the article is “defective” or “anomalous” amounts to a conclusion, observation, or judgment based on that classification output. See MPEP § 2106.04(a)(2)(III).
Claim 20 – Step 2A Prong Two – Do the additional elements integrate the exception into a
practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
“wherein each input in the plurality of inputs in the plurality of training data pairs includes an aggregation of: … encoded time series data representing a history of measurements of articles of a same type as the article of manufacture captured at a sequence of stations of the manufacturing process prior to the first station; - this clause constrains the information content/source (history/provenance). Merely specifying where the information comes from and what it contains (manufacturing context) constitutes pre-solution data gathering/field-of-use limits and does not integrate the abstract idea. See MPEP § 2106.05(g) and §2106.05(h).
controlling the manufacturing process by sending the defective or anomalous article of manufacture for inspection. – this limitation merely recites generic post-classification handling of the article at a high level of generality. The claim does not recite any particular manner of inspection, any specific inspection machine, any particular control logic for manufacturing equipment, or any technological manner in which the manufacturing process itself is improved. Instead, the limitation merely uses the result of the abstract analysis to decide that an article should be routed for inspection. This amounts to insignificant extra-solution activity and/or a field-of-use application of the abstract idea in a manufacturing environment, and therefore does not integrate the exception into a practical application. See MPEP § 2106.05(g); MPEP § 2106.04(d).
Claim 20 – Step 2B- Do the additional elements amount to “significantly more” than the judicial
exception?
No. When considered individually and in combination, the additional elements are well-understood,
routine, and conventional (WURC) activities. The additional elements:
“wherein each input in the plurality of inputs in the plurality of training data pairs includes an aggregation of: … encoded time series data representing a history of measurements of articles of a same type as the article of manufacture captured at a sequence of stations of the manufacturing process prior to the first station; - this clause constrains the information content/source (history/provenance). Generic and routine station/prior-station data capture, are well-understood, routine, conventional (WURC) activities. Using a conventional computing environment and standard data-collection provenance to implement an abstract supervised-training workflow (construct labeled pairs [Wingdings font/0xE0] aggregate numeric representations [Wingdings font/0xE0] minimize error) does not supply an inventive concept. See MPEP § 2106.05(d) and §2106.05(f).
controlling the manufacturing process by sending the defective or anomalous article of manufacture for inspection. – this limitation is recited at a high level of generality and does not require any specific technical mechanism for implementing the routing or inspection. It therefore amounts to well-understood, routine, and conventional (WURC) activity appended to the abstract analysis and does not supply an inventive concept. See MPEP § 2106.05(d).
Regarding claims 2-6
Claims 2-6 depend from claim 1 and were not amended in a manner that changes the nature of their additional limitations. The limitations of claims 2-6 were previously addressed in the prior Office Action and, when considered individually and in combination with amended claim 1, do not integrate the recited judicial exception into a practical application and do not amount to significantly more than the judicial exception itself. In particular, claim 2 merely further specifies sensor-based data capture, claims 3-5 merely specify particular classifier model types, and claim 6 merely further specifies the content of the encoded time series data. Accordingly, claims 2-6 remain rejected under 35 U.S.C. § 101.
Regarding claim 8
Claim 8 depends from claim 7 and was not amended in a manner that changes the nature of its additional limitations. The limitations of claim 8 were previously addressed in the prior Office Action and, when considered individually and in combination with amended claim 7, do not integrate the recited judicial exception into a practical application and do not amount to significantly more than the judicial exception itself. In particular, claim 8 merely further specifies that the measurements are sensor measurement data and that the claimed system includes one ore more sensors configured to capture the sensor measurement data. These additional limitations merely further recite data gathering using generic sensing components and do not add specific technological improvement or other meaningful sufficient to confer eligibility. Accordingly, claim 8 remains rejected under 35 U.S.C. § 101.
Regarding claims 10-19
Claims 10-19 depend from claim 9 and were not amended in a manner that changes the nature of their additional limitations. The limitations of claims 10-19 were previously addressed in the prior Office Action and, when considered individually and in combination with amended claim 9, do not integrate the recited judicial exception into a practical application and do not amount to significantly more than the judicial exception itself. The additional limitations of claims 10-19 merely further specify details of the training data, classifier architecture, model type, encoded time series data, and classifier output, and do not add a specific technological improvement or other meaningful limitation sufficient to confer eligibility. Accordingly, claims 10-19 remain rejected under 35 U.S.C. § 101.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over Ba et al. (US20230206040A1) henceforth “Ba” in view of Toroman et al. (US20230107337A1) henceforth “Toroman” in further view of Bufi et al. (WO2021062536A1) henceforth “Bufi”.
Regarding claim 1, Ba in view of Toroman and further in view of Bufi, teach a method for classifying an article of manufacture, comprising:
“receiving measurements related to an article of manufacture” – Ba teaches this limitation. Ba teaches:
“A processor may collect data from each of two or more stations of a set of stations..” (Ba, §Abstract)
“In some embodiments, the data may be sensor data (e.g., from IoT systems, cameras, thermal sensors, etc.) that reflects the performance or operation of the stations.” (Ba, pg. 1, ¶[0018])
This teaches receiving sensor measurements captured at a station in a manufacturing process.
“aggregating the feature vector of the article with encoded time series data representing a history of measurements of articles of a same type as the article of manufacture captured at a sequence of stations of the manufacturing process prior to the first station to generate an input to a classifier;” – Ba teaches this limitation in part (bolded). Ba teaches collection/capture of data at stations in a manufacturing setting:
“A processor may collect data from each of two or more stations of a set of stations..” (Ba, §Abstract)
“… determining the subsets of the set of stations that are related … utilizing historical data regarding the subsets of stations …” (Ba, pg. 2, ¶[0022])
Ba does not teach:
“applying a feature extractor to the received measurements to generate a feature vector of the article;”
“aggregating the feature vector of the article with encoded time series data representing a history of measurements of articles of a same type as the article of manufacture … ”
“and applying a classifier to the input to produce a predicted class of the article of manufacture.”
“determining that the article of manufacture is defective or anomalous in response to the predicted class of the article of manufacture;”
“controlling the manufacturing process by sending the defective or anomalous article of manufacturing for inspection.”
Toroman, however, teaches these limitations:
“applying a feature extractor to the received measurements to generate a feature vector of the article;” – Toroman teaches this limitation. Toroman teaches:
“The encodings of the set of tiles can be concatenated encoding or otherwise combined to form one feature vector for classification purposes.” (Toroman, pg. 9, ¶[0122])
“aggregating the feature vector of the article with encoded time series data representing a history of measurements of articles of a same type as the article of manufacture … ” Toroman teaches this limitation. Toroman teaches:
“encodings from different time scales are used to train a classification model … The encodings … can be concatenated … to form one feature vector for classification purposes.” (Toroman, pg. 9, ¶[0121]-[0122])
This teaches encoded time-series history associated with a sequence of stations and combining/aggregating that encoded history with the article’s feature vector to form classifier input.
“and applying a classifier to the input to produce a predicted class of the article of manufacture.” – Toroman teaches this limitation. Toroman teaches:
“encodings from different time scales are used to train a classification model that classifies the state of a system or machine …” (Toroman, pg. 9, ¶[0121])
This teaches applying a classifier to encoded inputs to output a classification (predicted class).
Ba nor Toroman teach these limitations:
“determining that the article of manufacture is defective or anomalous in response to the predicted class of the article of manufacture;”
“controlling the manufacturing process by sending the defective or anomalous article of manufacturing for inspection.”
Bufi, however, teaches these limitations:
“determining that the article of manufacture is defective or anomalous in response to the predicted class of the article of manufacture;” – Bufi teaches a defect-detection model that:
“generate[s] defect data describing a detected defect as an output” (Bufi, pg. 3, ¶[0012])
And teaches a programmable logic controller (“PLC”) device for:
“receiving the defect data from the node computing device and determining whether the defect is acceptable or unacceptable by comparing the defect data to tolerance data.” (Bufi, pg. 3, ¶[0012])
Bufi further teaches that:
“Articles 110 may be classified as defective or non-defective by the system 100.” (Bufi, pg. 11, ¶[0084])
“if the node device 148 finds a defect, the PLC 146 may receive an NG code from the node device 148 indicating that a defect has been found in the article 110 and that the article 110 is defective (i.e., ‘no good’).” (Bufi, pg. 25, ¶[0182])
“The NG code may include defect data describing the defect. The defect data may include a defect type or class” (Bufi, pg. 25, ¶[0183])
“the PLC 146 determines whether the defect is an NG (no good, confirm defect) or OK (reject defect) using the received defect data and defect specifications stored at the PLC 146.” (Bufi, pg. 39, ¶[0283])
Thus, Bufi teaches determining from classifier/defect-output data, including defect type/class information, whether the article is defective or anomalous.
“controlling the manufacturing process by sending the defective or anomalous article of manufacturing for inspection.” – Bufi teaches this limitation. Bufi teaches that:
“The visual inspection system 100 is located at an inspection station 108.” (Bufi, pg. 10, ¶[0078])
“may be integrated within other parts of the manufacturing or processing process (e.g., on a conveyor or the like).” (Bufi, pg. 10, ¶[0079])
“may include an automatic transport mechanism such as a conveyor belt, for transporting articles to and from the inspection station 108” (Bufi, pg. 10, ¶[0080])
“the system 100 can advantageously be configured to be positioned inline with respect to existing processes at the manufacturing facility” (Bufi, pg. 10, ¶[0081])
Bufi further teaches that:
“By identifying articles 110 as defective or non-defective, the inspected articles can be differentially treated based on the outcome of the visual inspection. Defective articles 110 may be discarded or otherwise removed from further processing. Non-defective articles 110 may continue with further processing.” (Bufi, pg. 11, ¶[0085])
Bufi also teaches that:
“Upon determining the defect data is unacceptable, the robotic arm may be configured to autonomously move the camera to an article identifier position” (Bufi, pg. 3, ¶[0014])
“The PLC device may be configured to generate a stop inspection command upon determining the defect data is unacceptable.” (Bufi, pg. 4, ¶[0022])
And Bufi discloses that an article loader may:
“deliver the article to either a defective article chute (if defects present) or to a tote or tray (if no defects).” (Bufi, pg. 17, ¶[]0126)
These teachings show manufacturing-process control in response to defect determination, including inline handling and differential routing of defective articles for inspection/disposition.
It would have been obvious to a POSITA to further modify the Ba/Toroman combination with the defect-decision and inspection-control teachings of Bufi so that, once Toroman’s classifier outputs a predicted class based on Ba’s multi-station manufacturing measurements and encoded operational history, the system determines whether the corresponding article is defective or anomalous and controls downstream manufacturing handling accordingly. Bufi expressly teaches using machine-learning-generated defect output in an inline manufacturing inspection system to determine whether defect data is acceptable or unacceptable, whether a part is NG or OK, and to control subsequent inspection operations and article handling in response to that determination. Combining these teachings would have been a predictable use of known quality-control techniques to improve manufacturing decision-making and inspection flow by identifying suspect articles based on classifier output and subjecting those articles to further inspection-related handling before continued downstream processing. Such a combination merely applies known ML-based classification output to known manufacturing quality-control handling for the predictable benefit of reducing propagation of defective parts and improving inspection efficiency.
Regarding claim 9, Ba in view of Toroman and further in view of Bufi, teach a method for training a classifier to classify articles of manufacture, comprising:
generating training data for the classifier, the training data including a plurality of training data pairs, wherein each of the plurality of training data pairs includes an input to the classifier and a predetermined output that the classifier is being trained to produce when the classifier is applied to the input,
and wherein each input in the plurality of inputs in the plurality of training data pairs includes an aggregation of: a feature vector of an article of manufacture based on one or more measurements related to the article captured at a first station of a manufacturing process;
and encoded time series data representing a history of measurements of articles of a same type as the article of manufacture captured at a sequence of stations of the manufacturing process prior to the first station;
iteratively adjusting parameters of the classifier by reducing an error in the outputs of the classifier generated when the classifier is applied to each of the inputs in the training data pairs to produce a predicted class of the article of manufacture;
determining that the article of manufacture is defective or anomalous in response to the predicted class of the article of manufacture; and
controlling the manufacturing process by sending the defective or anomalous article of manufacture for inspection.
Ba teaches these limitations:
“and wherein each input in the plurality of inputs in the plurality of training data pairs includes an aggregation of: a feature vector of an article of manufacture based on one or more measurements related to the article captured at a first station of a manufacturing process;” - Ba teaches this limitation in part (bolded). Ba teaches a multi-station context that expressly includes a first station:
“the engineered data may include data from station 1, station 2, station 3, etc. through station N.” (Ba, pg. 3, ¶[0037])
“the processor learns the parameters/residuals for the first station…” (Ba, pg. 3, ¶[0038])
and encoded time series data representing a history of measurements of articles of a same type as the article of manufacture captured at a sequence of stations of the manufacturing process prior to the first station; - Ba teaches this limitation in part (bolded). Ba teaches multi-station line and prior-station history (station 1…N), first station, and station-level data/forecasting:
“… the engineered data may include data from station 1, station 2, station 3, etc. through station N.” (Ba, pg. 3, ¶[0037])
and iteratively adjusting parameters of the classifier by reducing an error in the outputs of the classifier generated when the classifier is applied to each of the inputs in the training data pairs. – Ba teaches this limitation in part (bolded). Ba teaches iterative parameter adjustment driven by error/residual during training:
“the processor … tunes parameters for each/any of stations 1 through station N … if the residual … deviates from zero, the tuning parameters … are updated …” (Ba, pg. 3, ¶[0038])
Ba, however, does not teach:
“generating training data for the classifier, the training data including a plurality of training data pairs, wherein each of the plurality of training data pairs includes an input to the classifier and a predetermined output that the classifier is being trained to produce when the classifier is applied to the input”
“and wherein each input in the plurality of inputs in the plurality of training data pairs includes an aggregation of:”
“… generated when the classifier is applied to each of the inputs in the training data pairs”
Toroman teaches these limitations:
“generating training data for the classifier, the training data including a plurality of training data pairs, wherein each of the plurality of training data pairs includes an input to the classifier and a predetermined output that the classifier is being trained to produce when the classifier is applied to the input” – Toroman teaches this. Toroman teaches:
“In an embodiment, encodings from different time scales are used to train a classification model that classifies the state of a system or machine …” (Toroman, p. 9, ¶[0121])
“Machine learning component 120 automatically generates a training instance for each tile selection. Each training instance includes a label … and … an encoding from each of the identified training tiles for supervised learning.” (Toroman, p. 9, ¶[0123])
This teaches generating training data pairs (inputs + predetermined outputs/labels) for training a classifier.
“and wherein each input in the plurality of inputs in the plurality of training data pairs includes an aggregation of:” – Toroman teaches this limitation. Toroman teaches forming a feature vector:
“The encodings of the set of tiles can be … combined to form one feature vector for classification purposes.” (Toroman, pg. 9, ¶[0121])
“… generated when the classifier is applied to each of the inputs in the training data pairs” – Toroman teaches this limitation. Toroman teaches training instances with labels/encodings:
“Machine learning component 120 automatically generates a training instance for each tile selection. Each training instance includes a label identifying the manually selected state/classification and, as feature values to include in the training instance, an encoding from each of the identified training tiles for supervised learning” (Toroman, p. 9, ¶[0123])
Ba nor Toroman teach these limitations:
“determining that the article of manufacture is defective or anomalous in response to the predicted class of the article of manufacture;”
“controlling the manufacturing process by sending the defective or anomalous article of manufacturing for inspection.”
Bufi, however, teaches these limitations:
“determining that the article of manufacture is defective or anomalous in response to the predicted class of the article of manufacture;” – Bufi teaches a defect-detection model that:
“generate[s] defect data describing a detected defect as an output” (Bufi, pg. 3, ¶[0012])
And teaches a programmable logic controller (“PLC”) device for:
“receiving the defect data from the node computing device and determining whether the defect is acceptable or unacceptable by comparing the defect data to tolerance data.” (Bufi, pg. 3, ¶[0012])
Bufi further teaches that:
“Articles 110 may be classified as defective or non-defective by the system 100.” (Bufi, pg. 11, ¶[0084])
“if the node device 148 finds a defect, the PLC 146 may receive an NG code from the node device 148 indicating that a defect has been found in the article 110 and that the article 110 is defective (i.e., ‘no good’).” (Bufi, pg. 25, ¶[0182])
“The NG code may include defect data describing the defect. The defect data may include a defect type or class” (Bufi, pg. 25, ¶[0183])
“the PLC 146 determines whether the defect is an NG (no good, confirm defect) or OK (reject defect) using the received defect data and defect specifications stored at the PLC 146.” (Bufi, pg. 39, ¶[0283])
Thus, Bufi teaches determining from classifier/defect-output data, including defect type/class information, whether the article is defective or anomalous.
“controlling the manufacturing process by sending the defective or anomalous article of manufacturing for inspection.” – Bufi teaches this limitation. Bufi teaches that:
“The visual inspection system 100 is located at an inspection station 108.” (Bufi, pg. 10, ¶[0078])
“may be integrated within other parts of the manufacturing or processing process (e.g., on a conveyor or the like).” (Bufi, pg. 10, ¶[0079])
“may include an automatic transport mechanism such as a conveyor belt, for transporting articles to and from the inspection station 108” (Bufi, pg. 10, ¶[0080])
“the system 100 can advantageously be configured to be positioned inline with respect to existing processes at the manufacturing facility” (Bufi, pg. 10, ¶[0081])
Bufi further teaches that:
“By identifying articles 110 as defective or non-defective, the inspected articles can be differentially treated based on the outcome of the visual inspection. Defective articles 110 may be discarded or otherwise removed from further processing. Non-defective articles 110 may continue with further processing.” (Bufi, pg. 11, ¶[0085])
Bufi also teaches that:
“Upon determining the defect data is unacceptable, the robotic arm may be configured to autonomously move the camera to an article identifier position” (Bufi, pg. 3, ¶[0014])
“The PLC device may be configured to generate a stop inspection command upon determining the defect data is unacceptable.” (Bufi, pg. 4, ¶[0022])
And Bufi discloses that an article loader may:
“deliver the article to either a defective article chute (if defects present) or to a tote or tray (if no defects).” (Bufi, pg. 17, ¶[]0126)
These teachings show manufacturing-process control in response to defect determination, including inline handling and differential routing of defective articles for inspection/disposition.
It would have been obvious to a POSITA to further modify the Ba/Toroman combination with the defect-decision and inspection-control teachings of Bufi so that, once the classifier is trained using Ba’s multi-station manufacturing measurements and Toroman’s encoded operational history and classification framework, the trained classifier produces a predicted class for an article of manufacture, and that predicted class is used to determine whether the corresponding article is defective or anomalous and to control downstream manufacturing handling accordingly. Claim 9 expressly recites a training method in which classifier parameters are iteratively adjusted “to produce a predicted class of the article of manufacture”, followed by determining defectiveness or anomaly and sending the article for inspection. Bufi teaches using AI-generated defect output in an inline manufacturing inspection system to determine whether an article is defective or non-defective, including that “Article 110 may be classified as defective or non-defective by the system 100”, and that defective articles may be “discarded or otherwise removed from further processing”, while non-defective articles continue. Bufi also teaches that the node device sends defect data including “a defect type” to the PLC, that is a defect is found an NG code is sent, and that the PLC determines whether the defect is “NG (no good, confirm defect) or OK (reject defect)”. Combining these teachings would have been a predictable use of known quality-control techniques to train a classifier in the Ba/Toroman system for the purpose of producing classification output that can be used in the same known manner as Bufi to identify suspect articles and subject them to further inspection-related handling before continued downstream processing, thereby improving quality control and inspection efficiency.
Regarding claims 2-6
Claims 2-6 depend from amended claim 1 and were not amended in a manner that changes the nature of their additional limitations. The limitations of claims 2-6 were previously addressed in the prior Office Action and, when considered individually and in combination with amended claim 1, do not alter the conclusion of obviousness over Ba in view or Toroman and further in view of Bufi. In particular, claim 2 merely further specifies that the measurements are captured by sensors at the recited stations; claims 3-5 merely further specify particular classifier model types; and claim 6 merely further specifies that the encoded time series data includes one or more predicted measurements of the article at the first station. These additional limitations were previously shown to be taught or suggested by Ba and/or Toroman, and the newly added defect/anomaly-determination and manufacturing-control limitations of base claim 1 are taught or at least suggested by Bufi. Accordingly, claims 2-6 are rejected under 35 U.S.C. § 103 for the same reasons set forth for amended claim 1, mutatis mutandis.
Regarding claims 7-8
Claims 7-8 are system claims analogous in scope to claim 1. The system recitations (one or more processors, one or more non-transitory memories, and the explicit sensor /feature extractor / aggregator / classifier components) merely implement, as structural components configured to perform, the same functional limitations analyzed for claim 1. No additional material limitation changes the substance of the combination. Accordingly, claims 7-8 are rejected under 35 U.S.C. § 103 for the same reasons set forth for claim 1, mutatis mutandis. The cited references, Ba in view of Toroman and further in view of Bufi, and the motivation to combine set forth for claim 1 apply equally to claims 7-8.
Regarding claims 10-19
Claims 10-19 depend from amended claim 9 and were not amended in a manner that changes the nature of their additional limitations. The limitations of claims 10-19 were previously addressed in the prior Office Action and, when considered individually and in combination with amended claim 9, do not alter the conclusion of obviousness over Ba in view of Toroman and further in view of Bufi. In particular, claims 10-11 merely further specify sensor-based capture of the training measurements; claims 12-14 merely further specify the type of feature extractor; claim 15 merely further specifies that the feature vector is a one-dimensional vector including a plurality of elements; claim 16 merely further specifies that the encoded time series data includes one or more predicted measurements of the article at the first station; and claims 17-19 merely further specify particular classifier model types. These additional limitations were previously shown to be taught or suggested by Ba and/or Toroman in the non-final Office Action, while the newly added limitations of amended claim 9 concerning producing a predicted class, determining defectiveness or anomaly, and controlling manufacturing handling are taught or at least suggested by Bufi in the manner discussed above. Accordingly, claims 10-19 are rejected under 35 U.S.C. § 103 for the same reasons set forth for amended claim 9, mutatis mutandis.
Regarding claim 20
Claim 20 is a system claim analogous in scope to claim 9. The system recitations (one or more processors and one or more non-transitory memories storing instructions) merely implement, as structural components configured to perform, the same training, predicted-class generation, defect/anomaly determination and manufacturing-control functions analyzed for claim 9. The amendment to recite “one or more non-transitory memories” does not materially change the substance of the obviousness analysis. The cited references, Ba in view of Toroman and further in view of Bufi, and the motivation to combine set forth for claim 9 apply equally to claim 20.
Conclusion
THIS ACTION IS MADE FINAL. 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 Paul Coleman whose telephone number is (571)272-4687. The examiner can normally be reached Mon-Fri.
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, David Yi can be reached at (571) 270-7519. 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.
/PAUL COLEMAN/ Examiner, Art Unit 2126
/DAVID YI/ Supervisory Patent Examiner, Art Unit 2126