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 .
Response to Arguments
Applicant's arguments filed 08/28/2025 have been fully considered but they are not persuasive.
Regarding the 101 rejections, on pages 10-11 of “Remarks” applicant contends that the amended claim 1 does not recite abstract ideas under Step 2A Prong 1 because of the mention of the claimed limitations being performed on a specific machine that includes a processor, memory, and three models. The examiner respectfully disagrees. The limitations relating to learning the first and second models are claimed in such a broad way that, under the broadest reasonable interpretation, includes a step of evaluation and judgement and could be performed mentally or with pen and paper like adjusting model parameters using a simple model with an input output function, which is either a mental process of evaluation/judgement (MPEP 2106)). Similarly, the same mental processes are applied to learning the pseudo first model and learning the pseudo second model as these models are interpreted as being the same models as the first and second models but just given a pseudo data label. The limitations relating to the calculating an improvement degree using an improvement degree model are also claimed in such a broad way that, under the broadest reasonable interpretation, includes a step of evaluation and judgement and could be performed mentally or with pen and paper like adjusting an improvement model’s parameters by comparing outputs of multiple models, which is either a mental process of evaluation/judgement (MPEP 2106)). Lastly, the additional elements of the memory and processor, under the broadest reasonable interpretation, merely recites steps that apply generic computer components to perform judicial exceptions, which represents merely adding the words “apply it”, or an equivalent, which are not indicative of an inventive concept (MPEP 2106.05(f)).
On pages 11-12 of “Remarks” applicant contends that the amended claim 1 provides a practical application under Step 2A Prong 2. Specifically, provides a technical improvement to computer technology by reducing the processing load of the processor. The examiner respectfully disagrees. Applicant cites “As a result, the model that outputs the improvement degree can be learned in advance, and therefore it is not necessary to calculate the improvement degree by learning the first model and the second model each time new data is input, and the processing load of model learning can be reduced” from paragraph 142 of the Specification as support for the technical improvement. However, the claims do not reflect this proposed improvement of learning the improvement degree model in advance to prevent the calculating the improvement degree of the first and second models. Specifically, the amended claim 1 does not appear to have support for the pre-learning the improvement degree model as well as mentioning that this pre-learning reduces a computational load.
On pages 12-13 of “Remarks” applicant contends that the amended claim 1 recites additional elements that are not well understood, routine, or conventional activities under Step 2B. The examiner respectfully disagrees. As discussed above, the amended limitations of claim 1 still recite mental process abstract ideas. Additionally, the mention of performing the identified abstract ideas using a processor and memory, under the broadest reasonable interpretation, merely recite steps that apply generic computer components to perform an abstract idea which represents merely adding the words “apply it”, or an equivalent, which are not indicative of an inventive concept (MPEP 2106.05(f)). Under Step 2B, the courts have found that adding the words “apply it”, or an equivalent, with the judicial exception does not qualify as significantly more under Step 2B (MPEP 2106.05). Therefore, applicant’s arguments regarding the 101 rejections are not persuasive.
Regarding the 103 rejections, applicant’s arguments about reference(s) Li, Yuan, and Okamoto have been fully considered but are not persuasive.
Alleged no teaching of learning an improvement degree model based on the pseudo first model, the pseudo second model, and the pseudo-improvement degree
In Remarks/Arguments pg. 14-15, applicant contends:
“Yuan describes "[t]he ML controller 407 includes a component identification module 413 and an importance module 415. The component identification module 413 is configured to identify components within a received input (set of observed data). The importance module 415 is configured to importance of each component based on the change in the confidence measure after the component has been adjusted or perturbed. This may be through the importance module inputting the adjusted input into a machine learning model itself, or by the importance module sending the adjusted input to an external system that calculates and returns the corresponding confidence measure." See Yuan at [0117].
Yuan describes that the ML controller 407 includes the importance module 415 that determines the importance of each component by inputting the adjusted input into the machine learning model. However, Yuan does not describe learning the machine learning model based on a pseudo-first model, a pseudo-second model, and a pseudo- improvement degree.
Further, Okamoto describes "[t]he training the first learning model through machine learning includes training the first generator to generate, from the first label, a pseudo sample fitting the first sample for each of the plurality of first learning datasets. The second trainer trains a second learning model including a second generator through machine learning using the obtained plurality of second learning datasets." See Okamoto at [0021].
Okamoto at best describes training the first generator to generate, from the first label, the pseudo sample fitting the first sample for each of the plurality of first learning datasets. However, Okamoto does not describe learning an improvement degree model based on a pseudo-first model, a pseudo-second model, and a pseudo-improvement degree.
Li does not remedy the above noted deficiency of Yuan and Okamoto.
Therefore, the Applicant submits that the combination of Li, Yuan, and Okamoto does not teach, suggest or render obvious the feature of "learn an improvement degree model based on the pseudo-first model, the pseudo-second model, and the pseudo- improvement degree," as recited in amended independent claim 1.”
The relevant claim limitations appear to be: learn an improvement degree model based on the pseudo-first model, the pseudo-second model, and the pseudo-improvement degree; in claim 1. As noted in the previous Office Action, Li, Yuan, and Okamoto teaches:
(Li, ⁋4, “the inference computing apparatus to perform the following operations: receiving a first inference model from a model training apparatus, the first inference model being obtained by the model training apparatus through a model training based on a first training sample library”).
(Li, ⁋6, “In some embodiments of the present disclosure, updating the first inference model includes: performing a model training based on a second training sample library to obtain a second inference model, or sending a model update request to the model training apparatus to obtain the second inference model, the second training sample library includes training samples from historical data and/or training samples that are from the inference result and subjected to a re-judging”).
(Yuan, ⁋117, “The importance module 415 is configured to importance of each component based on the change in the confidence measure after the component has been adjusted or perturbed. This may be through the importance module inputting the adjusted input into a machine learning model itself, or by the importance module sending the adjusted input to an external system that calculates and returns the corresponding confidence measure.”, and Yuan, ⁋14, “obtaining a measure of confidence in a second prediction, the second prediction being generated through inputting the adjusted input into the machine learning model; and determining the influence of the component on the first prediction by calculating a difference between the measure of confidence in the first prediction and the measure of confidence in the second prediction.”).
(Okamoto, ⁋21, “the first generator to generate, from the first label, a pseudo sample fitting the first sample for each of the plurality of first learning datasets.”).
As noted in the previous office action, Li teaches training a first and second model on a base dataset and an augmented dataset. Yuan was relied upon to teach an improvement model that determines the improvement degree between a first prediction and a second prediction. Specifically, the importance module is interpreted as the improvement degree model as it determines the adjusted components that affect the predictions between a first model prediction and a second model prediction. Under the broadest reasonable interpretation, determining the adjusted components which affect the confidence measures of the first and second models are interpreted as learning an improvement degree model. While Li in view of Yuan teaches learning an improvement degree model based on the first model, the second model, and the improvement degree, the combination does not explicitly teach the pseudo element. As noted in the 101 analysis, the pseudo first model and the pseudo second model are interpreted as being the same models as the first and second models but just given pseudo data label. Okamoto was relied upon to teach using pseudo data when training models. Thus, the combination of Li, Yuan, and Okamoto teaches learning an improvement degree model based on the pseudo-first model, the pseudo-second model, and the pseudo-improvement degree. Therefore, applicant’s arguments are not persuasive.
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 1-2 and 4-14 are rejected under 35 U.S.C 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding claim 1 and analogous claim 13, in step 1 of the 101 analysis set forth in MPEP 2106, the claim recites An information processing apparatus comprising:. The claim recites an apparatus. An apparatus is one of the four statutory categories of invention.
In Step 2A, Prong 1 of the 101 analysis set forth in MPEP 2106, the examiner has determined that the following limitations recite a process that, under broadest reasonable interpretation, covers a mental process or mathematical concept but for the recitation of generic computer components:
learn a first model based on new data from a terminal device held by a user, wherein the shared data corresponds to additional data to the new data; join the shared data with the new data to obtain joined data; (i.e., the broadest reasonable interpretation includes a step of evaluation and judgement and could be performed mentally or with pen and paper like changing parameters of a model based on the model’s performance given certain data, which is either a mental process of evaluation/judgement (MPEP 2106)).
learn a second model based on the joined data; (i.e., the broadest reasonable interpretation includes a step of evaluation and judgement and could be performed mentally or with pen and paper like changing parameters of a model based on the model’s performance given certain data, which is either a mental process of evaluation/judgement (MPEP 2106)).
select pseudo-new data and pseudo-additional data from the shared data stored in the memory; (i.e., the broadest reasonable interpretation includes a step of evaluation and judgement and could be performed mentally or with pen and paper like selecting data to train a model with, which is either a mental process of evaluation/judgement (MPEP 2106)).
learn a pseudo-first model based on the pseudo-new data; learn a pseudo-second model based on the pseudo-additional data; (i.e., the broadest reasonable interpretation includes a step of evaluation and judgement and could be performed mentally or with pen and paper like changing parameters of a model based on the model’s performance given certain data, which is either a mental process of evaluation/judgement (MPEP 2106)).
calculate a pseudo-improvement degree based on the pseudo-first model and the pseudo-second model; learn an improvement degree model based on the pseudo-first model, the pseudo-second model, and the pseudo-improvement degree; (i.e., the broadest reasonable interpretation includes a step of evaluation and judgement and could be performed mentally or with pen and paper like comparing multiple models’ outputs to each other, which is either a mental process of evaluation/judgement (MPEP 2106)).
calculate, via the improvement degree model, an improvement degree indicating a degree of improvement in output precision of the second model compared to output precision of the first model; (i.e., the broadest reasonable interpretation includes a step of evaluation and judgement and could be performed mentally like comparing the output of the second model to the first model and determining whether the second model output is better than the first model output, which is either a mental process of evaluation/judgement (MPEP 2106)).
and generate presentation information based on the improvement degree. (i.e., the broadest reasonable interpretation includes a step of evaluation and judgement and could be performed mentally like creating an visual representation of which model output is better, which is either a mental process of evaluation/judgement (MPEP 2106)).
If the claim limitations, under their broadest reasonable interpretation, covers activities classified under Mental processes: concepts performed in the human mind (including observation, evaluation, judgement, or opinion) (see MPEP 2106.04(a)(2), subsection (III)) or Mathematical concepts: mathematical relationships, mathematical formulas or equations, or mathematical calculations (see MPEP 2106.04(a)(2), subsection (I)). Accordingly, the claim recites an abstract idea.
In Step 2A, Prong 2 of the 101 analysis, set forth in MPEP 2106, the examiner has determined that the following additional elements do not integrate this judicial exception into a practical application:
…a memory configured to store shared data; and a processor configured to:… (i.e., the generic computer components recited in this limitation merely add the words “apply it”, or an equivalent, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f))).
Since the claim does not contain any other additional elements, that amount to integration into a practical application, the claim is directed to an abstract idea.
In Step 2B of the 101 analysis set forth in the 2019 PEG, the examiner has determined that the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception:
Regarding limitation (VIII), under the broadest reasonable interpretation, merely recites steps that apply generic computer components to perform judicial exceptions, which represents merely adding the words “apply it”, or an equivalent, which are not indicative of an inventive concept (MPEP 2106.05(f)). Considering additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea. Therefore, the claim is not patent eligible.
Regarding claim 2, it is dependent upon claim 1 and fails to resolve the deficiencies identified above by integrating the judicial exception into a practical application, or introducing significantly more than the judicial exception. For example, claim 2 recites wherein the processor is further configured to join the shared data having similar features to the new data. Under the broadest reasonable interpretation, the limitations recite combining data with similar features which is a step of evaluation and judgement which can be performed mentally or with pen and paper. The steps evaluation and judgement are mental processes thus, claim 2 does not solve the deficiencies of claim 1.
Regarding claim 4, it is dependent upon claim 1 and fails to resolve the deficiencies identified above by integrating the judicial exception into a practical application, or introducing significantly more than the judicial exception. For example, claim 4 recites wherein the processor is further configured to learn the improvement degree model based on an explanatory variable that includes features of the pseudo-new data and features of the pseudo-additional data, and an objective variable that includes the pseudo-improvement degree. Under the broadest reasonable interpretation, the limitations merely recite steps of mere data gathering of data to train a model with, which has been recognized by the courts as being well-understood, routine, and conventional functions. Specifically, the courts have recognized computer functions directed to mere data gathering as well-understood, routine, and conventional functions when they are claimed in a merely generic manner or as insignificant extra-solution activity (MPEP 2106.05(g)). Therefore, claim 4 does not solve the deficiencies of claim 1.
Regarding claim 5, it is dependent upon claim 4 and fails to resolve the deficiencies identified above by integrating the judicial exception into a practical application, or introducing significantly more than the judicial exception. For example, claim 5 recites wherein the explanatory variable further includes information regarding a behavior history of the user for the pseudo-new data and the pseudo-additional data. Under the broadest reasonable interpretation, the limitations merely recite steps of mere data gathering of data to train a model with, which has been recognized by the courts as being well-understood, routine, and conventional functions. Specifically, the courts have recognized computer functions directed to mere data gathering as well-understood, routine, and conventional functions when they are claimed in a merely generic manner or as insignificant extra-solution activity (MPEP 2106.05(g)). Therefore, claim 5 does not solve the deficiencies of claim 4.
Regarding claim 6, it is dependent upon claim 1 and fails to resolve the deficiencies identified above by integrating the judicial exception into a practical application, or introducing significantly more than the judicial exception. For example, claim 6 recites wherein the processor is further configured to generate the additional data having the improvement degree satisfying a specific condition, as the presentation information. Under the broadest reasonable interpretation, the limitations recite creating data that has a high chance of improving the model performance which is a step of evaluation and judgement which can be performed mentally or with pen and paper. The steps evaluation and judgement are mental processes thus, claim 6 does not solve the deficiencies of claim 1.
Regarding claim 7, it is dependent upon claim 6 and fails to resolve the deficiencies identified above by integrating the judicial exception into a practical application, or introducing significantly more than the judicial exception. For example, claim 7 recites wherein the processor is further configure to generate, in a case of generation of a plurality of pieces of the additional data as the presentation information, recommendation level information based on the improvement degree. Under the broadest reasonable interpretation, the limitations recite creating data that has a recommendation level for a high chance of improving the model performance which is a step of evaluation and judgement which can be performed mentally or with pen and paper. The steps evaluation and judgement are mental processes thus, claim 7 does not solve the deficiencies of claim 6.
Regarding claim 8, it is dependent upon claim 6 and fails to resolve the deficiencies identified above by integrating the judicial exception into a practical application, or introducing significantly more than the judicial exception. For example, claim 8 recites wherein the processor is further configured to generate, in a case of generation of the additional data as the presentation information, appending information regarding the additional data. Under the broadest reasonable interpretation, the limitations recite creating additional data that has a high chance of improving the model performance which is a step of evaluation and judgement which can be performed mentally or with pen and paper. The steps evaluation and judgement are mental processes thus, claim 8 does not solve the deficiencies of claim 6.
Regarding claim 9, it is dependent upon claim 1 and fails to resolve the deficiencies identified above by integrating the judicial exception into a practical application, or introducing significantly more than the judicial exception. For example, claim 9 recites wherein the new data and the additional data are time-series data, and the processor is further configured to correct, in a case where the new data and the additional data are inconsistent in time-series, the additional data to be consistent with the time-series of the new data. Under the broadest reasonable interpretation, merely recite steps that amount to indicating a field of use or technological environment in which to apply a judicial exception (MPEP 2106.05(h)). Additionally, under the broadest reasonable interpretation, the limitations recite correcting outlier data which is a step of evaluation and judgement which can be performed mentally or with pen and paper. The steps evaluation and judgement are mental processes thus, claim 9 does not solve the deficiencies of claim 1.
Regarding claim 10, it is dependent upon claim 1 and fails to resolve the deficiencies identified above by integrating the judicial exception into a practical application, or introducing significantly more than the judicial exception. For example, claim 10 recites wherein the processor is further configured to acquire the new data as the shared data from the terminal device. Under the broadest reasonable interpretation, the limitations merely recite steps of mere data gathering of data, which has been recognized by the courts as being well-understood, routine, and conventional functions. Specifically, the courts have recognized computer functions directed to mere data gathering as well-understood, routine, and conventional functions when they are claimed in a merely generic manner or as insignificant extra-solution activity (MPEP 2106.05(g)). Therefore, claim 10 does not solve the deficiencies of claim 1.
Regarding claim 11, it is dependent upon claim 10 and fails to resolve the deficiencies identified above by integrating the judicial exception into a practical application, or introducing significantly more than the judicial exception. For example, claim 11 recites wherein the processor is further configured to: execute an anonymization process on the shared data satisfying a specific condition; and stores the shared data in the memory based on the anonymization process. Under the broadest reasonable interpretation, the limitations merely recite steps of mere data gathering of data, which has been recognized by the courts as being well-understood, routine, and conventional functions. Specifically, the courts have recognized computer functions directed to mere data gathering as well-understood, routine, and conventional functions when they are claimed in a merely generic manner or as insignificant extra-solution activity (MPEP 2106.05(g)). Therefore, claim 11 does not solve the deficiencies of claim 10.
Regarding claim 12, it is dependent upon claim 1 and fails to resolve the deficiencies identified above by integrating the judicial exception into a practical application, or introducing significantly more than the judicial exception. For example, claim 12 recites wherein the processor is further configured to: execute a specific process on the additional data; and join the additional data subjected to the specific process with the new data. Under the broadest reasonable interpretation, the limitations recite combining data which is a step of evaluation and judgement which can be performed mentally or with pen and paper. The steps evaluation and judgement are mental processes thus, claim 12 does not solve the deficiencies of claim 1.
Regarding claim 14, in step 1 of the 101 analysis set forth in MPEP 2106, the claim recites A non-transitory computer-readable medium. The claim recites a computer-readable medium which is interpreted as an article of manufacture. An article of manufacture is one of the four statutory categories of invention. For the Step 2A/2B analyses, since claim 14 is analogous to claim 1 it is rejected under the same rationales as claim 1.
The additional limitation below fails to resolve the deficiencies identified above by integrating the judicial exception into a practical application, or introducing significantly more than the judicial exception.
A non-transitory computer-readable medium having stored thereon computer-executable instructions that, when executed by a processor, cause the processor to execute operations, the operations comprising (i.e., the generic computer components recited in this limitation merely add the words “apply it”, or an equivalent, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f))).
Considering additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea. Therefore, the claim is not patent eligible.
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, 4-8, 10, and 12-14 are rejected under 35 U.S.C. 103 as being unpatentable over Li, et al., US Pre-Grant Publication 2021/0209488A1 (“Li”) in view of Yuan, et al., US Pre-Grant Publication 2020/0334492A1 (“Yuan”) and further in view of Okamoto, et al., US Pre-Grant Publication 2022/0300809A1 (“Okamoto”).
Regarding claim 1 and analogous claims 13 and 14, Li discloses:
An information processing apparatus, comprising: a memory configured to store shared data; and a processor configured to: (Li, ⁋4, “The inference computing apparatus includes at least one processor and a memory. The memory has stored therein program instructions that, when executed by the at least one processor, cause the inference computing apparatus to perform the following operations [An information processing apparatus, comprising: a memory… and a processor configured to:]”, and Li, ⁋93, “the inference computing apparatus 100 includes the second training sample library, which is stored in the memory 120 [a memory configured to store shared data;].”).
learn a first model based on new data from a terminal device held by a user, (Li, ⁋4, “the inference computing apparatus [An information processing apparatus comprising:] to perform the following operations: receiving a first inference model from a model training apparatus [learn a first model], the first inference model being obtained by the model training apparatus through a model training based on a first training sample library [based on new data],” and Li, ⁋140, “For example, referring to FIG. 5, the model training apparatus 200 further includes a network interface 240. The model training apparatus 200 can communicate with other devices (e.g., user-side device and/or the model training apparatus 200) through the network interface 240 to realize information interaction [from a terminal device held by a user,].”).
wherein the shared data corresponds to additional data to the new data; join the shared data with the new data to obtain joined data; learn a second model based on the joined data; select pseudo-new data and pseudo-additional data from the shared data stored in the memory; (Li, ⁋6, “In some embodiments of the present disclosure, updating the first inference model includes: performing a model training based on a second training sample library to obtain a second inference model [learn a second model based on the joined data;], or sending a model update request to the model training apparatus to obtain the second inference model, the second training sample library [select pseudo-new data and pseudo-additional data from the shared data stored in the memory;] includes training samples from historical data and/or training samples that are from the inference result and subjected to a re-judging; the second training sample having a combination of historical samples and samples subjected to re-judging is interpreted as adding new samples to an original dataset (i.e. wherein the shared data corresponds to additional data to the new data; join the shared data with the new data to obtain joined data;)”).
learn a pseudo-first model based on the pseudo-new data; (Li, ⁋4, “the inference computing apparatus to perform the following operations: receiving a first inference model from a model training apparatus [learn a pseudo-first model], the first inference model being obtained by the model training apparatus through a model training based on a first training sample library [based on the pseudo-new data;]”).
learn a pseudo-second model based on the pseudo-additional data; (Li, ⁋6, “In some embodiments of the present disclosure, updating the first inference model includes: performing a model training based on a second training sample library to obtain a second inference model; training the second model on joined data means that it must also be trained with the additional data (i.e. learn a pseudo-second model based on the pseudo-additional data;), or sending a model update request to the model training apparatus to obtain the second inference model, the second training sample library includes training samples from historical data and/or training samples that are from the inference result and subjected to a re-judging”).
While Li teaches the use of a first and second model where the second model is trained with augmented training data, Li does not explicitly teach:
calculate a pseudo-improvement degree based on the pseudo-first model and the pseudo-second model;
learn an improvement degree model based on the pseudo-first model, the pseudo-second model, and the pseudo-improvement degree;
calculate, via the improvement degree model, an improvement degree indicating a degree of improvement in output precision of the second model compared to output precision of the first model;
and generate presentation information based on the improvement degree.
Yuan teaches:
calculate a pseudo-improvement degree based on the pseudo-first model and the pseudo-second model; (Yuan, ⁋14, “obtaining a measure of confidence in a second prediction, the second prediction being generated through inputting the adjusted input into the machine learning model; and determining the influence of the component on the first prediction by calculating a difference between the measure of confidence [calculate a pseudo-improvement degree] in the first prediction [based on the pseudo-first model] and the measure of confidence in the second prediction [and the pseudo-second model;].”).
learn an improvement degree model based on the pseudo-first model, the pseudo-second model, and the pseudo-improvement degree; (Yuan, ⁋117, “The importance module 415 [learn an improvement degree model] is configured to importance of each component [based on the pseudo-first model, the pseudo-second model,] based on the change in the confidence measure after the component has been adjusted or perturbed. This may be through the importance module inputting the adjusted input into a machine learning model itself, or by the importance module sending the adjusted input to an external system that calculates and returns the corresponding confidence measure.”, and Yuan, ⁋14, “obtaining a measure of confidence in a second prediction, the second prediction being generated through inputting the adjusted input into the machine learning model; and determining the influence of the component on the first prediction by calculating a difference between the measure of confidence [and the pseudo-improvement degree;] in the first prediction [based on the pseudo-first model,] and the measure of confidence in the second prediction [the pseudo-second model,].”).
calculate, via the improvement degree model, an improvement degree indicating a degree of improvement in output precision of the second model compared to output precision of the first model; (Yuan, ⁋14, “obtaining a measure of confidence in a second prediction, the second prediction being generated through inputting the adjusted input into the machine learning model; and determining the influence of the component on the first prediction by calculating a difference between the measure of confidence [calculate, via the improvement degree model, an improvement degree indicating a degree of improvement in output precision] in the first prediction [compared to output precision of the first model;] and the measure of confidence in the second prediction [in output precision of the second model].”).
and generate presentation information based on the improvement degree. (Yuan, ⁋14, “The method further comprises outputting an indication of the influence of one or more of the components [and generate presentation information based on the improvement degree.]”).
Li and Yuan are both in the same field of endeavor (i.e. machine learning). It would have been obvious for a person having ordinary skill in the art before the effective filing date of the claimed invention to combine Li and Yuan to teach the above limitation(s). The motivation for doing so is that comparing the predictions of models on augmented datasets identifies per sample influence on model performance, thus improving the visibility of model performance (cf. Yuan, ⁋10-11, “Whilst it is possible to determine the influence of an embedded feature on a prediction by a machine learning model, this can be difficult to interpret by a user, as the feature may not relate to an interpretable real-world concept. Furthermore, this can require the machine learning model to be retrained, which can be computationally expensive. There is therefore a need for an improved means of identifying the influence of data on machine learning predictions.”).
While the Li in view of Yuan teaches finding an improvement degree between two models with augmented datasets, the combination does not explicitly teach:
using pseudo data.
Okamoto teaches using pseudo data (Okamoto, ⁋21, “the first generator to generate, from the first label, a pseudo sample [pseudo data] fitting the first sample for each of the plurality of first learning datasets.”).
Li, in view of Yuan, and Okamoto are both in the same field of endeavor (i.e. machine learning). It would have been obvious for a person having ordinary skill in the art before the effective filing date of the claimed invention to combine Li, in view of Yuan, and Okamoto to teach the above limitation(s). The motivation for doing so is that creating synthetic data samples improves model robustness as well as reduce data collection costs by providing training samples that have not occurred (cf. Okamoto, ⁋18-19, “In either situation, estimation may be performed with lower accuracy using learning data including any combination of feature types for which no sample is obtained. To improve the accuracy of visual inspection, samples are to be collected thoroughly for all the combinations of the feature types. However, such thorough collection of samples is costly. In response to the above issue, an aspect of the present invention are directed to a technique for reducing the costs for collecting various samples of a predetermined type of data including at least two features.”).
Regarding claim 4, Li in view of Yuan and Okamoto teaches the information processing apparatus according to claim 1. Okamoto also teaches the pseudo data as seen in claim 1. Yuan further teaches wherein the processor is further configured to learn the improvement degree model based on an explanatory variable that includes features of the pseudo-new data and features of the pseudo-additional data, and an objective variable that includes the pseudo-improvement degree. (Yuan, ⁋114, “The methods are therefore an efficient means of providing influence scores and are applicable to any form of machine learning prediction. As the embodiments directly adjust observations [and features of the pseudo-additional data,] prior to their processing for use in the prediction, they are able to provide influence scores for directly observed components [that includes features of the pseudo-new data] that are easy for the end user to understand (relative to machine learning features); the adjusted and original data are interpreted as the explanatory variables as they determine what the influence scores are (i.e. wherein the processor is further configured to learn the improvement degree model based on an explanatory variable). This helps provide an improved means of explaining the origin of machine learning predictions. The influence scores can be used to advise users as to how to improve the machine learning model; the influence score is interpreted as the improvement degree and using the score to improve a model is interpreted as an objective variable (i.e. and an objective variable that includes the pseudo-improvement degree.) or how to achieve improved results (e.g. through editing the observations).”).
Regarding claim 5, Li in view of Yuan and Okamoto teaches the information processing apparatus according to claim 4. Okamoto also teaches the pseudo data as seen in claim 4. Yuan also teaches using wherein the explanatory variable further includes information regarding…for the pseudo-new data and the pseudo-additional data as seen in claim 4. Li further teaches information regarding a behavior history of the user (Li, ⁋16, “the first inference model being obtained by the model training apparatus through a model training based on a first training sample library, the first training sample library including training samples from historical data [information regarding a behavior history of the user] generated in a manufacturing stage,”).
Regarding claim 6, Li in view of Yuan and Okamoto teaches the information processing apparatus according to claim 1. Yuan further teaches wherein the processor is further configured to generate the additional data having the improvement degree satisfying a specific condition, as the presentation information. (Yuan, ⁋17, “Outputting an indication of the influence of one or more components may comprise outputting an indication a set of one or more components (e.g. highlighting the set within the input) and outputting a corresponding influence for each component in the set [wherein the processor is further configured to generate the additional data having the improvement degree]. Alternatively, or in addition, the indication of the influence might be through a ranking or one or more components by influence, or by outputting only a set of one or more of the most influential components and/or a set of one or more of the least influential components [wherein the processor is further configured to generate the additional data having the improvement degree].”).
Regarding claim 7, Li in view of Yuan and Okamoto teaches the information processing apparatus according to claim 6. Yuan further teaches wherein the processor is further configure to generate, in a case of generation of a plurality of pieces of the additional data as the presentation information, recommendation level information based on the improvement degree. (Yuan, ⁋17, “Outputting an indication of the influence of one or more components may comprise outputting an indication a set of one or more components (e.g. highlighting the set within the input) and outputting a corresponding influence for each component in the set [wherein the processor is further configure to generate, in a case of generation of a plurality of pieces of the additional data as the presentation information,]. Alternatively, or in addition, the indication of the influence might be through a ranking or one or more components by influence, or by outputting only a set of one or more of the most influential components and/or a set of one or more of the least influential components [recommendation level information based on the improvement degree.].”).
Regarding claim 8, Li in view of Yuan and Okamoto teaches the information processing apparatus according to claim 6. Yuan further teaches wherein the processor is further configured to generate, in a case of generation of the additional data as the presentation information, appending information regarding the additional data. (Yuan, ⁋72, “Once influence scores have been calculated [wherein the processor is further configured to generate, in a case of generation of the additional data as the presentation information,] for each subgroup, the subgroups are ranked in order of their influence 117 and the ranked list is output to the user. This allows the user to evaluate the influence of the subgroups (observed clusters or components) [appending information regarding the additional data.] within the input upon the prediction.”).
Regarding claim 10, Li in view of Yuan and Okamoto teaches the information processing apparatus according to claim 1. Li further teaches wherein the processor is further configured to acquire the new data as the shared data from the terminal device. (Li, ⁋198, “The original data collecting module 17 is configured to collect the original data from the user-side device 3 to obtain the data to be processed, and send the data to be processed to the inference model module 13 [wherein the processor is further configured to acquire the new data as the shared data from the terminal device.].”).
Regarding claim 12, Li in view of Yuan and Okamoto teaches the information processing apparatus according to claim 1. Li further teaches Yuan further teaches wherein the processor is further configured to: execute a specific process on the additional data; and join the additional data subjected to the specific process with the new data. (Li, ⁋91, “Here, the expression “from the inference result and subjected to a re-judging” described here and below means performing a re-judging on the inference result [wherein the processor is further configured to: execute a specific process on the additional data;] of the first inference model…Then the product images of the display panel marked with a correct result can be obtained after the re-judging of the inference result, and these images can be used as a source of training samples of the second training library [and join the additional data subjected to the specific process with the new data.].”).
Claim 2 is rejected under 35 U.S.C. 103 as being unpatentable over Li, et al., US Pre-Grant Publication 2021/0209488A1 (“Li”) in view of Yuan, et al., US Pre-Grant Publication 2020/0334492A1 (“Yuan”) and further in view of Okamoto, et al., US Pre-Grant Publication 2022/0300809A1 (“Okamoto”) and Goodsitt, et al., US Pre-Grant Publication 2023/0289665A1 (“Goodsitt”).
Regarding claim 2, Li in view of Yuan and Okamoto teaches the information processing apparatus according to claim 1. While the combination teaches joining shared data to the new data as additional data, the combination does not explicitly teach wherein the processor is further configured to join the shared data having similar features to the new data.
Goodsitt teaches wherein the processor is further configured to join the shared data having similar features to the new data. (Goodsitt, ⁋6, “synthetic datasets can be generated and used to train a model. Synthetic datasets can be based on the original datasets, and/or can include information that is similar to the original datasets. While it is beneficial to use synthetic datasets to train models, it is possible that a model trained with a synthetic dataset can produce misclassifications. Some systems attempt to address these misclassifications by feeding the same synthetic dataset back into a model being trained (e.g., along with the original dataset) [wherein the processor is further configured to join the shared data having similar features to the new data.]”).
Li, in view of Yuan and Okamoto, and Goodsitt are both in the same field of endeavor (i.e. machine learning). It would have been obvious for a person having ordinary skill in the art before the effective filing date of the claimed invention to combine Li, in view of Yuan and Okamoto, and Goodsitt to teach the above limitation(s). The motivation for doing so is that adding more similar samples to a dataset improves model robustness (cf. Goodsitt, ⁋5, “original datasets can suffer from a lack of sufficient samples of data to train a model. Problems associated with a small dataset are numerous, but can include (i) over-fitting, which can be more difficult to avoid, and which can result in overfitting the validation set as well, (ii) outliers, which can become much more dangerous, and (iii) noise.”).
Claim 9 is rejected under 35 U.S.C. 103 as being unpatentable over Li, et al., US Pre-Grant Publication 2021/0209488A1 (“Li”) in view of Yuan, et al., US Pre-Grant Publication 2020/0334492A1 (“Yuan”) and further in view of Okamoto, et al., US Pre-Grant Publication 2022/0300809A1 (“Okamoto”) and Danichev, et al., US Pre-Grant Publication 2018/0357261A1 (“Danichev”).
Regarding claim 9, Li in view of Yuan and Okamoto teaches the information processing apparatus according to claim 1. Li further teaches wherein the new data and the additional data are time-series data, (Li, ⁋4, “the first training sample library including training samples from historical data [wherein the new data…are time-series data] generated in a manufacturing stage,” and Li, ⁋6, “the second training sample library includes training samples from historical data [and the additional data are time-series data,] and/or training samples that are from the inference result and subjected to a re-judging;”).
While the combination teaches the use of time-series data for the new and additional data, the combination does not explicitly teach and the processor is further configured to correct, in a case where the new data and the additional data are inconsistent in time-series, the additional data to be consistent with the time-series of the new data.
Danichev teaches and the processor is further configured to correct, in a case where the new data and the additional data are inconsistent in time-series, the additional data to be consistent with the time-series of the new data. (Danichev, ⁋24, “If the alignment determiner 116 determines that two time-series dataset are not aligned [and the processor is further configured to correct, in a case where the new data and the additional data are inconsistent in time-series,], then the smoothness evaluator 118, process selector 120, and dataset aligner 122 may be used to temporally align the time-series datasets [the additional data to be consistent with the time-series of the new data.].”).
Li, in view of Yuan and Okamoto, and Danichev are both in the same field of endeavor (i.e. time-series data). It would have been obvious for a person having ordinary skill in the art before the effective filing date of the claimed invention to combine Li, in view of Yuan and Okamoto, and Danichev to teach the above limitation(s). The motivation for doing so is that aligning time-series data reduces the errors that can arise from having misaligned timestamps (cf. Danichev, see ⁋11-12).
Claim 11 is rejected under 35 U.S.C. 103 as being unpatentable over Li, et al., US Pre-Grant Publication 2021/0209488A1 (“Li”) in view of Yuan, et al., US Pre-Grant Publication 2020/0334492A1 (“Yuan”) and further in view of Okamoto, et al., US Pre-Grant Publication 2022/0300809A1 (“Okamoto”) and Bayardo, et al., Non-Patent Literature “Data Privacy Through Optimal k-Anonymization” (“Bayardo”).
Regarding claim 11, Li in view of Yuan and Okamoto teaches the information processing apparatus according to claim 10. While the combination teaches the use of an acquisition unit and a storage unit, the combination does not explicitly teach performs and stores anonymization processing on the shared data satisfying a predetermined condition in the storage unit.
Bayardo teaches wherein the processor is further configured to: execute an anonymization process on the shared data satisfying a specific condition; and stores the shared data in the memory based