DETAILED ACTION
This action is responsive to the Application 18/631,200 filed on 04/10/2024. Claims 1-17 are pending in the case. Claims 1, 9 and 10 are independent claims.
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 .
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.
Priority
Foreign Priority of 10/19/2023 of acknowledged and receipt is acknowledged of certified copies of papers required by 37 CFR 1.55.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 04/10/2024 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
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 1-17 are rejected under 35 U.S.C. 101 because the claims are directed towards a judicial exception without significantly more.
Regarding Claim 1:
Subject Matter Eligibility Analysis Step 2A Prong 1:
The claim recites comparing an output…with a first input to extract a first local attribution for a feature of the first input which, under the broadest reasonable interpretation, covers performance of the limitation in the mind. The limitations encompass using judgement to compare values and making an evaluation to determine a value.
The claim recites extracting a portion of the first local attribution of which an absolute value is a threshold or more… which, under the broadest reasonable interpretation, covers performance of the limitation in the mind. The limitations encompass evaluation to choose a set of data. See 2106.04.(a)(2).III.C. Additionally, the broadest reasonable interpretation of the limitation falls under the abstract idea of Mathematical Calculations (see MPEP 2106.04(a)(2)(I)(C)).
The claim recites and updating the first input with a second input by applying an extraction result to the first input which, under the broadest reasonable interpretation, covers performance of the limitation in the mind. The limitations encompass using evaluation to put two numbers together. See 2106.04.(a)(2).III.C. Additionally, the broadest reasonable interpretation of the limitation falls under the abstract idea of Mathematical Calculations (see MPEP 2106.04(a)(2)(I)(C)).
Subject Matter Eligibility Analysis Step 2A Prong 2:
of an artificial neural network model specifies a particular technological environment in which the abstract idea is to take place, i.e. a field of use (see MPEP 2106.05(h))
using a mask extractor recites a generic computer on which to perform the abstract idea, e.g. "apply it on a computer" (see MPEP 2106.05(f))
Subject Matter Eligibility Analysis Step 2B:
Additional elements (a) do not integrate the abstract idea into a practical application nor do the additional limitation provide significantly more than the abstract idea because the limitation merely specifies a field of use in which the abstract idea is to take place, i.e. a field of use (see MPEP 2106.05(h)).
Additional elements (b) do not integrate the abstract idea into a practical application nor do the additional limitation provide significantly more than the abstract idea because the limitation amount to no more than mere instructions to apply the exception using a generic computer component. Please see MPEP §2106.05(f).
The additional element(s) (a) and (b) in the claim do/does not include any additional elements , when considered separately and in combination, that amount to an integration of the judicial exception into a practical application, nor significantly more than the judicial exception for the reasons set forth in step 2A prong 2 analysis above. The claim is not patent eligible.
Regarding Claim 2:
The rejection of claim 1 is incorporated and further claim recites further additional
elements/limitations:
Subject Matter Eligibility Analysis Step 2A Prong 1:
The claim recites comparing the output…with the second input to extract a second local attribution for the second input which, under the broadest reasonable interpretation, covers performance of the limitation in the mind. The limitations encompass using judgement to compare values and making an evaluation to determine a set of data. See 2106.04.(a)(2).III.C.
Subject Matter Eligibility Analysis Step 2A Prong 2:
acquiring an aggregated attribution by aggregating the first local attribution and the second local attribution which amount to mere extra solution activity of obtaining and/or gathering data over a network, see MPEP §2106.05(g)
of the artificial neural network model specifies a particular technological environment in which the abstract idea is to take place, i.e. a field of use (see MPEP 2106.05(h))
Subject Matter Eligibility Analysis Step 2B:
Further, additional element (a) of obtaining a network input is well understood, routine, and conventional activity of “transmitting or receiving data over a network" (see MPEP 2106.05(d)(II)(i) using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362)
Additional elements (b) do not integrate the abstract idea into a practical application nor do the additional limitation provide significantly more than the abstract idea because the limitation merely specifies a field of use in which the abstract idea is to take place, i.e. a field of use (see MPEP 2106.05(h)).
The additional element(s) (a) and (b) in the claim do/does not include any additional elements , when considered separately and in combination, that amount to an integration of the judicial exception into a practical application, nor significantly more than the judicial exception for the reasons set forth in step 2A prong 2 analysis above. The claim is not patent eligible.
Regarding Claim 3:
The rejection of claim 2 is incorporated and further claim recites further additional
elements/limitations:
Subject Matter Eligibility Analysis Step 2A Prong 1:
The claim recites performing postprocessing on the first local attribution and the second local attribution, wherein the postprocessing includes normalization and upsampling which is an abstract idea (Mathematical Calculations (see MPEP 2106.04(a)(2)(I)(C))).
Subject Matter Eligibility Analysis Step 2A Prong 2:
The claim does not contain elements that would warrant a Step 2A Prong 2 analysis.
Subject Matter Eligibility Analysis Step 2B:
The claim does not include any additional element, when considered separately and in combination, that amount to an integration of the judicial exception into a practical application, nor to significantly more than the judicial exception. The claim is not patent eligible.
Regarding Claim 4:
The rejection of claim 1 is incorporated and further claim recites further additional
elements/limitations:
Subject Matter Eligibility Analysis Step 2A Prong 1:
The claim recites set the threshold on the basis of a distribution of the first local attribution which is an abstract idea (Mathematical Relationships (see MPEP 2106.04(a)(2)(I)(A)))).
The claim recites determine an attribution of a portion of the first local attribution which is the threshold or less to be 0 which, under the broadest reasonable interpretation, covers performance of the limitation in the mind. The limitations encompass using judgement to evaluate values to be equal to or less than a number. See 2106.04.(a)(2).III.C. Additionally, the broadest reasonable interpretation of the limitation falls under the abstract idea of Mathematical Relationships (see MPEP 2106.04(a)(2)(I)(A)))).
Subject Matter Eligibility Analysis Step 2A Prong 2:
wherein the mask extractor comprises a first mask configured to recites a generic computer on which to perform the abstract idea, e.g. "apply it on a computer" (see MPEP 2106.05(f))
Subject Matter Eligibility Analysis Step 2B:
Additional elements (a) do not integrate the abstract idea into a practical application nor do the additional limitation provide significantly more than the abstract idea because the limitation amount to no more than mere instructions to apply the exception using a generic computer component. Please see MPEP §2106.05(f).
The additional element(s) (a) in the claim do/does not include any additional elements , when considered separately and in combination, that amount to an integration of the judicial exception into a practical application, nor significantly more than the judicial exception for the reasons set forth in step 2A prong 2 analysis above. The claim is not patent eligible.
Regarding Claim 5:
The rejection of claim 4 is incorporated and further claim recites further additional
elements/limitations:
Subject Matter Eligibility Analysis Step 2A Prong 1:
The claim recites wherein the updating of the first input with the second input comprises removing the portion of the first local attribution of which the attribution is determined to be 0 which, under the broadest reasonable interpretation, covers performance of the limitation in the mind with physical aid. The limitations encompass using evaluation to remove data from a set. See 2106.04.(a)(2).III.C.
Subject Matter Eligibility Analysis Step 2A Prong 2:
The claim does not contain elements that would warrant a Step 2A Prong 2 analysis.
Subject Matter Eligibility Analysis Step 2B:
The claim does not include any additional element, when considered separately and in combination, that amount to an integration of the judicial exception into a practical application, nor to significantly more than the judicial exception. The claim is not patent eligible.
Regarding Claim 6:
The rejection of claim 1 is incorporated and further claim recites further additional
elements/limitations:
Subject Matter Eligibility Analysis Step 2A Prong 1:
The claim recites remove noise and outliers of the first local attribution which, under the broadest reasonable interpretation, covers performance of the limitation in the mind with physical aid. The limitations encompass using evaluation to remove data from a set. See 2106.04.(a)(2).III.C.
Subject Matter Eligibility Analysis Step 2A Prong 2:
wherein the mask extractor comprises a second mask configured to recites a generic computer on which to perform the abstract idea, e.g. "apply it on a computer" (see MPEP 2106.05(f))
Subject Matter Eligibility Analysis Step 2B:
Additional elements (a) do not integrate the abstract idea into a practical application nor do the additional limitation provide significantly more than the abstract idea because the limitation amount to no more than mere instructions to apply the exception using a generic computer component. Please see MPEP §2106.05(f).
The additional element(s) (a) in the claim do/does not include any additional elements , when considered separately and in combination, that amount to an integration of the judicial exception into a practical application, nor significantly more than the judicial exception for the reasons set forth in step 2A prong 2 analysis above. The claim is not patent eligible.
Regarding Claim 7:
The rejection of claim 1 is incorporated and further claim recites further additional
elements/limitations:
Subject Matter Eligibility Analysis Step 2A Prong 1:
The claim recites wherein the first local attribution is a discrete sequence of anchor points which, under the broadest reasonable interpretation, covers performance of the limitation in the mind. The limitations encompass using judgement to compare values and making an evaluation to determine sequence of values .
Subject Matter Eligibility Analysis Step 2A Prong 2:
The claim does not contain elements that would warrant a Step 2A Prong 2 analysis.
Subject Matter Eligibility Analysis Step 2B:
The claim does not include any additional element, when considered separately and in combination, that amount to an integration of the judicial exception into a practical application, nor to significantly more than the judicial exception. The claim is not patent eligible.
Regarding Claim 8:
The rejection of claim 1 is incorporated and further claim recites further additional
elements/limitations:
Subject Matter Eligibility Analysis Step 2A Prong 1:
The claim recites wherein the extracting of the first local attribution comprises generating an attribution heatmap of the first local attribution which, under the broadest reasonable interpretation, covers performance of the limitation in the mind with physical aid. The limitations encompass using evaluation to map data. See 2106.04.(a)(2).III.C.
Subject Matter Eligibility Analysis Step 2A Prong 2:
The claim does not contain elements that would warrant a Step 2A Prong 2 analysis.
Subject Matter Eligibility Analysis Step 2B:
The claim does not include any additional element, when considered separately and in combination, that amount to an integration of the judicial exception into a practical application, nor to significantly more than the judicial exception. The claim is not patent eligible.
Regarding Claim 9:
Subject Matter Eligibility Analysis Step 2A Prong 1:
The claim recites comparing an output…with a first input to extract a first local attribution for a feature of the first input which, under the broadest reasonable interpretation, covers performance of the limitation in the mind. The limitations encompass using judgement to compare values and making an evaluation to determine a value.
The claim recites extracting a portion of the first local attribution of which an absolute value is a threshold or more… which, under the broadest reasonable interpretation, covers performance of the limitation in the mind. The limitations encompass evaluation to choose a set of data. See 2106.04.(a)(2).III.C. Additionally, the broadest reasonable interpretation of the limitation falls under the abstract idea of Mathematical Calculations (see MPEP 2106.04(a)(2)(I)(C)).
The claim recites and updating the first input with a second input by applying an extraction result to the first input which, under the broadest reasonable interpretation, covers performance of the limitation in the mind. The limitations encompass using evaluation to put two numbers together. See 2106.04.(a)(2).III.C. Additionally, the broadest reasonable interpretation of the limitation falls under the abstract idea of Mathematical Calculations (see MPEP 2106.04(a)(2)(I)(C)).
Subject Matter Eligibility Analysis Step 2A Prong 2:
A non-transitory computer-readable recording medium storing instructions that are executed by a processor to perform the operations recites a generic computer on which to perform the abstract idea, e.g. "apply it on a computer" (see MPEP 2106.05(f))
of an artificial neural network model specifies a particular technological environment in which the abstract idea is to take place, i.e. a field of use (see MPEP 2106.05(h))
using a mask extractor recites a generic computer on which to perform the abstract idea, e.g. "apply it on a computer" (see MPEP 2106.05(f))
Subject Matter Eligibility Analysis Step 2B:
Additional elements (b) do not integrate the abstract idea into a practical application nor do the additional limitation provide significantly more than the abstract idea because the limitation merely specifies a field of use in which the abstract idea is to take place, i.e. a field of use (see MPEP 2106.05(h)).
Additional elements (a) and (c) do not integrate the abstract idea into a practical application nor do the additional limitation provide significantly more than the abstract idea because the limitation amount to no more than mere instructions to apply the exception using a generic computer component. Please see MPEP §2106.05(f).
The additional element(s) (a) (b) and (c) in the claim do/does not include any additional elements , when considered separately and in combination, that amount to an integration of the judicial exception into a practical application, nor significantly more than the judicial exception for the reasons set forth in step 2A prong 2 analysis above. The claim is not patent eligible.
Regarding Claim 10:
Subject Matter Eligibility Analysis Step 2A Prong 1:
The claim recites comparing an output…with a first input to extract a first local attribution for a feature of the first input which, under the broadest reasonable interpretation, covers performance of the limitation in the mind. The limitations encompass using judgement to compare values and making an evaluation to determine a value.
The claim recites extracting a portion of the first local attribution of which an absolute value is a threshold or more… which, under the broadest reasonable interpretation, covers performance of the limitation in the mind. The limitations encompass evaluation to choose a set of data. See 2106.04.(a)(2).III.C. Additionally, the broadest reasonable interpretation of the limitation falls under the abstract idea of Mathematical Calculations (see MPEP 2106.04(a)(2)(I)(C)).
The claim recites and updating the first input with a second input by applying an extraction result to the first input which, under the broadest reasonable interpretation, covers performance of the limitation in the mind. The limitations encompass using evaluation to put two numbers together. See 2106.04.(a)(2).III.C. Additionally, the broadest reasonable interpretation of the limitation falls under the abstract idea of Mathematical Calculations (see MPEP 2106.04(a)(2)(I)(C)).
Subject Matter Eligibility Analysis Step 2A Prong 2:
An electronic device for distilling input features, comprising: a memory configured to store instructions; and a processor configured to execute the instructions, wherein the processor is configured to recites a generic computer on which to perform the abstract idea, e.g. "apply it on a computer" (see MPEP 2106.05(f))
of an artificial neural network model specifies a particular technological environment in which the abstract idea is to take place, i.e. a field of use (see MPEP 2106.05(h))
using a mask extractor recites a generic computer on which to perform the abstract idea, e.g. "apply it on a computer" (see MPEP 2106.05(f))
Subject Matter Eligibility Analysis Step 2B:
Additional elements (b) do not integrate the abstract idea into a practical application nor do the additional limitation provide significantly more than the abstract idea because the limitation merely specifies a field of use in which the abstract idea is to take place, i.e. a field of use (see MPEP 2106.05(h)).
Additional elements (a) and (c) do not integrate the abstract idea into a practical application nor do the additional limitation provide significantly more than the abstract idea because the limitation amount to no more than mere instructions to apply the exception using a generic computer component. Please see MPEP §2106.05(f).
The additional element(s) (a) (b) and (c) in the claim do/does not include any additional elements , when considered separately and in combination, that amount to an integration of the judicial exception into a practical application, nor significantly more than the judicial exception for the reasons set forth in step 2A prong 2 analysis above. The claim is not patent eligible.
Regarding claim 11:
The rejection of claim 10 is incorporated and further claim recites further additional elements/limitations:
Claim 11 is rejected under that same 101 claim analysis due to the substantially similarity of the limitations and additional elements of claim 2 found in claim 11.
Regarding claim 12:
The rejection of claim 11 is incorporated and further claim recites further additional elements/limitations:
Claim 12 is rejected under that same 101 claim analysis due to the substantially similarity of the limitations and additional elements of claim 3 found in claim 12.
Regarding claim 13:
The rejection of claim 10 is incorporated and further claim recites further additional elements/limitations:
Claim 13 is rejected under that same 101 claim analysis due to the substantially similarity of the limitations and additional elements of claim 4 found in claim 13.
Regarding claim 14:
The rejection of claim 13 is incorporated and further claim recites further additional elements/limitations:
Claim 14 is rejected under that same 101 claim analysis due to the substantially similarity of the limitations and additional elements of claim 5 found in claim 14.
Regarding claim 15:
The rejection of claim 10 is incorporated and further claim recites further additional elements/limitations:
Claim 15 is rejected under that same 101 claim analysis due to the substantially similarity of the limitations and additional elements of claim 6 found in claim 15.
Regarding claim 16:
The rejection of claim 10 is incorporated and further claim recites further additional elements/limitations:
Claim 16 is rejected under that same 101 claim analysis due to the substantially similarity of the limitations and additional elements of claim 7 found in claim 16.
Regarding claim 17:
The rejection of claim 10 is incorporated and further claim recites further additional elements/limitations:
Claim 17 is rejected under that same 101 claim analysis due to the substantially similarity of the limitations and additional elements of claim 8 found in claim 17.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claim(s) 1-2, 4-5, 7-11, 13-14 and 16-17 is/are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Lee et al(“Attribution Mask: Filtering Out Irrelevant Features By Recursively Focusing Attention on Inputs of DNNs” henceforth known as Lee)
Regarding claim 1:
Lee discloses comparing an output of an artificial neural network model with a first input to extract a first local attribution for a feature of the first input(Lee, Page 4, Col. 2, Algorithm 1, where the recursive masking loop of Algorithm 1 corresponds to comparing an output of an artificial neural network model with a first input to extract a first local attribution for a feature of the first input as the first iteration of the recursive masking loop takes the first input z0 of a deep neural network and extracts a local attribution by calculating a first local attribution A1,k)
Lee discloses extracting a portion of the first local attribution of which an absolute value is a threshold or more(Lee, Page 4, Col. 2, Paragraphs 3-4, “The mask is calculated using MinMax normalization…where the initial mask m0,k is 1, and mi,k, Ai ∈ Rd0. This method normalizes the mask value to between zero and 1” where the MinMax modifies attribution data to fit within boundaries of 0 and 1 corresponds to extracting a portion of the first local attribution of which an absolute value is a threshold or more) using a mask extractor(Lee, Page 4, Col. 2, Algorithm 1, where Algorithm 1 corresponds to a mask extractor as it calculates a mask mi,k and where the first iteration of the recursive masking loop transforming the attribution A1,k through a MinMax function corresponds to extracting a portion of the first local attribution as the function limits the original local attribution values. (See also Lee, Page 4, Col. 2, Paragraph 3, “The mask is calculated using MinMax normalization”))
Lee discloses and updating the first input with a second input by applying an extraction result to the first input(Lee, Page 4, Col. 2, Algorithm 1, where the first iteration of the masking loop takes the first input z0 and the first local attribution A1,k to extract the mask
m
1
,
k
to apply to
z
0
to produce a second input
z
0
∘
m
1
,
k
*
used in the next iteration corresponds to updating the first input with a second input by applying an extraction result to the first input as Algorithm 1 uses an extraction result of a calculated attribution mask for the first input and applies the mask to the previous input producing a second input)
Regarding claim 2:
The rejection of claim 1 with prior art is incorporated and further:
Lee discloses comparing the output of the artificial neural network model with the second input to extract a second local attribution for the second input(Lee, Page 4, Col. 2, Algorithm 1, “Ai,k =
E
x
p
l
(
f
,
z
0
∘
m
i
-
1
,
k
*
,
k
)
” where the first and second iterations of the recursive masking loop generating the attribution A1,k and A2,k corresponds to a first and second local attribution with the second local attribution being calculated with the model, second input and target(
E
x
p
l
(
f
,
z
0
∘
m
1
,
k
*
,
k
)
) corresponding to comparing the output of the artificial neural network model with the second input to extract a second local attribution for the second input as the model is used with the second input to calculate a second local attribution.) and acquiring an aggregated attribution by aggregating the first local attribution and the second local attribution(Lee, Page 4, Col. 2, Algorithm 1, “mi,k = MinMax(Ai,k)” where the MinMax function performing a cumulative normalization on each attribution corresponding to acquiring an aggregated attribution by aggregating the first local attribution and the second local attribution as the normalization is an aggregated attribution representation as it is derived from the first and second attributions)
Regarding claim 4:
The rejection of claim 1 with prior art is incorporated and further:
Lee discloses wherein the mask extractor comprises a first mask configured to set the threshold on the basis of a distribution of the first local attribution and determine an attribution of a portion of the first local attribution which is the threshold or less to be 0(Lee, Page 4, Col. 2, Paragraphs 3-4, “The mask is calculated using MinMax normalization…where the initial mask m0,k is 1, and mi,k, Ai ∈ Rd0. This method normalizes the mask value to between zero and 1” where the MinMax function setting any attribution to the minimum of 0 and the maximum of any attribution to 1 and every intermediate attribution between 0 and 1 corresponds to set the threshold on the basis of a distribution of the first local attribution and determine an attribution of a portion of the first local attribution which is the threshold or less to be 0 as the MinMax function is based on the distribution of the attribution and a portion of the attribution which is the Min threshold is set to be 0.
Regarding claim 5:
The rejection of claim 4 with prior art is incorporated and further:
Lee discloses wherein the updating of the first input with the second input comprises removing the portion of the first local attribution of which the attribution is determined to be 0(Lee, Page 4, Col. 2, Paragraphs 5, “Next, the attribution mask is multiplied by the input features and then the masked features are fed into the DNN” where the minimum attribution has a corresponding mask value of 0 and the paper multiplies the attribution mask by the input features of the second input
z
0
∘
m
1
,
k
is multiplied by zero corresponds to updating of the first input with the second input comprises removing the portion of the first local attribution of which the attribution is determined to be 0 as the first input
z
0
is transformed into second input
z
0
∘
m
1
,
k
and the feature(s) of the first input being multiplied by zero becoming zero corresponds to moving a portion of the first local attribution which is determined to be zero. )
Regarding claim 7:
The rejection of claim 1 with prior art is incorporated and further:
Lee discloses wherein the first local attribution is a discrete sequence of anchor points(Lee, Page 1 , Col. 2, Figure 1, and Lee, Page 4, Col. 2, Algorithm 1, where recursively creating masked inputs/iteration states and recalculating attribution at each one corresponds to a first local attribution is a discrete sequence of anchor points as the attribution is created with the masked inputs/iteration states)
Regarding claim 8:
The rejection of claim 1 with prior art is incorporated and further:
Lee discloses wherein the extracting of the first local attribution comprises generating an attribution heatmap of the first local attribution(Lee, Page 1 , Col. 2, Figure 1, and Lee, Page 1 , Col. 2, Figure 1, and Lee, Page 4, Col. 2, Algorithm 1, where the initial mask = 1 and the first iteration A1,k = Expl(f, z0, k) is an attribution calculated from a first local attribution and Figure 1 shows the visual/spatial representation of A1,k corresponds to extracting of the first local attribution comprises generating an attribution heatmap of the first local attribution as the visual/spatial representation of A1,k represents a heatmap)
Regarding claim 9:
Lee discloses non-transitory computer-readable recording medium storing instructions that are executed by a processor to perform the operations of(Lee, Page 5, Col. 1, Paragraph 5, “By allocating 88 batches for each of the 4 GPUs, the total batch size was 352” where the graphics processing unit and computer corresponds to having non-transitory computer-readable recording medium that are executed by a processor to perform the operations)
Lee discloses comparing an output of an artificial neural network model with a first input to extract a first local attribution for a feature of the first input(Lee, Page 4, Col. 2, Algorithm 1, where the recursive masking loop of Algorithm 1 corresponds to comparing an output of an artificial neural network model with a first input to extract a first local attribution for a feature of the first input as the first iteration of the recursive masking loop takes the first input z0 of a deep neural network and extracts a local attribution by calculating a first local attribution A1,k)
Lee discloses extracting a portion of the first local attribution of which an absolute value is a threshold or more(Lee, Page 4, Col. 2, Paragraphs 3-4, “The mask is calculated using MinMax normalization…where the initial mask m0,k is 1, and mi,k, Ai ∈ Rd0. This method normalizes the mask value to between zero and 1” where the MinMax modifies attribution data to fit within boundaries of 0 and 1 corresponds to extracting a portion of the first local attribution of which an absolute value is a threshold or more) using a mask extractor(Lee, Page 4, Col. 2, Algorithm 1, where Algorithm 1 corresponds to a mask extractor as it calculates a mask mi,k and where the first iteration of the recursive masking loop transforming the attribution A1,k through a MinMax function corresponds to extracting a portion of the first local attribution as the function limits the original local attribution values. (See also Lee, Page 4, Col. 2, Paragraph 3, “The mask is calculated using MinMax normalization”))
Lee discloses and updating the first input with a second input by applying an extraction result to the first input(Lee, Page 4, Col. 2, Algorithm 1, where the first iteration of the masking loop takes the first input z0 and the first local attribution A1,k to extract the mask
m
1
,
k
to apply to
z
0
to produce a second input
z
0
∘
m
1
,
k
*
used in the next iteration corresponds to updating the first input with a second input by applying an extraction result to the first input as Algorithm 1 uses an extraction result of a calculated attribution mask for the first input and applies the mask to the previous input producing a second input)
Regarding claim 10:
Lee discloses a memory configured to store instructions; and a processor configured to execute the instructions, wherein the processor is configured to(Lee, Page 5, Col. 1, Paragraph 5, “By allocating 88 batches for each of the 4 GPUs, the total batch size was 352” where the graphics processing unit and computer corresponds to having a memory for storing instructions and a processing for executing instructions)
Lee discloses comparing an output of an artificial neural network model with a first input to extract a first local attribution for a feature of the first input(Lee, Page 4, Col. 2, Algorithm 1, where the recursive masking loop of Algorithm 1 corresponds to comparing an output of an artificial neural network model with a first input to extract a first local attribution for a feature of the first input as the first iteration of the recursive masking loop takes the first input z0 of a deep neural network and extracts a local attribution by calculating a first local attribution A1,k)
Lee discloses extracting a portion of the first local attribution of which an absolute value is a threshold or more(Lee, Page 4, Col. 2, Paragraphs 3-4, “The mask is calculated using MinMax normalization…where the initial mask m0,k is 1, and mi,k, Ai ∈ Rd0. This method normalizes the mask value to between zero and 1” where the MinMax modifies attribution data to fit within boundaries of 0 and 1 corresponds to extracting a portion of the first local attribution of which an absolute value is a threshold or more) using a mask extractor(Lee, Page 4, Col. 2, Algorithm 1, where Algorithm 1 corresponds to a mask extractor as it calculates a mask mi,k and where the first iteration of the recursive masking loop transforming the attribution A1,k through a MinMax function corresponds to extracting a portion of the first local attribution as the function limits the original local attribution values. (See also Lee, Page 4, Col. 2, Paragraph 3, “The mask is calculated using MinMax normalization”))
Lee discloses and updating the first input with a second input by applying an extraction result to the first input(Lee, Page 4, Col. 2, Algorithm 1, where the first iteration of the masking loop takes the first input z0 and the first local attribution A1,k to extract the mask
m
1
,
k
to apply to
z
0
to produce a second input
z
0
∘
m
1
,
k
*
used in the next iteration corresponds to updating the first input with a second input by applying an extraction result to the first input as Algorithm 1 uses an extraction result of a calculated attribution mask for the first input and applies the mask to the previous input producing a second input)
Regarding claim 11:
The rejection of claim 10 is incorporated in claim 11. Claim 11 is rejected under the same rationale as set forth in the rejection of claim 2.
Regarding claim 13:
The rejection of claim 10 is incorporated in claim 13. Claim 13 is rejected under the same rationale as set forth in the rejection of claim 4.
Regarding claim 14:
The rejection of claim 13 is incorporated in claim 14. Claim 14 is rejected under the same rationale as set forth in the rejection of claim 5.
Regarding claim 16:
The rejection of claim 10 is incorporated in claim 16. Claim 16 is rejected under the same rationale as set forth in the rejection of claim 7.
Regarding claim 17:
The rejection of claim 10 is incorporated in claim 17. Claim 17 is rejected under the same rationale as set forth in the rejection of claim 8.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claim(s) 3 and 12 is/are rejected under 35 U.S.C. 103 as being unpatentable over Lee et al(“Attribution Mask: Filtering Out Irrelevant Features By Recursively Focusing Attention on Inputs of DNNs” henceforth known as Lee) in view of Mundhenk et al( “Efficient Saliency Maps for Explainable AI” henceforth known as Mundhenk)
Regarding claim 3:
The rejection of claim 2 with prior art is incorporated and further:
Lee discloses wherein the acquiring of the aggregated attribution performing postprocessing on the first local attribution and the second local attribution, wherein the postprocessing includes normalization (Lee, Page 4, Col. 2, Paragraph 3, “The mask is calculated using MinMax normalization”)
Lee does not explicitly disclose, however Mundhenk does disclose upsampling
and upsampling(Mundhenk, Page 5, Paragraph 2, “We then create a combined saliency map by taking the weighted average of the maps. Since they are at different scales, they are upsampled via bilinear interpolation to match the dimensions of the input image.”)
References Lee and Mundhenk are analogous art because they are from the same field of endeavor of using explainable AI for deep neural networks that identify parts of an image that are important to model predictions.
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Lee and Mundhenk before him or her, to modify the postprocessing of Lee to include the upsampling of Mundhenk to match the dimensions of other images/maps to perform operations. The suggestion/motivation for doing so would have been Mundhenk, Page 5, Paragraph 2, “…saliency maps that have been bilinear interpolated (upsampled) to the original input image size p,q, they are then combined”
Regarding claim 12:
The rejection of claim 11 is incorporated in claim 12. Claim 12 is rejected under the same rationale as set forth in the rejection of claim 3.
Claim(s) 6 and 15 is/are rejected under 35 U.S.C. 103 as being unpatentable over Lee et al(“Attribution Mask: Filtering Out Irrelevant Features By Recursively Focusing Attention on Inputs of DNNs” henceforth known as Lee) in view of Kim et al(“Why are Saliency Maps Noisy? Cause of and Solution to Noisy Saliency Maps” henceforth known as Kim)
Regarding claim 6:
The rejection of claim 1 with prior art is incorporated and further:
Lee does not disclose, however Kim discloses wherein the mask extractor comprises a second mask configured to remove noise(Kim, Page 4, Col. 1, Paragraph 4, “To this end, we propose RectifiedGradient, or RectGrad in short, where the gradient propagates only through units whose importance scores exceed some threshold. By construction. RectGrad drops irrelevant features while retaining relevant features in a layer-wise fashion.” Where the layer-by-layer threshold acts as a mask by blocked weak and irrelevant signals before they reach the final saliency map(See also Kim, Page 7, Col. 1, Paragraph 1, “We also observed that RectGrad does not suffer from this problem since it thresholds irrelevant features at every layer and hence stops noise accumulation.”)) and outliers of the first local attribution(Kim, Page 23, Paragraph 1, “To visualize the attributions, we summed up the attributions along the color channel and then capped low outlying values to 0.5th percentile and high outlying values to 99.5th percentile for RGB images. We only capped outlying values for grayscale Images”)
References Lee and Kim are analogous art because they are from the same field of endeavor of using gradient-based attribution methods for identifying important image features in neural networks..
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Lee and Kim before him or her, to modify the noise condition of Lee to use the layer-by-layer noise and outlier mask detection of Kim to reduce noise and irrelevant features. The suggestion/motivation for doing so would have been Kim, Page 6, Co. 2, Paragraph 3, “We observed that the background noise in saliency maps is due to noise accumulation. Specifically, irrelevant features may have relatively small gradient at high intermediate layers. However, since gradient is calculated by successive multiplication, the noise grows exponentially as gradient is propagated towards the input layer. This results in confusing attribution maps which assign high attribution to irrelevant regions. We also observed that RectGrad does not suffer from this problem since it thresholds irrelevant features at every layer and hence stops noise accumulation.”
Regarding claim 15:
The rejection of claim 10 is incorporated in claim 15. Claim 15 is rejected under the same rationale as set forth in the rejection of claim 6.
Relevant Art:
While not used in the current rejection the following prior arts were found the be relevant to the disclosure:
Giyoung Jeon, “Distilled Gradient Aggregation: Purify Features for Input Attribution in the Deep Neural Network” as a possible 102 reference, however possible disclosure as 2 of the inventors are listed as 2 out of the three authors.
Ruth Fong, “Understanding Deep Networks via Extremal Perturbations and Smooth Masks” discusses identify the salient channels with feature inversion to explain model behavior.
Ruth C. Fong “Interpretable Explanations of Black Boxes by Meaningful Perturbation” discusses creating a learned mask and attribution of features based on perturbation
Sara Hooker, “A Benchmark for Interpretability Methods in Deep Neural Networks” that discusses interpretability of neural network models
Siyue Wu, “AD-KD: Attribution-Driven Knowledge Distillation for Language Model Compression” that uses a teacher-student driven attribution that uses the teacher as a mask for the student to learn
Weiqi Wang, “Outlier Denoising Using a Novel Statistics-Based Mask Strategy for Compressive Sensing” that discusses using masks for denoising
Daniel Smilkov, “SmoothGrad: removing noise by adding noise” that discusses-based sensitivity maps and visualization of maps
Julius Adebay, “Sanity Checks for Saliency Maps” that discusses using salient maps for attribution and finding noise/outliers in data
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to CHARLES JEFFREY JONES JR whose telephone number is (703)756-1414. The examiner can normally be reached Monday - Friday 8:00 - 5:00 EST.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Kakali Chaki can be reached at 571-272-3719. 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.
/C.J.J./Examiner, Art Unit 2122
/KAKALI CHAKI/Supervisory Patent Examiner, Art Unit 2122
Prosecution Insights