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 .
Status of the Claims
Claims 1-21 are pending for examination.
Claims 1, 11 and 12 are independent Claims.
Claims 1-21 are rejected under 35 U.S.C. §112(b).
Claims 1-21 are rejected under 35 U.S.C. §103.
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 1-20 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.
Claims 1, 10, 11 and 12 recite the limitation “the calculated degrees of association” in “operation as a weight determining unit …” limitation. There is insufficient antecedent basis for this limitation in the claim.
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.
Claim(s) 1, 2, 7-12 and 21 is/are rejected under 35 U.S.C. 103 as being unpatentable over Matthew et al. (U.S. 20200034962 hereinafter Matthew) in view of Colt et al. (U.S. 2017/0115658 hereinafter Colt) in further view of Hwang et al. (U.S. 2018/0357539 hereinafter Hwang).
As Claim 1, Matthew teaches a parameter adjustment apparatus comprising a processor configured with a program to perform (Matthew (¶0009, processor and a computer readable storage media) operations comprising:
operation as an information acquiring unit for acquiring historical existing task information regarding a plurality of existing inference tasks (Matthew (¶0033 line 1-7), “As an example, an image can be input to multiple pre-trained neural networks 202, 204, 206. Each of those networks 202, 204, 206 produce a respective feature map 208, 210, 212 of the image. The feature maps 208, 210, 212 can identify, for example, objects within the image (such as an apple or strawberry) as well as aspects of those objects (such as a bruise or blemish on fruit)”), wherein
each existing inference task is solved by integrating inference results of the plurality of existing inference models that are associated with the existing inference task (Matthew (¶0034 line 3-15), “In this example, there are two pre-trained neural networks 302, 306, which each produce respective results 304, 308 based on the common inputs (existing inference task) provided to the neural networks 302, 306. These results 304, 308 are concatenated and combined 310, then those concatenated, combined results are input into an additional neural network 312. From that additional neural network 312, the system produces new results "Result C" 314, which were not found by either of the two initial neural networks 302, 306.”), and
for specifying priorities of inference results of the plurality of existing inference models for each existing inference task, when integrating the inference results of the existing inference models (Matthew (¶0026 last 11 lines), model producing the higher number of false positive can be weighted lower when making the ultimate categorization of the system);
a plurality of object inference models is generated based on object learning data obtained by the plurality of different sensor groups in the object environment such that the object inference models of the generated plurality of object inference models are respectively configured to carry out object related tasks related to the object inference task (Matthew (¶0033 last 7 lines), with the features originally identified by the pre-trained neural networks 202, 204, 206, and with the newly identified features identified by the additional neural network, the system can identify and classify the objects within the image),
the object inference task is solved by integrating inference results of the plurality of object inference models (Matthew (¶0034 line 6-11, fig. 4), result 304 and 308, from the first and second model, are concatenated and combined to produce new “Result C” 314), and
the determined plurality of object weights of the object weight set specify, when integrating the inference results of the plurality of object inference models, priorities of inference results of each of the plurality of object inference models (Matthew (¶0026 last 11 lines), “one model produces false positive thirty percent of the time and the other four model produce false positives less than 20 percent of the time”. “Model producing the higher number of false positives can be weighted lower when making the ultimate categorization of the system”).
Matthew may not explicitly disclose:
each existing inference task is associated with a plurality of existing inference models, for each existing inference task, the plurality of existing inference models being generated based on sets of existing learning data respectively obtained by a plurality of different sensor groups in a plurality of different learning environments such that plurality of existing inference models are respectively configured to carry out existing related tasks related to the existing inference task,
Colt teaches:
each existing inference task is associated with a plurality of existing inference models, for each existing inference task, the plurality of existing inference models being generated based on sets of existing learning data respectively obtained by a plurality of different sensor groups in a plurality of different learning environments (Colt (¶0064 line 3-9, ¶00663-10), sensor readings are collected for the first and second manufacturing area) such that plurality of existing inference models are respectively configured to carry out existing related tasks related to the existing inference task (Colt (¶0065 line 3-6, ¶0067 line 3-6), first model is generated for the first manufacture area. Second model is generated for second manufacture area),
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify an ensemble model of Matthew instead be an ensemble model taught by Colt, with a reasonable expectation of success. The motivation would be to provide a “linkage is determined by a high-level overview of activities/conditions within the item being manufactured, which was not available when the lower-level first and second models were being created” (Colt (¶0068 last 4 lines)).
Matthew in view of Colt may not explicitly disclose:
the historical existing task information indicates a plurality of existing weight sets that are each constituted by a plurality of existing weights and that have been respectively determined for the plurality of existing inference tasks (Hwang (¶0110 line 1-6), “it is preferable to perform the selective re-learning of the model by re-learning only the weight that is influenced by the new task”),
operation as a degree of association calculating unit configured to calculate a degree of association between an object inference task in an object environment and each of the plurality of existing inference tasks in the different learning environments (Hwang (¶0112 line 4-7), “if the new task arrives at the model, it is suitable for a sparse linear model for predicting task t by using the highest hidden unit of the neural network by the following Equation 2.”), according to similarities in task objectives between the object inference task and the plurality of existing inference tasks (Hwang (¶0113 line 5-8), “The units and the weights within the network affected by the learning may be identified by constructing the sparse connection at the layer.”); and
operation as a weight determining unit configured to determine a plurality of object weights constituting an object weight set, according to the calculated degrees of association, from the plurality of existing weights of each of the plurality of existing weight sets that are respectively associated with the plurality of existing inference tasks different from the object inference task (Hwang (¶0113 line 8-11), “it is possible to identify all units (and input features) having a path by performing a breadth-first search in the network, starting at the selected node. Next, only the weights of the selected sub-network S(W s') may be learned.”), wherein
Hwang teaches:
the historical existing task information indicates a plurality of existing weight sets that are each constituted by a plurality of existing weights and that have been respectively determined for the plurality of existing inference tasks (Hwang (¶0110 line 1-6), “it is preferable to perform the selective re-learning of the model by re-learning only the weight that is influenced by the new task”),
operation as a degree of association calculating unit configured to calculate a degree of association between an object inference task in an object environment and each of the plurality of existing inference tasks in the different learning environments (Hwang (¶0112 line 4-7), “if the new task arrives at the model, it is suitable for a sparse linear model for predicting task t by using the highest hidden unit of the neural network by the following Equation 2.”), according to similarities in task objectives between the object inference task and the plurality of existing inference tasks (Hwang (¶0113 line 5-8), “The units and the weights within the network affected by the learning may be identified by constructing the sparse connection at the layer.”); and
operation as a weight determining unit configured to determine a plurality of object weights constituting an object weight set, according to the calculated degrees of association, from the plurality of existing weights of each of the plurality of existing weight sets that are respectively associated with the plurality of existing inference tasks different from the object inference task (Hwang (¶0113 line 8-11), “it is possible to identify all units (and input features) having a path by performing a breadth-first search in the network, starting at the selected node. Next, only the weights of the selected sub-network S(W s') may be learned.”), wherein
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify an ensemble model of Matthew in view of Colt instead be a comparison between new task and old task for updating a subset of weight, with a reasonable expectation of success. The motivation would be to solve the problem of new task as “The simplest way to train a model for a series of tasks is to re-learn the entire model each time a new task arrives. However, the retraining may be costly for the deep neural networks. Therefore, it is preferable to perform the selective re-learning of the model by re-learning only the weight that is influenced by the new task” (Hwang (¶0110 line 1-6)).
As Claim 2, besides claim 1, Matthew in view of Colt in further view of Hwang teaches wherein the processor is configured with the program to perform operations further comprising operation as an output unit configured to output weight information indicating the object weight set, to an inference apparatus that uses the plurality of object inference models (Matthew (¶0026 last 11 lines), weights of the models are modified based on accuracy, speed or efficiency of the respective models).
As Claim 7, besides Claim 1, Matthew in view of Colt in further view of Hwang teaches wherein integrating the inference results of the object inference models is constituted by comprises performing a weighted majority decision on the inference results of the plurality of object inference models (Matthew (¶0026 last 11 lines), “one model produces false positive thirty percent of the time and the other four model produce false positives less than 20 percent of the time”. “Model producing the higher number of false positives can be weighted lower when making the ultimate categorization of the system”).
As Claim 8, besides Claim 1, Matthew in view of Colt in further view of Hwang teaches wherein the inference results of the object inference models are constituted by comprise numerical values, and integrating the inference results of the object inference models is constituted by comprises weighting inference results of the plurality of object inference models according to the plurality of object weights, and calculating an average or a total sum of the weighted inference results (Matthew (¶0026 last 11 lines), “one model produces false positive thirty percent of the time and the other four model produce false positives less than 20 percent of the time”. “Model producing the higher number of false positives can be weighted lower when making the ultimate categorization of the system”).
As Claim 9, besides Claim 1, Matthew in view of Colt in further view of Hwang teaches wherein the object inference models include comprise one or more computation parameters to be used in a shared computation of the object related tasks, and integrating the inference results of the object inference models is constituted by comprises weighting values of the one or more computation parameters of the object inference models according to the object weights, and calculating an average or a total sum of values of the weighted one or more computation parameters of the object inference models (Matthew (¶0026 last 11 lines), “one model produces false positive thirty percent of the time and the other four model produce false positives less than 20 percent of the time”. “Model producing the higher number of false positives can be weighted lower when making the ultimate categorization of the system”).
As Claim 10, Matthew teaches an inference apparatus comprising a processor configured with a program to perform operations comprising:
operation as a data acquiring unit configured to acquire object data (Matthew (¶0033 line 1), an image can be input);
operation as an output unit configured to output information regarding a result of solving the object inference task (Matthew (¶0034 line 6-11, fig. 4), result 304 and 308, from the first and second model, are concatenated and combined to produce new “Result C” 314).
The rest of the Claim is rejected for the same reasons as Claim 1.
As Claim 11 and 12, the Claims are rejected for the same reasons as Claim 1.
As Claim 21, besides Claim 1, Matthew in view of Colt in further view of Hwang teaches wherein the task objectives include at least one of a type of object to be inspected, a type of defect to be detected, and a location-related attribute including a production line or factory (Colt (¶0065 line 3-6, ¶0067 line 3-6), first model is generated for the first manufacture area. Second model is generated for second manufacture area).
Claim(s) 3 and 13 is/are rejected under 35 U.S.C. 103 as being unpatentable over Matthew and Colt in view of Hwang in further view of Kitsunezuka (U.S. 2019/0028215 hereinafter Kitsunezuka).
As Claim 3, Matthew in view of Colt in further view of Hwang may not explicitly disclose:
wherein the processor is configured with the program to perform operations such that operation as the degree of association calculating unit comprises calculating the degrees of association from distances between locations of inference points in the different learning environment related to the existing inference tasks and a location of an inference point in the object environment related to the object inference task.
Kitsunezuka teaches:
wherein the processor is configured with the program to perform operations such that operation as the degree of association calculating unit comprises calculating the degrees of association from distances between locations of inference points in the different learning environment related to the existing inference tasks and a location of an inference point in the object environment related to the object inference task (Kitsunezuka (¶0044 line 18-21), The weighting coefficient calculated by the weighting coefficient calculation unit 0216 is smaller in value as a distance between the estimation location and the observation location is longer, and is larger in value as the impact level is larger).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify an weight calculation of Matthew in view of Colt in further view of Hwang instead be a weight taught by Kitsunezuka, with a reasonable expectation of success. The motivation would be to “create a model using data at a multitude of observation locations. Therefore, even when the observation amount of a sensor is largely deviated from the actual value of the estimation location due to the impact of an obstacle, Kriging method can decrease the impact on the final estimation impact” (Kitsunezuka (¶0005 last 10 lines)).
As Claim 13, the Claim is rejected for the same reasons as Claim 3.
Claim(s) 4 and 14 is/are rejected under 35 U.S.C. 103 as being unpatentable over Matthew and Colt in view of Hwang in further view of Clarke et al. (U.S. 4,863,268 hereinafter Clarke).
As Claim 4, besides Claim 1, Matthew in view of Colt in further view of Hwang does not explicitly disclose:
wherein the processor is configured with the program to perform operations such that operation as the degree of association calculating unit comprises calculating the degrees of association from semantic similarities between terms that respectively indicate the existing inference tasks and the object inference task, the semantic similarities being measured by language processing.
Clarke teaches:
wherein the processor is configured with the program to perform operations such that operation as the degree of association calculating unit comprises calculating the degrees of association from semantic similarities between terms that respectively indicate the existing inference tasks and the object inference task, the semantic similarities being measured by language processing (Clarke (col. 7 line 4-8), defect type is entered using voice recognition).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine input of Matthew in view of Colt in further view of Hwang instead be a voice input taught by Clarke, with a reasonable expectation of success. The motivation would be to provide “an easier type of analysis than one which has to find and quantify the defects” (Clarke (col. 7 line 9-11)).
As Claim 14, the Claim is rejected for the same reasons as Claim 4.
Claim(s) 5-6, 16 and 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Matthew and Colt in view of Hwang in further view of Ishii et al. (U.S. 2016/0360966 hereinafter Ishii).
As Claim 5, besides Claim 1, Matthew in view of Colt in further view of Hwang does not explicitly disclose:
wherein the processor is configured with the program to perform operations such that operation as the degree of association calculating unit comprises calculating the degrees of association from degrees of similarity between material objects respectively related to the existing inference tasks and the object inference task.
Ishii teaches:
wherein the processor is configured with the program to perform operations such that operation as the degree of association calculating unit comprises calculating the degrees of association from degrees of similarity between material objects respectively related to the existing inference tasks and the object inference task (Ishii (¶0394 line 11-ends), system compares light quantity distribution of test object to the light quantity distribution of the optical models in order to select a suited model).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to replace object condition of Matthew in view of Colt in further view of Hwang instead be a object-environment and model-environment taught by Ishii, with a reasonable expectation of success. The motivation would be to allow “the error in the amounts of the detection light due to a different in the state of installation of four probes that are adjacent to each other […] can be reduced” (Ishii (¶0393)).
As Claim 6, besides Claim 1, Matthew in view of Colt in further view of Hwang does not explicitly disclose:
wherein the processor is configured with the program to perform operations such that operation as the degree of association calculating unit comprises calculating the degrees of association from degrees of similarity between environments in which the existing inference tasks and the object inference task are respectively carried out.
Ishii teaches:
wherein the processor is configured with the program to perform operations such that operation as the degree of association calculating unit comprises calculating the degrees of association from degrees of similarity between environments in which the existing inference tasks and the object inference task are respectively carried out (Ishii (¶0394 line 11-ends), system compares light quantity distribution of test object to the light quantity distribution of the optical models in order to select a suited model).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to replace object condition of Matthew in view of Colt in further view of Hwang instead be a object-environment and model-environment taught by Ishii, with a reasonable expectation of success. The motivation would be to allow “the error in the amounts of the detection light due to a different in the state of installation of four probes that are adjacent to each other […] can be reduced” (Ishii (¶0393)).
As Claim 16, the Claims are rejected for the same reasons as Claim 5.
As Claim 19, the Claims are rejected for the same reasons as Claim 6.
Claim(s) 15 is/are rejected under 35 U.S.C. 103 as being unpatentable over Matthew, Colt and Hwang in view of Kisunezuka in further view of Clarke.
As Claim 15, besides Claim 3, Matthew and Colt in view of Hwang in further view of Kitsunezuka does not explicitly disclose:
wherein the processor is configured with the program to perform operations such that operation as the degree of association calculating unit comprises calculating the degrees of association from semantic similarities between terms that respectively indicate the existing inference tasks and the object inference task, the semantic similarities being measured by language processing.
Clarke teaches:
wherein the processor is configured with the program to perform operations such that operation as the degree of association calculating unit comprises calculating the degrees of association from semantic similarities between terms that respectively indicate the existing inference tasks and the object inference task, the semantic similarities being measured by language processing (Clarke (col. 7 line 4-8), defect type is entered using voice recognition).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine input of Matthew and Colt in view of Hwang in further view of Kitsunezuka instead be a voice input taught by Clarke, with a reasonable expectation of success. The motivation would be to provide “an easier type of analysis than one which has to find and quantify the defects” (Clarke (col. 7 line 9-11)).
Claim(s) 17 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Matthew, Colt and Hwang in view of Kisunezuka in further view of Ishii.
As Claim 17, besides Claim 3, Matthew and Colt in view of Hwang in further view of Kitsunezuka does not explicitly disclose:
wherein the processor is configured with the program to perform operations such that operation as the degree of association calculating unit comprises calculating the degrees of association from degrees of similarity between material objects respectively related to the existing inference tasks and the object inference task.
Ishii teaches:
wherein the processor is configured with the program to perform operations such that operation as the degree of association calculating unit comprises calculating the degrees of association from degrees of similarity between material objects respectively related to the existing inference tasks and the object inference task (Ishii (¶0394 line 11-ends), system compares light quantity distribution of test object to the light quantity distribution of the optical models in order to select a suited model).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to replace object condition of Matthew and Colt in view of Hwang in further view of Kitsunezuka instead be a object-environment and model-environment taught by Ishii, with a reasonable expectation of success. The motivation would be to allow “the error in the amounts of the detection light due to a different in the state of installation of four probes that are adjacent to each other […] can be reduced” (Ishii (¶0393)).
As Claim 20, besides Claim 3, Matthew and Colt in view of Hwang in further view Kitsunezuka does not explicitly disclose:
wherein the processor is configured with the program to perform operations such that operation as the degree of association calculating unit comprises calculating the degrees of association from degrees of similarity between environments in which the existing inference tasks and the object inference task are respectively carried out.
Ishii teaches:
wherein the processor is configured with the program to perform operations such that operation as the degree of association calculating unit comprises calculating the degrees of association from degrees of similarity between environments in which the existing inference tasks and the object inference task are respectively carried out (Ishii (¶0394 line 11-ends), system compares light quantity distribution of test object to the light quantity distribution of the optical models in order to select a suited model).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to replace object condition of Matthew and Colt in view of Hwang in further view of Kitsunezuka instead be a object-environment and model-environment taught by Ishii, with a reasonable expectation of success. The motivation would be to allow “the error in the amounts of the detection light due to a different in the state of installation of four probes that are adjacent to each other […] can be reduced” (Ishii (¶0393)).
Claim(s) 18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Matthew, Colt and Hwang in view of Clark in further view of Ishii.
As Claim 18, besides Claim 4, Matthew and Colt in view of Hwang in further view of Clark does not explicitly disclose:
wherein the processor is configured with the program to perform operations such that operation as the degree of association calculating unit comprises calculating the degrees of association from degrees of similarity between material objects respectively related to the existing inference tasks and the object inference task.
Ishii teaches:
wherein the processor is configured with the program to perform operations such that operation as the degree of association calculating unit comprises calculating the degrees of association from degrees of similarity between material objects respectively related to the existing inference tasks and the object inference task (Ishii (¶0394 line 11-ends), system compares light quantity distribution of test object to the light quantity distribution of the optical models in order to select a suited model).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to replace object condition of Matthew and Colt in view of Hwang in further view of Clark instead be a object-environment and model-environment taught by Ishii, with a reasonable expectation of success. The motivation would be to allow “the error in the amounts of the detection light due to a different in the state of installation of four probes that are adjacent to each other […] can be reduced” (Ishii (¶0393)).
Response to Arguments
Response to Claim Rejections under 35 U.S.C. §112(b)
As Claims 1, 10, 11 and 12, Applicant argue that the amended claims would clear the antecedent basis issues (fourth paragraph of page 12 in the remarks).
Applicant’s arguments are not persuasive. Applicant did not address 35 U.S.C. §112(b) rejection. Therefore, 35 U.S.C. §112(b) rejections are not withdrawn.
Response to Claim Rejections under 35 U.S.C. §103
As Claims 1, 11 and 12, Applicant argues that cited references do not disclose “the historical existing task information …”, “calculating a degree of association …”, “determining a plurality of object weights …” (first paragraph of page 15 in the remarks).
Applicant’s arguments are moot because new reference Hwang teaches the limitation(s).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to NHAT HUY T NGUYEN whose telephone number is (571)270-7333. The examiner can normally be reached M-F: 12:00-8:00 EST.
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, Viker Lamardo can be reached at 571-270-5871. 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.
/NHAT HUY T NGUYEN/Primary Examiner, Art Unit 2147
Prosecution Insights