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
Response to Applicant's Arguments Regarding the 101 Rejection
The applicant presents arguments traversing the 101 rejections. The Examiner has fully considered the Applicant's arguments but finds them unpersuasive for the reasons set forth below.
Applicant’s Argument:
The applicant argues (Page 10)
PNG
media_image1.png
471
1268
media_image1.png
Greyscale
Examiner’s Response:
In response, the examiner maintains the rejection. The argument conflates the abstract idea with the environment in which it is performed. Under its broadest reasonable interpretation, the recited identifying of tokens “of a predefined importance” remains an act of evaluation and judgment that can be practically performed in the mind. The added clause “wherein the ML inference is applied to time series data by a time series forecasting or anomaly detection model” does not change that; it identifies the field of use and is an additional element analyzed at Prong Two. A claim does not leave the mental-process or mathematical-concept groupings merely because the data being evaluated comes from, or relates to, a real model. By linking the abstract idea to a field of use i.e. the time series data and time series forecasting or anomaly detection model does not cause the claims to be significantly more than the abstract idea. Therefore, the rejection is maintained as shown below.
Applicant’s Argument:
The applicant argues (Page 10-11)
PNG
media_image2.png
615
1252
media_image2.png
Greyscale
PNG
media_image3.png
614
1258
media_image3.png
Greyscale
Examiner’s Response:
In response, the examiner maintains the rejection. The cited passages describe the alleged advance as providing more “informative” or “probative” explanation data for black-box time series models. Claim 1 recites only identifying important tokens (mental), three mathematical computations (distribution, quantiles, and the saliency value), and presenting the result on a generic GUI. It recites no step that applies the saliency information to improve the model, the computer, or any technology. This does not cause the claims to be significantly more than the abstract idea. Therefore, the rejection is maintained as shown below.
Applicant’s Argument:
The applicant argues (Page 11)
PNG
media_image4.png
423
1314
media_image4.png
Greyscale
Examiner’s Response:
In response, the examiner maintains the rejection. The applicant alleges that by “providing an explainability technique for computer implemented time series forecasting and anomaly detection models”, this qualifies as significantly more and not well-understood, routine and/or do not constitute a conventional activity. However, the recitation tying the claim to a “time series forecasting or anomaly detection model” is a field-of-use/technological-environment limitation that is not significantly more (MPEP 2106.05(h). Additionally, Presenting the computed result via a generic GUI is well-understood, routine, and conventional (MPEP 2106.05(d) and the applicant’s own specification supplies the Berkheimer support by describing only generic hardware (paragraphs [0054]–[0062]) and conventional presentation (paragraphs [0030]–[0033]).
Response to Applicant's Arguments Regarding the 103 Rejection
The Applicant has amended independent claims 1, 10, and 15 and presents arguments traversing the § 103 rejections. The Examiner has fully considered the Applicant's arguments but finds them unpersuasive for the reasons set forth below.
Applicant’s Argument:
The Applicant argues (page 12):
"Nayebi, Takagi, and Venkat, when taken alone or in combination, fail to disclose or suggest all aspects recited in the subject claims. For example, amended independent claim 1 recites, in part, 'dividing an aggregated importance of features of the frequency distribution information by a width of the quantile.' Nayebi, Takagi, and Venkat, when taken alone or in combination, fail to disclose or suggest at least such features related to normalizing the importance feature aggregation to the quantile width."
Examiner's Response:
It appears the Applicant is arguing that none of the cited references teach the specific mathematical operation of dividing aggregated importance by a width of the quantile. This argument is not persuasive. Nayebi explicitly teaches this operation in Equation 5 (Section 3.3, page 8): φ(i,t) = φ(ωk_i) / |ωk_i|. The numerator φ(ωk_i) is the aggregated importance (Shapley value) of the features within time window ωk_i -- this is the "aggregated importance of features of the frequency distribution information" because it represents the total importance contribution of the features falling within a particular interval of the distribution. The denominator |ωk_i| is the size (the width) of the time window. This equation explicitly performs the operation of dividing aggregated importance by the width of the grouping interval. When combined with Takagi's teaching of using quantiles as a distribution type (Para 0041), the grouping intervals become quantiles, and dividing by window size becomes dividing by quantile width. The mathematical operation claimed, dividing an aggregate by the extent of its grouping interval, is directly disclosed by Nayebi Equation 5 and merely applied to quantile-structured intervals via Takagi.
Applicant’s Argument:
The Applicant argues (page 12):
"The Office Action cites Nayebi in alleged support of most aspects of independent claim 1, including 'calculating spatial saliency information based on the frequency distribution information and quantile information.' See, Office Action, page 20. Nayebi, as cited, discloses calculating Shapley values for specific time windows, and time window variations, including fixed-length windows, sliding window, and variable-length windows. See, Nayebi, sections 3.3.1, 3.3.2, and 3.3.3. Nayebi discloses calculating Shapley values for the time windows, but Nayebi fails to disclose or suggest at least 'dividing an aggregated importance of features of the frequency distribution information by a width of the quantile,' as recited in amended independent claim 1."
Examiner's Response:
It appears the Applicant is arguing that while Nayebi calculates Shapley values for time windows, Nayebi does not perform the specific operation of dividing aggregated importance by quantile width.
This argument is not persuasive. The Applicant acknowledges that Nayebi calculates Shapley values for time windows but contends that this does not amount to dividing aggregated importance by a width of the quantile. However, Nayebi's Equation 4 (Section 3.3, page 7) computes the Shapley value for an entire time window, this IS the aggregated importance of the features within that window, because the equation treats the window as a single feature and computes its collective contribution to the model output. Nayebi's Equation 5 (Section 3.3, page 8) then divides this aggregated window importance by the window's cardinality |ωk_i|, which represents the width/size of the window. The Examiner notes that the claim recites "a width of the quantile" without specifying a particular unit of measurement. Under the broadest reasonable interpretation, the cardinality of a grouping interval (number of elements it spans) is a measure of its width. Moreover, Takagi teaches using quantiles as a distribution type (Para 0041), and when this is applied to Nayebi's framework, the windows become quantile-based groupings, and the existing division by window size (Equation 5) directly yields division by quantile width.
Applicant’s Argument:
The Applicant argues (pages 12-13):
"Rather, Nayebi only contemplates calculating Shapley values using WindowSHAP, but Nayebi fails to contemplate spatial saliency explanation at all. Indeed, the Shapley values in Nayebi are not spatial at all, e.g., the Shapley values are not based on spatial grouping information. Nayebi also fails to contemplate normalizing the values for the time window width."
Examiner's Response:
It appears the Applicant is raising three arguments,
Nayebi does not teach spatial saliency explanation
Nayebi's Shapley values are not spatial because they are not based on spatial grouping information
Nayebi does not normalize values for the time window width.
Regarding spatial saliency explanation: This argument is not persuasive. Nayebi's entire WindowSHAP framework is directed to calculating saliency (importance) information for spatial groupings (time windows) along the temporal axis of time series data. Section 3.3 (page 7) states: "WindowSHAP is designed on the idea of constructing windows from either nearby or non-adjacent temporal steps. In this method, we compute Shapley values for each individual time window rather than for all possible combinations of variable-time points." Computing importance values for spatial groupings (windows) of a time series is spatial saliency explanation -- it explains which spatial regions of the time series contribute most to the model output. The fact that Nayebi does not use the specific phrase "spatial saliency" does not mean the concept is absent. Under the broadest reasonable interpretation, a Shapley value computed for a time window IS a spatial saliency value -- it quantifies the saliency (importance) of a spatial (temporal) region.
Regarding spatial grouping: This argument is not persuasive. Nayebi's time windows ARE spatial groupings. Each window ωk_i groups contiguous time points into a spatial region along the time axis. Dynamic WindowSHAP (Section 3.3.3, page 10) creates variable-length windows based on the density of importance, splitting high-importance regions into smaller intervals. These variable-length windows, derived from importance distribution analysis, are spatial groupings functionally analogous to quantiles. When combined with Takagi's explicit teaching of quantiles as a distribution type (Para 0041), the spatial groupings are formally structured as quantiles.
Regarding normalization by window width: This argument is incorrect with respect to Nayebi's disclosure. Nayebi Equation 5 (Section 3.3, page 8) explicitly divides the aggregated importance of a time window by its size: φ(i,t) = φ(ωk_i) / |ωk_i|. The denominator |ωk_i| is the width/size of the time window. This IS normalization by window width. The Examiner notes that the Applicant's statement that "Nayebi also fails to contemplate normalizing the values for the time window width" is contradicted by Equation 5, which performs precisely this normalization.
Applicant’s Argument:
The Applicant argues (page 13):
"Takagi and Venkat fail to cure the deficiencies of Nayebi with respect to at least such aspects of amended independent claim 1, at least because these references are also completely silent regarding the aspects."
Examiner's Response:
This argument is not persuasive because, as explained above, Nayebi itself teaches dividing aggregated importance by the width of a grouping interval (Equation 5). Therefore, Takagi and Venkat are not relied upon to "cure" this particular aspect. Takagi is relied upon to teach: (1) displaying importance information via a graphical user interface (Para 0024), and (2) using quantiles as a distribution type for importance analysis (Para 0041). When Takagi's quantile-based distribution analysis is combined with Nayebi's window-based framework, Nayebi's existing Equation 5 operation (dividing aggregated importance by window size) yields the claimed "dividing an aggregated importance of features of the frequency distribution information by a width of the quantile" because the windows, structured as quantiles per Takagi, have their size/width defined by the quantile boundaries.
Regarding Independent Claims 10 and 15
Independent claims 10 and 15 are amended with substantially the same limitations as amended independent claim 1. Therefore, claims 10 and 15 are rejected for the same reasons above with respect to claim 1.
Regarding Dependent Claims
Dependent claims 2-9, 11-14, and 16-20 depend from independent claims 1, 10, and 15 respectively and do not recite additional limitations that overcome the deficiencies identified above. These claims are rejected as set forth in the prior Office Action, which is incorporated herein, with the modifications to the independent claim rejections as set forth below.
Claim Objections
In response to the amendments made regarding the claim objections presented in the previous office action, the examiner withdraws the claim objections.
Specification and Drawings Objections
In response to the amendments made regarding the specification and drawings objections presented in the previous office action, the examiner withdraws these objections.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The analysis of the claims will follow the 2019 Revised Patent Subject Matter Eligibility Guidance, 84 Fed. Reg. 50 (“2019 PEG”).
Claim 1
Step 1: The claim recites a method; therefore, it is directed to the statutory category of processes.
Step 2A prong 1: The claim recites the following abstract ideas:
identifying, based on a token-based importance method, a plurality of tokens each having at least a predefined importance to a machine learning (ML) inference, (…) This limitation falls within the metal process grouping because it describes an act of evaluation or judgement. A person can mentally judge and identify items based on importance.
generating frequency distribution information by binning the plurality of tokens; (This limitation falls with the mathematical concepts grouping because it involves statistical math operations of organizing data to determine a distribution.)
generating, based on the frequency distribution information, quantile information based on the plurality; (This limitation falls within the mathematical concepts grouping because it involves finding quantile information which involve a mathematical calculation to divide a probability distribution into intervals of equal probability.)
calculating based on the frequency distribution information and the quantile information, the spatial saliency information including a spatial saliency value for a quantile information, wherein the calculating includes dividing an aggregated importance of features of the frequency distribution information by a width of the quantile; (This limitation falls within the mathematical concepts grouping because it recites the act of calculating a numerical value using mathematical methods. The newly added “wherein ... dividing an aggregated importance ... by a width of the quantile” recites a mathematical formula (Specification, paragraphs [0028] – [0029]) and therefore reinforces, rather than removes, the mathematical-concept characterization.)
Step 2A prong 2: This judicial exception is not integrated into a practical application. The claim further recites:
wherein the ML inference is applied to time series data by a time series forecasting or anomaly detection model; (This limitation merely confines the abstract idea to a particular technological environment / field of use. The claim recites nothing about how the model operates and effects no change to the model; the model is named only as the source and context of the importance values being evaluated and computed upon. Generally linking an abstract idea to a particular technological environment does not integrate it into a practical application (MPEP 2106.05(h))
and presenting the spatial saliency information for the time series forecasting or anomaly detection model via a graphical user interface. (This limitation is merely a post-solution step of presenting the data that does not meaningfully limit the claim. Presenting is recited at a high level of generality, and the added phrase “for the time series forecasting or anomaly detection model” merely identifies what the displayed data relates to. Simply implementing the abstract idea in a generic method is not a practical application of the abstract idea. Therefore, this step is an insignificant extra-solution activity (MPEP 2106.05(g)).)
Step 2B:
wherein the ML inference is applied to time series data by a time series forecasting or anomaly detection model; (This limitation merely confines the abstract idea to a particular technological environment / field of use. The claim recites nothing about how the model operates and effects no change to the model; the model is named only as the source and context of the importance values being evaluated and computed upon. Generally linking an abstract idea to a particular technological environment does not integrate it into a practical application (MPEP 2106.05(h))
and presenting the spatial saliency information for the time series forecasting or anomaly detection model via a graphical user interface. (MPEP 2106.05(d)(II) indicates that merely presenting offers and gathering statistics is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer.)
The additional elements considered individually or in combination do not amount to significantly more than the judicial exception. Therefore, the claim is not patent eligible.
Claim 2
Step 1: A process, as above.
Step 2A prong 1: See the rejection of Claim 1 above, which claim 2 depends on. Claim 2 further recites:
further comprising generating the ML inference based on time series data, wherein the ML inference includes at a predicted value of a time stamp. (This limitation falls within the metal process grouping because it describes an act of evaluation or judgement. A person can mentally make an inference that’s based on time series data at a particular time stamp.
Step 2A prong 2: The claim does not recite additional elements therefore the judicial exception is not
integrated into a practical application.
Step 2B: The claim does not recite additional elements that amount to significantly more than the
judicial exception. Therefore, the claim is not patent eligible.
Claim 3
Step 1: A process, as above.
Step 2A prong 1: See the rejection of Claim 1 above, which claim 3 depends on. Claim 3 further recites:
wherein identifying the plurality of tokens comprises: identifying the plurality of the tokens as a predefined number of tokens having the highest importance values according to the token-based importance method. (This limitation falls within mental process grouping of abstract ideas. A person can perform this limitation in their mind or using pen and paper. The process amounts to simply observation, evaluation, and sorting, which are concepts performed in the human mind.)
Step 2A prong 2: The claim does not recite additional elements therefore the judicial exception is not
integrated into a practical application.
Step 2B: The claim does not recite additional elements that amount to significantly more than the
judicial exception. Therefore, the claim is not patent eligible.
Claim 4
Step 1: A process, as above.Step 2A prong 1: See the rejection of Claim 1 above, which claim 4 depends on. Claim 4 further recites:
wherein token-based importance method includes a local interpretable model-agnostic explanations (LIME) method or a Shapley additive explanations (SHAP) method. (This limitation falls within the mathematical concepts grouping because it recites mathematical algorithms (LIME and SHAP) which are used to calculate feature contribution values and importance scores.
Step 2A prong 2: The claim does not recite additional elements therefore the judicial exception is not
integrated into a practical application.
Step 2B: The claim does not recite additional elements that amount to significantly more than the
judicial exception. Therefore, the claim is not patent eligible.
Claim 5
Step 1: A process, as above.Step 2A prong 1: See the rejection of Claim 1 above, which claim 5 depends on. Claim 5 further recites:
wherein generating frequency distribution information comprises generating a frequency distribution histogram based on the plurality of tokens. (This limitation falls within the mathematical concepts grouping because creating a histogram relies on the mathematical operations of defining numerical ranges and calculating how data falls within those ranges. As stated in paragraph [0028] – “the SSE module 116 generates a frequency distribution histogram HistLKT,Nb by binning LKT into a predefined number of bins NB, and determines the quantiles of the HistLKT,Nb where NQ is the number of quantiles.”
Step 2A prong 2: The claim does not recite additional elements therefore the judicial exception is not
integrated into a practical application.
Step 2B: The claim does not recite additional elements that amount to significantly more than the
judicial exception. Therefore, the claim is not patent eligible.
Claim 6
Step 1: A process, as above.Step 2A prong 1: See the rejection of Claim 1 above, which claim 6 depends on. Claim 6 further recites:
wherein calculating the spatial saliency information comprises: determining an aggregated importance of a timestamp range of the quantile of the quantile information; and determining the spatial saliency value based on the aggregated importance and a size of the quantile. (This limitation falls within the mathematical concepts grouping because it describes a mathematical formula for density. The specification explicitly defines this calculation as a mathematical equation (see paragraph [0028]).
Step 2A prong 2: The claim does not recite additional elements therefore the judicial exception is not
integrated into a practical application.
Step 2B: The claim does not recite additional elements that amount to significantly more than the
judicial exception. Therefore, the claim is not patent eligible.
Claim 7
Step 1: A process, as above.Step 2A prong 1: See the rejection of Claim 1 above, which claim 7 depends on.
Step 2A prong 2: This judicial exception is not integrated into a practical application. The claim further recites:
wherein presenting the spatial saliency information via the graphical user interface comprises generating the graphical user interface to include a table presenting the spatial saliency information; and applying, based on the spatial saliency value, within the graphical user interface, a graphical effect to table information associated with the quantile of the quantile information. (The additional elements merely present the result of the abstract idea (the calculated saliency information) in a table format with visual effects. This is a post-solution step of displaying data, which is an insignificant addition to the claim that does not meaningfully limit the claim. Simply implementing the abstract idea and displaying the result in a generic table is not practical application of the abstract idea. Therefore, this is insignificant extra-solution step (MPEP 2106.05(g)).)
Step 2B: The claim does not recite additional elements that amount to significantly more than the
judicial exception. Therefore, the claim is not patent eligible.
wherein presenting the spatial saliency information via the graphical user interface comprises generating the graphical user interface to include a table presenting the spatial saliency information; and applying, based on the spatial saliency value, within the graphical user interface, a graphical effect to table information associated with the quantile of the quantile information. (MPEP 2106.05(d)(II) indicates that merely presenting offers and gathering statistics is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer)
The additional elements considered individually or in combination do not amount to significantly more than the judicial exception. Therefore, the claim is not patent eligible.
Claim 8
Step 1: A process, as above.Step 2A prong 1: See the rejection of Claim 1 above, which claim 8 depends on.
Step 2A prong 2: This judicial exception is not integrated into a practical application. The claim further recites:
wherein presenting the spatial saliency information comprises: generating the graphical user interface to include a graph representation of time sample information used to generate the ML inference, wherein the graph representation identifies the quantile of the quantile information; and applying, based on the spatial saliency value, within the graphical user interface, a graphical effect to graph information associated with the quantile of the quantile information. (The additional elements merely present the result of the abstract idea (the calculated saliency information) in a graph format with visual effects. This is a post-solution step of displaying data, which is an insignificant addition to the claim that does not meaningfully limit the claim. Simply implementing the abstract idea and displaying the result in a generic table is not practical application of the abstract idea. Therefore, this is insignificant extra-solution step (MPEP 2106.05(g)).)
Step 2B:
wherein presenting the spatial saliency information via the graphical user interface comprises: generating the graphical user interface to include a graph representation of time sample information used to generate the ML inference, wherein the graph representation identifies the quantile of the quantile information; and applying, based on the spatial saliency value, within the graphical user interface, a graphical effect to graph information associated with the quantile of the quantile information. (MPEP 2106.05(d)(II) indicates that merely presenting offers and gathering statistics is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer)
The additional elements considered individually or in combination do not amount to significantly more than the judicial exception. Therefore, the claim is not patent eligible.
Claim 9
Step 1: A process, as above.Step 2A prong 1: See the rejection of Claim 1 above, which claim 9 depends on.
Step 2A prong 2: This judicial exception is not integrated into a practical application. The claim further recites:
wherein presenting the spatial saliency information comprises transmitting, to a client device in response to a client request, the spatial saliency information for display via the graphical user interface. (This additional element recites a mere instruction to apply an exception with a recitation of the words "apply it" (or an equivalent) as identified in MPEP 2106.05(f), and does not provide integration into a practical application. This limitation amounts to transmitting data over a network, which is merely an instruction to apply the abstract idea using a generic computer component.
Step 2B:
wherein presenting the spatial saliency information comprises transmitting, to a client device in response to a client request, the spatial saliency information for display via the graphical user interface. (MPEP 2106.05(d)(II) indicates that merely “Receiving or transmitting data over a network” is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer)
The additional elements considered individually or in combination do not amount to significantly more than the judicial exception. Therefore, the claim is not patent eligible.
Claim 10Step 1: The claim recites a system; therefore, it is directed to the statutory category of machine.Step 2A prong 1: The claim recites the following abstract ideas:
identify, based on a token-based importance method, a plurality of tokens each having at least a predefined importance to a machine learning (ML) inference, (…) This limitation falls within the metal process grouping because it describes an act of evaluation or judgement. A person can mentally judge and identify items based on importance.
generate frequency distribution information by binning the plurality of tokens; (This limitation falls with the mathematical concepts grouping because it involves statistical math operations of organizing data to determine a distribution.)
generate, based on the frequency distribution information, quantile information based on the plurality; (This limitation falls within the mathematical concepts grouping because it involves finding quantile information which involve a mathematical calculation to divide a probability distribution into intervals of equal probability.)
determine, based on the frequency distribution information and the quantile information, the spatial saliency information including a spatial saliency value for a quantile information, wherein the calculating includes dividing an aggregated importance of features of the frequency distribution information by a width of the quantile; (This limitation falls within the mathematical concepts grouping because it recites the act of determining a numerical value using mathematical methods.)
Step 2A prong 2: This judicial exception is not integrated into a practical application. The claim further recites:
A system comprising: a memory storing instructions thereon; and at least one processor coupled with the memory and configured by the instructions to: (Adding the words “apply it” (or an equivalent) with the judicial exception, 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)).)
wherein the ML inference is applied to time series data by a time series forecasting or anomaly detection model; (This limitation merely confines the abstract idea to a particular technological environment / field of use. The claim recites nothing about how the model operates and effects no change to the model; the model is named only as the source and context of the importance values being evaluated and computed upon. Generally linking an abstract idea to a particular technological environment does not integrate it into a practical application (MPEP 2106.05(h))
and presenting the spatial saliency information for the time series forecasting or anomaly detection model via a graphical user interface. (This limitation is merely a post-solution step of presenting the data that does not meaningfully limit the claim. Presenting is recited at a high level of generality, and the added phrase “for the time series forecasting or anomaly detection model” merely identifies what the displayed data relates to. Simply implementing the abstract idea in a generic method is not a practical application of the abstract idea. Therefore, this step is an insignificant extra-solution activity (MPEP 2106.05(g)).)
Step 2B:
A system comprising: a memory storing instructions thereon; and at least one processor coupled with the memory and configured by the instructions to: (Adding the words “apply it” (or an equivalent) with the judicial exception, 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)).)
wherein the ML inference is applied to time series data by a time series forecasting or anomaly detection model; (This limitation merely confines the abstract idea to a particular technological environment / field of use. The claim recites nothing about how the model operates and effects no change to the model; the model is named only as the source and context of the importance values being evaluated and computed upon. Generally linking an abstract idea to a particular technological environment does not integrate it into a practical application (MPEP 2106.05(h))
and presenting the spatial saliency information for the time series forecasting or anomaly detection model via a graphical user interface. (MPEP 2106.05(d)(II) indicates that merely presenting offers and gathering statistics is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer.)
The additional elements considered individually or in combination do not amount to significantly more than the judicial exception. Therefore, the claim is not patent eligible.
Claim 11
Claim 11 is a system (machine) claim that recites identical limitations to method claim 2. Therefore, claim 11 is rejected using the same rationale as claim 2.
Claim 12
Claim 12 is a system (machine) claim that recites identical limitations to method claim 3. Therefore, claim 12 is rejected using the same rationale as claim 3.
Claim 13
Claim 13 is a system (machine) claim that recites identical limitations to method claim 4. Therefore, claim 13 is rejected using the same rationale as claim 4.
Claim 14
Claim 14 is a system (machine) claim that recites identical limitations to method claim 6. Therefore, claim 14 is rejected using the same rationale as claim 6.
Claim 15
Step 1: The claim recites a non-transitory computer-readable device; therefore, it is directed to the statutory category of machine.
Step 2A prong 1: The claim recites the following abstract ideas:
identifying, based on a token-based importance method, a plurality of tokens each having at least a predefined importance to a machine learning (ML) inference, (…) This limitation falls within the metal process grouping because it describes an act of evaluation or judgement. A person can mentally judge and identify items based on importance.
generating frequency distribution information by binning the plurality of tokens; (This limitation falls with the mathematical concepts grouping because it involves statistical math operations of organizing data to determine a distribution.)
generating, based on the frequency distribution information, quantile information based on the plurality; (This limitation falls within the mathematical concepts grouping because it involves finding quantile information which involve a mathematical calculation to divide a probability distribution into intervals of equal probability.)
calculating based on the frequency distribution information and the quantile information, the spatial saliency information including a spatial saliency value for a quantile information, wherein the calculating includes dividing an aggregated importance of features of the frequency distribution information by a width of the quantile; (This limitation falls within the mathematical concepts grouping because it recites the act of calculating a numerical value using mathematical methods.)
Step 2A prong 2: This judicial exception is not integrated into a practical application. The claim further recites:
A non-transitory computer-readable device having instructions thereon that, when executed by at least one computing device, cause the at least one computing device to perform operations comprising: (Adding the words “apply it” (or an equivalent) with the judicial exception, 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)).)
wherein the ML inference is applied to time series data by a time series forecasting or anomaly detection model; (This limitation merely confines the abstract idea to a particular technological environment / field of use. The claim recites nothing about how the model operates and effects no change to the model; the model is named only as the source and context of the importance values being evaluated and computed upon. Generally linking an abstract idea to a particular technological environment does not integrate it into a practical application (MPEP 2106.05(h))
and presenting the spatial saliency information for the time series forecasting or anomaly detection model via a graphical user interface. (This limitation is merely a post-solution step of presenting the data that does not meaningfully limit the claim. Presenting is recited at a high level of generality, and the added phrase “for the time series forecasting or anomaly detection model” merely identifies what the displayed data relates to. Simply implementing the abstract idea in a generic method is not a practical application of the abstract idea. Therefore, this step is an insignificant extra-solution activity (MPEP 2106.05(g)).)
Step 2B:
A non-transitory computer-readable device having instructions thereon that, when executed by at least one computing device, cause the at least one computing device to perform operations comprising: (Adding the words “apply it” (or an equivalent) with the judicial exception, 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)).)
wherein the ML inference is applied to time series data by a time series forecasting or anomaly detection model; (This limitation merely confines the abstract idea to a particular technological environment / field of use. The claim recites nothing about how the model operates and effects no change to the model; the model is named only as the source and context of the importance values being evaluated and computed upon. Generally linking an abstract idea to a particular technological environment does not integrate it into a practical application (MPEP 2106.05(h))
and presenting the spatial saliency information for the time series forecasting or anomaly detection model via a graphical user interface. (MPEP 2106.05(d)(II) indicates that merely presenting offers and gathering statistics is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer.)
The additional elements considered individually or in combination do not amount to significantly more than the judicial exception. Therefore, the claim is not patent eligible.
Claim 16
Claim 16 is a non-transitory computer-readable device type claim that recites identical limitations to method claim 2. Therefore, claim 16 is rejected using the same rationale as claim 2.
Claim 17
Claim 17 is a non-transitory computer-readable device type claim that recites identical limitations to method claim 3. Therefore, claim 17 is rejected using the same rationale as claim 3.
Claim 18
Claim 18 is a non-transitory computer-readable device type claim that recites identical limitations to method claim 4. Therefore, claim 18 is rejected using the same rationale as claim 4.
Claim 19
Claim 19 is a non-transitory computer-readable device type claim that recites identical limitations to method claim 5. Therefore, claim 19 is rejected using the same rationale as claim 5.
Claim 20
Claim 20 is a non-transitory computer-readable device type claim that recites identical limitations to method claim 6. Therefore, claim 20 is rejected using the same rationale as claim 6.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1-4, 6-18 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over non-patent literature Nayebi et al. ("WindowSHAP: An Efficient Framework for Explaining Timeseries Classifiers based on Shapley Values", hereinafter "Nayebi") in view of Takagi et al. (US-20220076049-A1), hereinafter "Takagi".
Claim 1
Nayebi teaches:
A method comprising: (Section 3.3, page 7, "We introduce our efficient framework called WindowSHAP... In this method, we compute Shapley values for each individual time window...")
identifying, based on a token-based importance method, a plurality of tokens each having at least a predefined importance to a machine learning (ML) inference, (Section 3.1, page 6, "Shapley values assign an importance (contribution) score φi to the ith feature, indicating how much the model output for a single instance is influenced by its ith feature." – EN: Shapley values are a token-based importance method. Each feature (token) is assigned an importance score, and selecting the features with at least a predefined importance value yields the claimed "plurality of tokens each having at least a predefined importance." See also Figure 7, page 18, "The top 15 variables depicted on the y axis are ranked according to their importance." – EN: This demonstrates selecting a plurality of tokens that each meet a predefined importance threshold.)
wherein the ML inference is applied to time series data by a time series forecasting or anomaly detection model; (Figure 1 caption, page 4, "Conceptual Demonstration of KernelSHAP vs WindowSHAP for a classification model for an individual instance, predicting whether there is an anomaly in a synthetically generated sequence." – EN: This denotes applying the ML inference to time series data by an anomaly detection model. See also Section 3.5, page 13, "A prediction model was developed and trained to predict an adverse event. Here, an adverse event is defined when intracranial pressure (ICP) is larger than 22 mmHg for at least 15 minutes." – EN: This is an anomaly detection model applied to time series data. See also Section 3.5, page 13, "The third clinical prediction model is based on the MIMIC data set and uses the initial 48 hours of clinical data to predict patient mortality in the subsequent 48 hours." – EN: This is a time series forecasting model that uses time series data to predict a future outcome.)
generating frequency distribution information by binning the plurality of tokens; (See Figure 1, page 4, and Figure 8, page 19 – EN: These depict bar charts showing the distribution of importance (Shapley values) across time bins, which is a visual representation of frequency distribution information generated by binning. Also see Section 3.3.1, page 8, "In this approach, the time-axis is segmented into fixed-length windows." See also Algorithm 1, page 9, which constructs time windows as ωk_i = {(i,t) | (k-1)·l < t ≤ min{L, k·l}} – EN: This is the mathematical definition of binning: partitioning the time axis into discrete intervals (bins) of length l. Each bin captures the features (tokens) that fall within its temporal range. The resulting set of bins with their associated importance values constitutes frequency distribution information. For example, information about how the important tokens are distributed across the binned time intervals.)
generating, based on the frequency distribution information, quantile information based on the plurality of tokens; (Section 3.3.3 Dynamic WindowSHAP, page 10, "In this approach, we divide the entire series into variable-length time windows. To accomplish this, we first define what the optimal split is using the following two objectives: 1. Keeping the number of time windows as few as possible... 2. Avoiding lengthy windows with large contribution scores to minimize information loss" – EN: Dynamic WindowSHAP generates variable-length windows (ranges) based on the importance density of the data. It splits regions with high importance into smaller windows, making spatial groupings (analogous to quantiles). For example, ranges that partition the distribution based on the concentration of importance. This is done based on the frequency distribution information because the splitting decision depends on the Shapley values (importance) calculated in prior iterations.)
calculating, based on the frequency distribution information and the quantile information, spatial saliency information including a spatial saliency value for a quantile of the quantile information, (Section 3.3, page 7, Equation 4 – EN: This equation calculates the Shapley value for the kth time window of variable i, which constitutes a saliency value for a particular spatial grouping (window/quantile). The calculation is based on both the distribution of importance across the data (frequency distribution information) and the window/quantile structure (quantile information). See also Section 1, page 3, "Instead of calculating Shapley values for every possible time step and variable combinations, we simply calculate Shapley values for each time window (see Figure 1 for conceptual demonstration)." – EN: This describes calculating importance values for spatial groupings rather than individual points, which is the essence of spatial saliency information.)
wherein the calculating includes dividing an aggregated importance of features of the frequency distribution information by a width of the quantile; (Section 3.3, page 8, Equation 5, "The Shapley value of any variable-time point combination can be estimated by distributing the importance of a time window equally among its time points, i.e., φ(i,t) = φ(ωk_i) / |ωk_i|, ∀(i,t) ∈ ωk_i" -- EN: The numerator φ(ωk_i) is the aggregated importance of the time window, the Shapley value assigned to the entire window, which represents the collective importance of all features within that temporal range of the frequency distribution. The denominator |ωk_i| is the size (width) of the time window, measured as the number of time points it spans. This equation performs the operation of dividing aggregated importance by the width of the grouping interval. In a time series with uniformly spaced time steps, the size |ωk_i| is directly proportional to the temporal width of the window. See also Section 3.3, page 7, Equation 4 -EN: Equation 4 computes the aggregated importance (Shapley value) for the window as a whole, treating each window as a single feature. Equation 5 then divides this aggregated importance by the window's size/width. Together, Equations 4 and 5 disclose the claimed "dividing an aggregated importance of features of the frequency distribution information by a width of the quantile." See also Section 5, page 19, "First, by aggregating nearby time steps as a time window, WindowSHAP lowers the dependence of the elements..." — EN: This confirms that the framework aggregates importance at the window level and then distributes (divides) it based on window size.)
presenting the spatial saliency information (Figure 8, page 19 -- EN: This shows bar charts presenting Shapley values for specific time windows. Figure 7, page 18 -- EN: This shows heatmaps depicting the importance of all time steps for important features, which constitutes a presentation of spatial saliency information.)
for the time series forecasting or anomaly detection model (…) (Figure 1 caption, page 4, "Conceptual Demonstration of KernelSHAP vs WindowSHAP for a classification model for an individual instance, predicting whether there is an anomaly in a synthetically generated sequence." – EN: This denotes applying the ML inference to time series data by an anomaly detection model. See also Section 3.5, page 13, "A prediction model was developed and trained to predict an adverse event. Here, an adverse event is defined when intracranial pressure (ICP) is larger than 22 mmHg for at least 15 minutes." – EN: This is an anomaly detection model applied to time series data. See also Section 3.5, page 13, "The third clinical prediction model is based on the MIMIC data set and uses the initial 48 hours of clinical data to predict patient mortality in the subsequent 48 hours." – EN: This is a time series forecasting model that uses time series data to predict a future outcome.)
Nayebi does not explicitly disclose:
"… via a graphical user interface."
However, Takagi teaches:
"via a graphical user interface" (Para 0024, "The display controller 14 causes the display 5 to display various types of information. For example, the display controller 14 visualizes the distribution. Specifically, the display controller 14 generates a graph showing a distribution for each of a plurality of feature amounts, and causes the display 5 to display the graph.")
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to combine the presentation of spatial saliency information of Nayebi, including the importance information for time series forecasting and anomaly detection models, with the display controller and graphical user interface of Takagi. This combination would allow users to observe and analyze behaviors of importances corresponding to features through a GUI. (Takagi Para 0046, "Since importances are visualized for each of the groups formed according to estimated output values, it is possible to observe and analyze behaviors of importances corresponding to estimated output values. Since importances of each feature amount are visualized for each group, it is possible to observe and analyze behaviors of importances corresponding to feature amounts, differences in behavior corresponding to estimated output values, and the like.")
Furthermore, it would have been obvious to apply the quantile-based distribution grouping of Takagi (Para 0041, "The type of distribution can be discretionarily selected from among a probability density function, an average, a standard deviation, a quantile, and the like.") to the time-window-based importance analysis of Nayebi, because Takagi teaches that such statistical distribution methods, including quantiles, are standard approaches for analyzing and visualizing feature importances in machine learning, and because applying quantile-based grouping to Nayebi's window-based framework would improve the analysis by providing statistically principled spatial groupings. When Nayebi's windows are structured as quantiles per Takagi's teaching, Nayebi's existing operation of dividing aggregated window importance by window size (Equation 5) directly yields the claimed operation of dividing aggregated importance by quantile width.
Claim 2
Nayebi further teaches:
generating the ML inference based on time series data, wherein the ML inference includes at a
predicted value of a time stamp. (Section 3.5 “We developed a prediction model to predict the long-
term functional outcome of patients”. Section 3.5 "The third clinical prediction model is based on the
MIMIC data set and uses the initial 48 hours of clinical data to predict patient mortality in the
subsequent 48 hours.” EN: This denotes the paper’s application to time-series predictive models used
to forecast patient outcomes at future time points.)
Claim 3
Nayebi further teaches:
The method of claim 1, wherein identifying the plurality of tokens comprises: identifying the plurality of tokens as a predefined number of tokens having the highest importance values according to the token-based importance method. (Figure 7 – “The top 15 variables depicted on the y axis are ranked according to their importance.” -- EN: This depicts a visualization where it identifies and presents exactly the top 15 variables based on their importance ranking. This figure is based on the Dynamic WindowSHAP algorithm which uses Shapley values, therefore it is based on SHAP (a token-based importance method.)
Claim 4
Nayebi further teaches:
The method of claim 1, wherein token-based importance method includes a local interpretable model-agnostic explanations (LIME) method or a Shapley additive explanations (SHAP) method. (Section 1 Introduction - “In summary, the main contributions of this study are as follows: Developing the WindowSHAP framework, a variation of Shapley additive explanations for time-series data…”)
Claim 6
Nayebi further teaches:
The method of claim 1, wherein calculating the spatial saliency information comprises: determining an aggregated importance of a timestamp range of the quantile of the quantile information; (Section 3.3 “Considering each window for each variable as a feature, the Shapley value for the 𝑘th time window of variable 𝑖 is calculated as see equation 4” and Section 5 “First, by aggregating nearby time steps as a time window, WindowSHAP lowers the dependence of the elements…”
PNG
media_image5.png
215
1448
media_image5.png
Greyscale
EN: this denotes determining the importance (shapley value) of a timestamp range (window). Equation 4 treats each window as a single feature, therefore equation 4 determines the collective (aggregated) importance of all the time steps within that time range (window) ). and determining the spatial saliency value based on the aggregated importance and a size of the quantile (Section 3.3
“The Shapley value of any variable-time point combination can be estimated by distributing the importance of a time window equally among its time points, i.e., see equation 5”)
PNG
media_image6.png
254
1125
media_image6.png
Greyscale
Claim 7
Nayebi further teaches:
The method of claim 1, wherein presenting the spatial saliency information via the graphical user interface (…) to include a table presenting the spatial saliency information; and applying, based on the spatial saliency value, within the graphical user interface, a graphical effect to table information associated with (…) (Section 4 “Figure 7. Heatmaps depicting the importance of all time steps for the important features for a certain patient record” EN: This denotes a Heatmap which includes a matrix of data organized in rows and columns and the graphical effect is the color intensity of the cell.)
Nayebi does not distinctly disclose:
“generating the graphical user interface…” and “…the quantile of the quantile information.“
However, Takagi teaches:
“generating the graphical user interface…” (Para 0024, “Specifically, the display controller 14 generates a graph showing a distribution for each of a plurality of feature amounts, and causes the display 5 to display the graph.”) “…the quantile of the quantile information.“ (Para 0041, The type of distribution can be discretionarily selected from among… a quantile…)
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to combine the heatmap (table information) and graphical effect of Nayebi with the statistical distribution analysis which includes quantiles and the display controller of Takagi. Combining the teaching of Nayebi and Takagi allows the heatmap (table information) to be structured in quantiles and be displayed via a graphical user interface. Thereby, allowing users observe and analyze behaviors of importances corresponding to features. (Takagi Para 0062, “The configuration also reduces the possibility of overlooking the case where there are feature amounts of high importances only in some input data samples out of all input data samples, the case where a distribution of importances is multimodal, or the like.” Takagi Para 0046, “Since importances are visualized for each of the groups formed according to estimated output values, it is possible to observe and analyze behaviors of importances corresponding to estimated output values. Since importances of each feature amount are visualized for each group, it is possible to observe and analyze behaviors of importances corresponding to feature amounts, differences in behavior corresponding to estimated output values, and the like.”)
Claim 8
Nayebi further teaches:
The method of claim 1, wherein presenting the spatial saliency information comprises: (…) to include a graph representation of time sample information used to generate the ML inference, (Figure 8 depicts a graph of time sample information used to generate the prediction (heart rate variable) wherein the graph representation identifies (…); (Figure 8 -- EN: this depicts (identifies) the two divided blocks, a blue block and red block) and applying, based on the spatial saliency value, within the graphical user interface, a graphical effect to graph information associated with (Figure 8, EN: this depicts the graphical effect (red and blue) and the height of the bars corresponds to the Shapley values (calculated in equation 5), this corresponds to “spatial saliency value” as stated in claim 6.)
Nayebi does not distinctly disclose:
“generating the graphical user interface” and “the quantile of the quantile information;”
However, Takagi teaches:
“generating the graphical user interface…” (Para 0024, “Specifically, the display controller 14 generates a graph showing a distribution for each of a plurality of feature amounts, and causes the display 5 to display the graph.”) “…the quantile of the quantile information.“ (Para 0041, The type of distribution can be discretionarily selected from among… a quantile…)
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to combine the graph representation of time samples and the graphical effect of Nayebi with the statistical distribution analysis which includes quantiles and the display controller of Takagi. Combining the teaching of Nayebi and Takagi allows the graph representation of time sample to be structured in quantiles and be displayed via a graphical user interface. Thereby, allowing users observe and analyze behaviors of importances corresponding to features. (Takagi Para 0062, “The configuration also reduces the possibility of overlooking the case where there are feature amounts of high importances only in some input data samples out of all input data samples, the case where a distribution of importances is multimodal, or the like.” Takagi Para 0046, “Since importances are visualized for each of the groups formed according to estimated output values, it is possible to observe and analyze behaviors of importances corresponding to estimated output values. Since importances of each feature amount are visualized for each group, it is possible to observe and analyze behaviors of importances corresponding to feature amounts, differences in behavior corresponding to estimated output values, and the like.”)
Claim 9
Nayebi further teaches:
the spatial saliency information (Section 3.3 “The Shapley value of any variable-time point combination can be estimated by distributing the importance of a time window equally among its time points, i.e., see equation 5” Examiner Note (EN): the numerator of the equation denotes the aggregated importance and the denominator is the size of the time window. The result 𝜙(i,t) is the importance density of the size of the window. Similar to Para 0028 of the instant application “determines the spatial saliency information 126 by calculating the density of feature contribution to a model prediction for each quantile” )
Nayebi does not disclose:
wherein presenting (…) comprises: transmitting, to a client device in response to a client request, (…) for display via the graphical user interface.
However, Takagi teaches:
wherein presenting (…) comprises: transmitting, to a client device in response to a client request, (…) for display via the graphical user interface. (Figure 1 depicts an input device 3, communication device 4, display 5, and display controller 14)
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to take the density information (spatial saliency information) of Nayebi and apply it to the configuration system of Takagi to present the information. Thereby, allowing users to observe and analyze behaviors of importances corresponding to features. (Takagi Para 0046, Since importances are visualized for each of the groups formed according to estimated output values, it is possible to observe and analyze behaviors of importances corresponding to estimated output values. Since importances of each feature amount are visualized for each group, it is possible to observe and analyze behaviors of importances corresponding to feature amounts, differences in behavior corresponding to estimated output values, and the like.)
Claim 10
Takagi teaches:
A system comprising: (Para 0019, “FIG. 1 shows a configuration example of an importance analysis apparatus 100 according to the present embodiment. The importance analysis apparatus 100 is a computer for analyzing an importance of a feature amount in machine learning.”) a memory storing instructions thereon; and at least one processor coupled with the memory and configured by the instructions to: (Para 0020, “The processing circuitry 1 includes a processor such as a central processing unit (CPU), and a memory such as a random access memory (RAM). The processing circuitry 1 executes importance visualization processing for calculating and visualizing an importance of a feature amount in machine learning.” Para 0025, “The storage device 2 stores results of various operations by the processing circuitry 1, various programs executed by the processing circuitry 1, and the like.”)
The rest of claim 10 recites substantially the same limitations to method claim 1. Therefore, claim 10 is rejected using the same rationale as claim 1.
Regarding claim 11,
Claim 11 is a system type claim that recite the same limitations as claim 2, Therefore, claim 11 is rejected using the same rationale as claim 2.
Regarding claim 12,
Claim 12 is a system type claim that recite the same limitations as claim 3, Therefore, claim 12 is rejected using the same rationale as claim 3.
Regarding claim 13,
Claim 13 is a system type claim that recite the same limitations as claim 4, Therefore, claim 13 is rejected using the same rationale as claim 4.
Regarding claim 14,
Claim 14 is a system type claim that recite the same limitations as claim 6, Therefore, claim 14 is rejected using the same rationale as claim 6.
Regarding claim 15,
Takagi teaches:
A non-transitory computer-readable device having instructions thereon that, (Para, 0002, “Embodiments described herein relate generally to an importance analysis apparatus, method, and non-transitory computer readable medium.” Para 0063, “The function of each unit according to the present embodiment, and the program for causing a computer to implement the function may be stored in a non-transitory computer readable medium.”) when executed by at least one computing device, (Para 0020, “The processing circuitry 1 executes a program stored in the storage device 2 to implement an estimation unit 11, an importance calculator 12, a distribution calculator 13, and a display controller 14.”) cause the at least one computing device to perform operations comprising: (Para 0020, “The processing circuitry 1 executes importance visualization processing for calculating and visualizing an importance of a feature amount in machine learning.”)
The rest of claim 15 recites identical limitations to method claim 1. Therefore, claim 15 is rejected using the same rationale as claim 1.
Regarding claim 16,
Claim 16 is a non-transitory computer-readable device type claim that recite the same limitation of claim 2. Therefore, claim 16 is rejected using the same rationale as claim 2.
Regarding claim 17,
Claim 17 is a non-transitory computer-readable device type claim that recite the same limitation of claim 3. Therefore, claim 17 is rejected using the same rationale as claim 3.
Regarding claim 18,
Claim 18 is a non-transitory computer-readable device type claim that recite the same limitation of claim 4. Therefore, claim 18 is rejected using the same rationale as claim 4.
Regarding claim 20,
Claim 20 is a non-transitory computer-readable device type claim that recite the same limitation of claim 6. Therefore, claim 20 is rejected using the same rationale as claim 6.
Claims 5 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over non-patent literature Nayebi et al. ("WindowSHAP: An Efficient Framework for Explaining Timeseries Classifiers based on Shapley Values", hereinafter "Nayebi") in view of Takagi et al. (US-20220076049-A1), hereinafter "Takagi" further in view of Venkataramani et al. (US-20170154181-A1), hereinafter "Venkataramani”.Regarding claim 5, as discussed above, Nayebi in view of Takagi teaches all of the limitations of claim 1
Nayebi further teaches:wherein generating frequency distribution information based on the plurality of tokens of the predefined importance comprises generating (…) based on the plurality of tokens of the predefined importance. (Figure 1 and Figure 8 – Examiner Note (EN): this depicts a visual distribution of importance (bar charts) across the time axis, which is similar to “frequency distribution” of importance. Creating a “frequency distribution” is simply a statistical method of visualizing where the “importance tokens” identified previously are located. Nayebi discloses determining the location and density of importance).Nayebi does not disclose:
“a frequency distribution histogram”
However,
Venkataramani teaches “a frequency distribution histogram” (figure 7 – EN: this depicts Histogram which includes frequency in the Y axis)
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to combine the importance density calculation of Nayebi with the statistical distribution analysis (frequency and quantiles) of Takagi to include a histogram as taught by Venkataramani in order to visualize and identify clusters of importance within the time series data while filtering out noise. (Venkataramani, Figure 7 Event Density Histogram)
Regarding claim 19,
Claim 19 is a non-transitory computer-readable device type claim that recite the same limitation of claim 5. Therefore, claim 19 is rejected using the same rationale as claim 5.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to NAYMUR RAHMAN ALI whose telephone number is (571)272-0007. The examiner can normally be reached Mon-Fri. 9:30-6:30 pm.
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, Alexey Shmatov can be reached at (571)270-3428. 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.
/NAYMUR RAHMAN ALI/Examiner, Art Unit 2123
/ALEXEY SHMATOV/Supervisory Patent Examiner, Art Unit 2123
Prosecution Insights