DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
This action is in response to amendments filed June 5th, 2026, in which claims 1, 8, and 9 have been amended and claim 11 has been added. No claims have been cancelled. The amendments have been entered, and claims 1-11 are currently pending in the case. Claims 1, 8, and 9 are independent claims.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on April 7th, 2026 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 § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claim 11 is rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim 11 recites the limitation "converting the similarity calculated for each of the at least two partial regions of the specific layer". Claim 11 is dependent upon claim 1, and therefore there is insufficient antecedent basis for this limitation in the claim. For examination purposes, this limitation has been interpreted as being dependent upon claim 2.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1 and 7-10 are rejected under 35 U.S.C. 103 as being unpatentable over Yeh et al. ("On Completeness-aware Concept-Based Explanations in Deep Neural Networks", Yeh et al., 2020) (hereinafter “Yeh”) in view of Nakakimura et al. (US 2017/0356889 A1) (hereinafter “Nakakimura”) in further view of CONTEXT (“CONTEXT Version 4: Neural Networks for Text Categorization”, 18 July 2017).
Regarding claim 1:
Yeh teaches, [a] method for discriminating a class of an object (Yeh, page 7, section 5.2, ¶1 “We perform experiments on Animals with Attribute (AwA) [Lampert et al., 2009] that contains 50 animal classes.”) …using a vector neural network type machine learning model (Yeh, page 6, definition 4.2, ¶1 “Given a prediction
f
x
=
h
ϕ
x
…”),
the machine learning model including, from an input data side:
a convolutional layer that receives the input data (Yeh, page 3, section 3, ¶1 “For DNNs that build up by processing parts of input at a time, such as those composed of convolutional layers…”),
a plurality of vector neuron layers that are consecutively arranged to receive a vector input from a preceding layer and output a vector output to a subsequent layer (Yeh, page 3, section 3, ¶1 “We assume that the pre-trained DNN model can be decomposed into two functions: the first part
Φ
⋅
maps input xi into an intermediate layer
Φ
(
x
i
)
, and the second part
h
⋅
maps the intermediate layer
Φ
(
x
i
)
to the output
h
(
Φ
(
x
i
)
)
, which is a probability vector for each class, and
h
y
(
Φ
(
x
i
)
)
, is the probability of data x being predicted as label y by the model f.”), and
a classification vector neuron layer that receives a vector input from a last layer of the plurality of vector neuron layers and outputs a classification result of the input data (Yeh, page 3, section 3, ¶1 “We assume that the pre-trained DNN model can be decomposed into two functions: the first part
Φ
⋅
maps input xi into an intermediate layer
Φ
(
x
i
)
, and the second part
h
⋅
maps the intermediate layer
Φ
(
x
i
)
to the output
h
(
Φ
(
x
i
)
)
, which is a probability vector for each class, and
h
y
(
Φ
(
x
i
)
)
, is the probability of data x being predicted as label y by the model f.” here label y can be considered the classification result of the input data (x)),
the method performed by one or more processors (Yeh, page 1, footnote 1 “The code is released at https://github.com/chihkuanyeh/concept_exp.” Here, the use of computer code inherently teaches a processor) and comprising:
(a1) obtaining from a non-transitory computer readable medium for each class of one or more classes discriminable by the machine learning model (Yeh, pages 7-8, section 5.2, ¶1 “For each class, the top concepts based on the conceptSHAP are the most important concepts to classify this class, as shown in Fig.3.”), a known feature … group obtained based on an output of a specific layer (Yeh, page 3, section 3, ¶1 “To choose the intermediate layer to apply concepts, we follow previous works on concept explanations Kim et al. [2018], Ghorbani et al. [2019] by starting from the layer closest to the prediction until we reached a layer that user is happy with…” here, the intermediate layer can be considered the specific layer) among the plurality of vector neuron layers arranged between the convolutional layer and the classification vector neuron layer of the machine learning model (Yeh, page 6, section 5.1, ¶1 “We take the last convolution layer as the feature layer
Φ
x
.” Here the feature layer can be considered a classification layer) when a plurality of pieces of training data are input to the convolutional layer of the machine learning model (Yeh, page 5, ¶4 “Intuitively, we require that the top-K nearest neighbor training input patches of each concept to be sufficiently close to the concept, and different concepts are as different as possible.”);
(b) executing a class discrimination processing of the object by inputting … the object as the input data to the convolutional layer of the machine learning model (Yeh, page 7, section 5.2, ¶1 “We use the Inception-V3 model, pre-trained on Imagenet [Szegedy et al., 2016], which yields 0.9 test accuracy.” Here, the use of the CNN model Inception-V3 can be considered the executing a class discrimination processing),
wherein:
the (b) includes:
(b0) obtaining a class discrimination result, which is an output from the classification vector neuron layer (Yeh, page 6, section 5.1, ¶1 “We take the last convolution layer as the feature layer
Φ
x
.” Here the feature layer can be considered a classification layer) upon inputting the … data of the object to the convolutional layer of the machine learning model, the class discrimination result indicating a class among the one or more classes determined for the object (Yeh, page 3, section 3, ¶1 “We assume that the pre-trained DNN model can be decomposed into two functions: the first part
Φ
⋅
maps input xi into an intermediate layer
Φ
(
x
i
)
, and the second part
h
⋅
maps the intermediate layer
Φ
(
x
i
)
to the output
h
(
Φ
(
x
i
)
)
, which is a probability vector for each class, and
h
y
(
Φ
(
x
i
)
)
, is the probability of data x being predicted as label y by the model f.” here label y can be considered the a class among the one or more classes determined for the object),
(b1) calculating a feature … of a vector output (Yeh, page 3, ¶3 “Based on this motivation, we define the concept product for part of data xt as
PNG
media_image1.png
28
338
media_image1.png
Greyscale
, where TH is a threshold which trims value less than β to 0.”) that is output from the specific layer among the plurality of vector neuron layers arranged between the convolutional layer and the classification vector neuron layer of the machine learning model when the … data of the object is input to the convolutional layer of the machine learning model Yeh, page 6, section 5.1, ¶1 “We take the last convolution layer as the feature layer
Φ
x
.”);
(b3) creating …text for the class…(Yeh, pages 8-9, section 5.3, ¶2 “For each concept, Table 2 shows (a) the top nearest neighbors based on the dot product of the concept and part of reviews (b) the most frequent words in the top-500 nearest neighbors (excluding stop words) (c) the conceptSHAP score for each concept.”) that is output from the specific layer, which is neither the convolutional layer that receives the spectral data nor the classification vector neuron layer that outputs the class discrimination result (Yeh, page 6, section 5.1, ¶1 “We take the last convolution layer as the feature layer
Φ
x
.” Here the feature layer is neither the convolutional layer that receives the spectral data nor the classification vector neuron layer that outputs the class discrimination result),…
Yeh does not teach “…using a spectrometer…;
…spectrum group;
(a2) measuring the object with the spectrometer and obtaining spectral data of the object from the spectrometer;
…spectral data…
…spectral data…
…calculating a feature spectrum;
…spectral data…
(b2) calculating a similarity between the feature spectrum and the known feature spectrum group;
…based on the calculated similarity between the feature spectrum and the known feature spectrum;”
However, Nakakimura teaches
…using a spectrometer… (Nakakimura, ¶3 “The present invention is preferably used to process three-dimensional spectral data obtained by, for example, a Liquid Chromatograph Mass Spectrometer (LC-MS), a Gas Chromatograph Mass Spectrometer (GC-MS), a liquid chromatograph using a multichannel type detector such as, e.g., a photodiode array (PDA) detector, a liquid chromatograph or a gas chromatograph using an ultraviolet-visible spectrophotometer or an infrared spectrophotometer capable of wavelength scanning as a detector, or an imaging mass spectrometer, etc.”)
…spectrum group (Nakakimura, ¶20 “a) a characteristic spectrum acquisition unit configured to perform multivariate analysis by considering a plurality of spectrums constituting a single three-dimensional spectral data obtained from a specific sample among a plurality of samples as a collection of a single spectrum not depending on a value of the second parameter, and based on a result of the multivariate analysis, one or a plurality of characteristic spectrums that characterize the specific sample is obtained;”)
(a2) measuring the object with the spectrometer and obtaining spectral data of the object from the spectrometer (Nakakimura, ¶3 “The present invention is preferably used to process three-dimensional spectral data obtained by, for example, a Liquid Chromatograph Mass Spectrometer (LC-MS), a Gas Chromatograph Mass Spectrometer (GC-MS), a liquid chromatograph using a multichannel type detector such as, e.g., a photodiode array (PDA) detector, a liquid chromatograph or a gas chromatograph using an ultraviolet-visible spectrophotometer or an infrared spectrophotometer capable of wavelength scanning as a detector, or an imaging mass spectrometer, etc.”) and
…spectral data… (Nakakimura, ¶20 “a) a characteristic spectrum acquisition unit configured to perform multivariate analysis by considering a plurality of spectrums constituting a single three-dimensional spectral data obtained from a specific sample among a plurality of samples as a collection of a single spectrum not depending on a value of the second parameter, and based on a result of the multivariate analysis, one or a plurality of characteristic spectrums that characterize the specific sample is obtained;”)
…spectral data… (Nakakimura, ¶20 “a) a characteristic spectrum acquisition unit configured to perform multivariate analysis by considering a plurality of spectrums constituting a single three-dimensional spectral data obtained from a specific sample among a plurality of samples as a collection of a single spectrum not depending on a value of the second parameter, and based on a result of the multivariate analysis, one or a plurality of characteristic spectrums that characterize the specific sample is obtained;”)
…calculating a feature spectrum (Nakakimura, ¶30 “Normally, since a plurality of characteristic spectrums are obtained, the spectrum similarity calculation unit calculates the similarity between each spectrum at each measurement time extracted from the three-dimensional spectral data to a single sample and a characteristic spectrum for each of three-dimensional spectral data with respect to a plurality of samples for each characteristic spectrum.”)
…spectral data… (Nakakimura, ¶20 “a) a characteristic spectrum acquisition unit configured to perform multivariate analysis by considering a plurality of spectrums constituting a single three-dimensional spectral data obtained from a specific sample among a plurality of samples as a collection of a single spectrum not depending on a value of the second parameter, and based on a result of the multivariate analysis, one or a plurality of characteristic spectrums that characterize the specific sample is obtained;”)
(b2) calculating a similarity between the feature spectrum and the known feature spectrum group (Nakakimura, ¶30 “Normally, since a plurality of characteristic spectrums are obtained, the spectrum similarity calculation unit calculates the similarity between each spectrum at each measurement time extracted from the three-dimensional spectral data to a single sample and a characteristic spectrum for each of three-dimensional spectral data with respect to a plurality of samples for each characteristic spectrum.”)
…based on the calculated similarity between the feature spectrum…and the known feature spectrum (Nakakimura, ¶30 “Normally, since a plurality of characteristic spectrums are obtained, the spectrum similarity calculation unit calculates the similarity between each spectrum at each measurement time extracted from the three-dimensional spectral data to a single sample and a characteristic spectrum for each of three-dimensional spectral data with respect to a plurality of samples for each characteristic spectrum.”);
Yeh and Nakakimura are analogous art because both references concern methods for data processing. Accordingly, it would have been obvious to a person of ordinary skill in the art, before the effective filing date of the claimed invention, to modify Yeh’s image analysis method to incorporate the spectrums taught by Nakakimura. The motivation for doing so would have been to improve the throughput of the whole difference analysis, as stated in Nakakimura, ¶38 “Further, in the three-dimensional spectral data processing device and processing method according to the present invention, since a second parameter such as a retention time, etc., is not taken into account when determining the characteristic spectrum, no alignment processing is required for aligning the retention time among a plurality of samples which are normally required when obtaining a two-dimensional characteristic data table including characteristic data for a plurality of samples, and the time and effort required for such processing can be saved. As a result, the throughput of the whole difference analysis can be improved.”
Yeh in view of Nakakimura does not teach “creating an explanatory text…
(b4) outputting, for display on a user device, the class discrimination result with the explanatory text indicating a reason for the class determined for the object.”
However, CONTEXT teaches creating an explanatory text… (CONTEXT, page 16, section 3.2 “Prediction file (text output) Optionally, predict writes prediction values in the text format, one data point per line.” The prediction values in text format can be considered an explanatory text)…
(b4) outputting, for display on a user device, the class discrimination result (CONTEXT, page 16, section 3.2 “Prediction file (text output) Optionally, predict writes prediction values in the text format, one data point per line.” The text output of CONTEXT can be considered outputting the explanatory text and as a data file will be read on a user device) with the explanatory text indicating a reason for the class determined for the object (CONTEXT, page 16, section 3.2 “reNet predictapplies a model saved during training to new data and write prediction values to a file.” Here, the prediction values can be considered a reason).
Yeh in view of Nakakimura and CONTEXT are analogous art because both references concern classification using convolution neural networks. Accordingly, it would have been obvious to a person of ordinary skill in the art, before the effective filing date of the claimed invention, to have combined the text output of CONTEXT to the teachings of Yeh in view of Nakakimura. The motivation for doing so would have been to have a structured method of outputting the discrimination result, as stated in CONTEXT, col 22, lines 41-48, “The application server provides access control services in cooperation with the data store and is able to generate content such as text, graphics, audio and/or video to be transferred to the user, which may be served to the user by the Web server in the form of HTML, XML or another appropriate structured language in this example.”.
Regarding claim 7:
Yeh in view of Nakakimura in view of CONTEXT teaches [t]he method according to claim 1, wherein the (b4) includes:
displaying a discrimination result list in which the class discrimination result and the explanatory text are arranged for two or more classes among a plurality of classes that are discriminable by the machine learning model (CONTEXT, page 16, section 3.2 “Prediction file (text output) Optionally, predict writes prediction values in the text format, one data point per line. However, this could be inefficient in space/time when the number of data points or the number of classes is large.
PNG
media_image2.png
226
651
media_image2.png
Greyscale
” here, the WriteText format of the prediction can be considered the explanatory text for the classes discriminated).
It would have been obvious to combine the teachings of Yeh in view of Ma and CONTEXT for the reasons set forth in connection with claim 1 above.
Regarding claim 8:
Yeh teaches, [a] … system including … an information processing device that executes a class discrimination processing of discriminating a class of an object (Yeh, page 7, section 5.2, ¶1 “We perform experiments on Animals with Attribute (AwA) [Lampert et al., 2009] that contains 50 animal classes.”) using …a vector neural network type machine learning model Yeh, page 6, definition 4.2, ¶1 “Given a prediction
f
x
=
h
ϕ
x
…”),
the machine learning model including, from an input data side:
a convolutional layer that receives the input data (Yeh, page 3, section 3, ¶1 “For DNNs that build up by processing parts of input at a time, such as those composed of convolutional layers…”),
a plurality of vector neuron layers that are consecutively arranged to receive a vector input from a preceding layer and output a vector output to a subsequent layer (Yeh, page 3, section 3, ¶1 “We assume that the pre-trained DNN model can be decomposed into two functions: the first part
Φ
⋅
maps input xi into an intermediate layer
Φ
(
x
i
)
, and the second part
h
⋅
maps the intermediate layer
Φ
(
x
i
)
to the output
h
(
Φ
(
x
i
)
)
, which is a probability vector for each class, and
h
y
(
Φ
(
x
i
)
)
, is the probability of data x being predicted as label y by the model f.”), and
a classification vector neuron layer that receives a vector input from a last layer of the plurality of vector neuron layers and outputs a classification result of the input data (Yeh, page 3, section 3, ¶1 “We assume that the pre-trained DNN model can be decomposed into two functions: the first part
Φ
⋅
maps input xi into an intermediate layer
Φ
(
x
i
)
, and the second part
h
⋅
maps the intermediate layer
Φ
(
x
i
)
to the output
h
(
Φ
(
x
i
)
)
, which is a probability vector for each class, and
h
y
(
Φ
(
x
i
)
)
, is the probability of data x being predicted as label y by the model f.” here label y can be considered the classification result of the input data (x)),
the method performed by one or more processors (Yeh, page 1, footnote 1 “The code is released at https://github.com/chihkuanyeh/concept_exp.” Here, the use of computer code inherently teaches a processor) and comprising:
(a1) obtaining from a non-transitory computer readable medium for each class of one or more classes discriminable by the machine learning model (Yeh, pages 7-8, section 5.2, ¶1 “For each class, the top concepts based on the conceptSHAP are the most important concepts to classify this class, as shown in Fig.3.”), a known feature … group obtained based on an output of a specific layer (Yeh, page 3, section 3, ¶1 “To choose the intermediate layer to apply concepts, we follow previous works on concept explanations Kim et al. [2018], Ghorbani et al. [2019] by starting from the layer closest to the prediction until we reached a layer that user is happy with…” here, the intermediate layer can be considered the specific layer) among the plurality of vector neuron layers arranged between the convolutional layer and the classification vector neuron layer of the machine learning model (Yeh, page 6, section 5.1, ¶1 “We take the last convolution layer as the feature layer
Φ
x
.” Here the feature layer can be considered a classification layer) when a plurality of pieces of training data are input to the convolutional layer of the machine learning model (Yeh, page 5, ¶4 “Intuitively, we require that the top-K nearest neighbor training input patches of each concept to be sufficiently close to the concept, and different concepts are as different as possible.”);
(b) executing a class discrimination processing of the object by inputting … the object as the input data to the convolutional layer of the machine learning model (Yeh, page 7, section 5.2, ¶1 “We use the Inception-V3 model, pre-trained on Imagenet [Szegedy et al., 2016], which yields 0.9 test accuracy.” Here, the use of the CNN model Inception-V3 can be considered the executing a class discrimination processing),
wherein:
the (b) includes:
(b0) obtaining a class discrimination result, which is an output from the classification vector neuron layer (Yeh, page 6, section 5.1, ¶1 “We take the last convolution layer as the feature layer
Φ
x
.” Here the feature layer can be considered a classification layer) upon inputting the … data of the object to the convolutional layer of the machine learning model, the class discrimination result indicating a class among the one or more classes determined for the object (Yeh, page 3, section 3, ¶1 “We assume that the pre-trained DNN model can be decomposed into two functions: the first part
Φ
⋅
maps input xi into an intermediate layer
Φ
(
x
i
)
, and the second part
h
⋅
maps the intermediate layer
Φ
(
x
i
)
to the output
h
(
Φ
(
x
i
)
)
, which is a probability vector for each class, and
h
y
(
Φ
(
x
i
)
)
, is the probability of data x being predicted as label y by the model f.” here label y can be considered the a class among the one or more classes determined for the object),
(b1) calculating a feature … of a vector output (Yeh, page 3, ¶3 “Based on this motivation, we define the concept product for part of data xt as
PNG
media_image1.png
28
338
media_image1.png
Greyscale
, where TH is a threshold which trims value less than β to 0.”) that is output from the specific layer among the plurality of vector neuron layers arranged between the convolutional layer and the classification vector neuron layer of the machine learning model when the … data of the object is input to the convolutional layer of the machine learning model Yeh, page 6, section 5.1, ¶1 “We take the last convolution layer as the feature layer
Φ
x
.”);
(b3) creating …text for the class…(Yeh, pages 8-9, section 5.3, ¶2 “For each concept, Table 2 shows (a) the top nearest neighbors based on the dot product of the concept and part of reviews (b) the most frequent words in the top-500 nearest neighbors (excluding stop words) (c) the conceptSHAP score for each concept.”) that is output from the specific layer, which is neither the convolutional layer that receives the spectral data nor the classification vector neuron layer that outputs the class discrimination result (Yeh, page 6, section 5.1, ¶1 “We take the last convolution layer as the feature layer
Φ
x
.” Here the feature layer is neither the convolutional layer that receives the spectral data nor the classification vector neuron layer that outputs the class discrimination result),…
Yeh does not teach “…using a spectrometer…;
…spectrum group;
(a2) measuring the object with the spectrometer and obtaining spectral data of the object from the spectrometer;
…spectral data…
…spectral data…
…calculating a feature spectrum;
…spectral data…
(b2) calculating a similarity between the feature spectrum and the known feature spectrum group;
…based on the calculated similarity between the feature spectrum and the known feature spectrum;”
However, Nakakimura teaches
…using a spectrometer… (Nakakimura, ¶3 “The present invention is preferably used to process three-dimensional spectral data obtained by, for example, a Liquid Chromatograph Mass Spectrometer (LC-MS), a Gas Chromatograph Mass Spectrometer (GC-MS), a liquid chromatograph using a multichannel type detector such as, e.g., a photodiode array (PDA) detector, a liquid chromatograph or a gas chromatograph using an ultraviolet-visible spectrophotometer or an infrared spectrophotometer capable of wavelength scanning as a detector, or an imaging mass spectrometer, etc.”)
…spectrum group (Nakakimura, ¶20 “a) a characteristic spectrum acquisition unit configured to perform multivariate analysis by considering a plurality of spectrums constituting a single three-dimensional spectral data obtained from a specific sample among a plurality of samples as a collection of a single spectrum not depending on a value of the second parameter, and based on a result of the multivariate analysis, one or a plurality of characteristic spectrums that characterize the specific sample is obtained;”)
(a2) measuring the object with the spectrometer and obtaining spectral data of the object from the spectrometer (Nakakimura, ¶3 “The present invention is preferably used to process three-dimensional spectral data obtained by, for example, a Liquid Chromatograph Mass Spectrometer (LC-MS), a Gas Chromatograph Mass Spectrometer (GC-MS), a liquid chromatograph using a multichannel type detector such as, e.g., a photodiode array (PDA) detector, a liquid chromatograph or a gas chromatograph using an ultraviolet-visible spectrophotometer or an infrared spectrophotometer capable of wavelength scanning as a detector, or an imaging mass spectrometer, etc.”) and
…spectral data… (Nakakimura, ¶20 “a) a characteristic spectrum acquisition unit configured to perform multivariate analysis by considering a plurality of spectrums constituting a single three-dimensional spectral data obtained from a specific sample among a plurality of samples as a collection of a single spectrum not depending on a value of the second parameter, and based on a result of the multivariate analysis, one or a plurality of characteristic spectrums that characterize the specific sample is obtained;”)
…spectral data… (Nakakimura, ¶20 “a) a characteristic spectrum acquisition unit configured to perform multivariate analysis by considering a plurality of spectrums constituting a single three-dimensional spectral data obtained from a specific sample among a plurality of samples as a collection of a single spectrum not depending on a value of the second parameter, and based on a result of the multivariate analysis, one or a plurality of characteristic spectrums that characterize the specific sample is obtained;”)
…calculating a feature spectrum (Nakakimura, ¶30 “Normally, since a plurality of characteristic spectrums are obtained, the spectrum similarity calculation unit calculates the similarity between each spectrum at each measurement time extracted from the three-dimensional spectral data to a single sample and a characteristic spectrum for each of three-dimensional spectral data with respect to a plurality of samples for each characteristic spectrum.”)
…spectral data… (Nakakimura, ¶20 “a) a characteristic spectrum acquisition unit configured to perform multivariate analysis by considering a plurality of spectrums constituting a single three-dimensional spectral data obtained from a specific sample among a plurality of samples as a collection of a single spectrum not depending on a value of the second parameter, and based on a result of the multivariate analysis, one or a plurality of characteristic spectrums that characterize the specific sample is obtained;”)
(b2) calculating a similarity between the feature spectrum and the known feature spectrum group (Nakakimura, ¶30 “Normally, since a plurality of characteristic spectrums are obtained, the spectrum similarity calculation unit calculates the similarity between each spectrum at each measurement time extracted from the three-dimensional spectral data to a single sample and a characteristic spectrum for each of three-dimensional spectral data with respect to a plurality of samples for each characteristic spectrum.”)
…based on the calculated similarity between the feature spectrum…and the known feature spectrum (Nakakimura, ¶30 “Normally, since a plurality of characteristic spectrums are obtained, the spectrum similarity calculation unit calculates the similarity between each spectrum at each measurement time extracted from the three-dimensional spectral data to a single sample and a characteristic spectrum for each of three-dimensional spectral data with respect to a plurality of samples for each characteristic spectrum.”);
Yeh and Nakakimura are analogous art because both references concern methods for data processing. Accordingly, it would have been obvious to a person of ordinary skill in the art, before the effective filing date of the claimed invention, to modify Yeh’s image analysis method to incorporate the spectrums taught by Nakakimura. The motivation for doing so would have been to improve the throughput of the whole difference analysis, as stated in Nakakimura, ¶38 “Further, in the three-dimensional spectral data processing device and processing method according to the present invention, since a second parameter such as a retention time, etc., is not taken into account when determining the characteristic spectrum, no alignment processing is required for aligning the retention time among a plurality of samples which are normally required when obtaining a two-dimensional characteristic data table including characteristic data for a plurality of samples, and the time and effort required for such processing can be saved. As a result, the throughput of the whole difference analysis can be improved.”
Yeh in view of Nakakimura does not teach “creating an explanatory text…
(b4) outputting, for display on a user device, the class discrimination result with the explanatory text indicating a reason for the class determined for the object.”
However, CONTEXT teaches creating an explanatory text… (CONTEXT, page 16, section 3.2 “Prediction file (text output) Optionally, predict writes prediction values in the text format, one data point per line.” The prediction values in text format can be considered an explanatory text)…
(b4) outputting, for display on a user device, the class discrimination result (CONTEXT, page 16, section 3.2 “Prediction file (text output) Optionally, predict writes prediction values in the text format, one data point per line.” The text output of CONTEXT can be considered outputting the explanatory text and as a data file will be read on a user device) with the explanatory text indicating a reason for the class determined for the object (CONTEXT, page 16, section 3.2 “reNet predictapplies a model saved during training to new data and write prediction values to a file.” Here, the prediction values can be considered a reason).
Yeh in view of Nakakimura and CONTEXT are analogous art because both references concern classification using convolution neural networks. Accordingly, it would have been obvious to a person of ordinary skill in the art, before the effective filing date of the claimed invention, to have combined the text output of CONTEXT to the teachings of Yeh in view of Nakakimura. The motivation for doing so would have been to have a structured method of outputting the discrimination result, as stated in CONTEXT, col 22, lines 41-48, “The application server provides access control services in cooperation with the data store and is able to generate content such as text, graphics, audio and/or video to be transferred to the user, which may be served to the user by the Web server in the form of HTML, XML or another appropriate structured language in this example.”.
Regarding claim 9:
Yeh teaches, [a] … non-transitory computer-readable storage medium storing a computer program causing a processor to (Yeh, page 1, footnote 1 “The code is released at https://github.com/chihkuanyeh/concept_exp.” Here, the use of computer code inherently teaches a storage medium and processor) execute a class discrimination processing of discriminating a class of an object (Yeh, page 7, section 5.2, ¶1 “We perform experiments on Animals with Attribute (AwA) [Lampert et al., 2009] that contains 50 animal classes.”) using …a vector neural network type machine learning model Yeh, page 6, definition 4.2, ¶1 “Given a prediction
f
x
=
h
ϕ
x
…”),
the machine learning model including, from an input data side:
a convolutional layer that receives the input data (Yeh, page 3, section 3, ¶1 “For DNNs that build up by processing parts of input at a time, such as those composed of convolutional layers…”),
a plurality of vector neuron layers that are consecutively arranged to receive a vector input from a preceding layer and output a vector output to a subsequent layer (Yeh, page 3, section 3, ¶1 “We assume that the pre-trained DNN model can be decomposed into two functions: the first part
Φ
⋅
maps input xi into an intermediate layer
Φ
(
x
i
)
, and the second part
h
⋅
maps the intermediate layer
Φ
(
x
i
)
to the output
h
(
Φ
(
x
i
)
)
, which is a probability vector for each class, and
h
y
(
Φ
(
x
i
)
)
, is the probability of data x being predicted as label y by the model f.”), and
a classification vector neuron layer that receives a vector input from a last layer of the plurality of vector neuron layers and outputs a classification result of the input data (Yeh, page 3, section 3, ¶1 “We assume that the pre-trained DNN model can be decomposed into two functions: the first part
Φ
⋅
maps input xi into an intermediate layer
Φ
(
x
i
)
, and the second part
h
⋅
maps the intermediate layer
Φ
(
x
i
)
to the output
h
(
Φ
(
x
i
)
)
, which is a probability vector for each class, and
h
y
(
Φ
(
x
i
)
)
, is the probability of data x being predicted as label y by the model f.” here label y can be considered the classification result of the input data (x)),
the method performed by one or more processors (Yeh, page 1, footnote 1 “The code is released at https://github.com/chihkuanyeh/concept_exp.” Here, the use of computer code inherently teaches a processor) and comprising:
(a1) obtain from a non-transitory computer readable medium for each class of one or more classes discriminable by the machine learning model (Yeh, pages 7-8, section 5.2, ¶1 “For each class, the top concepts based on the conceptSHAP are the most important concepts to classify this class, as shown in Fig.3.”), a known feature … group obtained based on an output of a specific layer (Yeh, page 3, section 3, ¶1 “To choose the intermediate layer to apply concepts, we follow previous works on concept explanations Kim et al. [2018], Ghorbani et al. [2019] by starting from the layer closest to the prediction until we reached a layer that user is happy with…” here, the intermediate layer can be considered the specific layer) among the plurality of vector neuron layers arranged between the convolutional layer and the classification vector neuron layer of the machine learning model (Yeh, page 6, section 5.1, ¶1 “We take the last convolution layer as the feature layer
Φ
x
.” Here the feature layer can be considered a classification layer) when a plurality of pieces of training data are input to the convolutional layer of the machine learning model (Yeh, page 5, ¶4 “Intuitively, we require that the top-K nearest neighbor training input patches of each concept to be sufficiently close to the concept, and different concepts are as different as possible.”);
(b) execute a class discrimination processing of the object by inputting … the object as the input data to the convolutional layer of the machine learning model (Yeh, page 7, section 5.2, ¶1 “We use the Inception-V3 model, pre-trained on Imagenet [Szegedy et al., 2016], which yields 0.9 test accuracy.” Here, the use of the CNN model Inception-V3 can be considered the executing a class discrimination processing),
wherein:
the (b) includes:
(b0) obtain a class discrimination result, which is an output from the classification vector neuron layer (Yeh, page 6, section 5.1, ¶1 “We take the last convolution layer as the feature layer
Φ
x
.” Here the feature layer can be considered a classification layer) upon inputting the … data of the object to the convolutional layer of the machine learning model, the class discrimination result indicating a class among the one or more classes determined for the object (Yeh, page 3, section 3, ¶1 “We assume that the pre-trained DNN model can be decomposed into two functions: the first part
Φ
⋅
maps input xi into an intermediate layer
Φ
(
x
i
)
, and the second part
h
⋅
maps the intermediate layer
Φ
(
x
i
)
to the output
h
(
Φ
(
x
i
)
)
, which is a probability vector for each class, and
h
y
(
Φ
(
x
i
)
)
, is the probability of data x being predicted as label y by the model f.” here label y can be considered the a class among the one or more classes determined for the object),
(b1) calculate a feature … of a vector output (Yeh, page 3, ¶3 “Based on this motivation, we define the concept product for part of data xt as
PNG
media_image1.png
28
338
media_image1.png
Greyscale
, where TH is a threshold which trims value less than β to 0.”) that is output from the specific layer among the plurality of vector neuron layers arranged between the convolutional layer and the classification vector neuron layer of the machine learning model when the … data of the object is input to the convolutional layer of the machine learning model Yeh, page 6, section 5.1, ¶1 “We take the last convolution layer as the feature layer
Φ
x
.”);
(b3) create …text for the class…(Yeh, pages 8-9, section 5.3, ¶2 “For each concept, Table 2 shows (a) the top nearest neighbors based on the dot product of the concept and part of reviews (b) the most frequent words in the top-500 nearest neighbors (excluding stop words) (c) the conceptSHAP score for each concept.”) that is output from the specific layer, which is neither the convolutional layer that receives the spectral data nor the classification vector neuron layer that outputs the class discrimination result (Yeh, page 6, section 5.1, ¶1 “We take the last convolution layer as the feature layer
Φ
x
.” Here the feature layer is neither the convolutional layer that receives the spectral data nor the classification vector neuron layer that outputs the class discrimination result),…
Yeh does not teach “…using a spectrometer…;
…spectrum group;
(a2) measuring the object with the spectrometer and obtaining spectral data of the object from the spectrometer;
…spectral data…
…spectral data…
…calculating a feature spectrum;
…spectral data…
(b2) calculating a similarity between the feature spectrum and the known feature spectrum group;
…based on the calculated similarity between the feature spectrum and the known feature spectrum;”
However, Nakakimura teaches
…using a spectrometer… (Nakakimura, ¶3 “The present invention is preferably used to process three-dimensional spectral data obtained by, for example, a Liquid Chromatograph Mass Spectrometer (LC-MS), a Gas Chromatograph Mass Spectrometer (GC-MS), a liquid chromatograph using a multichannel type detector such as, e.g., a photodiode array (PDA) detector, a liquid chromatograph or a gas chromatograph using an ultraviolet-visible spectrophotometer or an infrared spectrophotometer capable of wavelength scanning as a detector, or an imaging mass spectrometer, etc.”)
…spectrum group (Nakakimura, ¶20 “a) a characteristic spectrum acquisition unit configured to perform multivariate analysis by considering a plurality of spectrums constituting a single three-dimensional spectral data obtained from a specific sample among a plurality of samples as a collection of a single spectrum not depending on a value of the second parameter, and based on a result of the multivariate analysis, one or a plurality of characteristic spectrums that characterize the specific sample is obtained;”)
(a2) measure the object with the spectrometer and obtaining spectral data of the object from the spectrometer (Nakakimura, ¶3 “The present invention is preferably used to process three-dimensional spectral data obtained by, for example, a Liquid Chromatograph Mass Spectrometer (LC-MS), a Gas Chromatograph Mass Spectrometer (GC-MS), a liquid chromatograph using a multichannel type detector such as, e.g., a photodiode array (PDA) detector, a liquid chromatograph or a gas chromatograph using an ultraviolet-visible spectrophotometer or an infrared spectrophotometer capable of wavelength scanning as a detector, or an imaging mass spectrometer, etc.”) and
…spectral data… (Nakakimura, ¶20 “a) a characteristic spectrum acquisition unit configured to perform multivariate analysis by considering a plurality of spectrums constituting a single three-dimensional spectral data obtained from a specific sample among a plurality of samples as a collection of a single spectrum not depending on a value of the second parameter, and based on a result of the multivariate analysis, one or a plurality of characteristic spectrums that characterize the specific sample is obtained;”)
…spectral data… (Nakakimura, ¶20 “a) a characteristic spectrum acquisition unit configured to perform multivariate analysis by considering a plurality of spectrums constituting a single three-dimensional spectral data obtained from a specific sample among a plurality of samples as a collection of a single spectrum not depending on a value of the second parameter, and based on a result of the multivariate analysis, one or a plurality of characteristic spectrums that characterize the specific sample is obtained;”)
…calculating a feature spectrum (Nakakimura, ¶30 “Normally, since a plurality of characteristic spectrums are obtained, the spectrum similarity calculation unit calculates the similarity between each spectrum at each measurement time extracted from the three-dimensional spectral data to a single sample and a characteristic spectrum for each of three-dimensional spectral data with respect to a plurality of samples for each characteristic spectrum.”)
…spectral data… (Nakakimura, ¶20 “a) a characteristic spectrum acquisition unit configured to perform multivariate analysis by considering a plurality of spectrums constituting a single three-dimensional spectral data obtained from a specific sample among a plurality of samples as a collection of a single spectrum not depending on a value of the second parameter, and based on a result of the multivariate analysis, one or a plurality of characteristic spectrums that characterize the specific sample is obtained;”)
(b2) calculating a similarity between the feature spectrum and the known feature spectrum group (Nakakimura, ¶30 “Normally, since a plurality of characteristic spectrums are obtained, the spectrum similarity calculation unit calculates the similarity between each spectrum at each measurement time extracted from the three-dimensional spectral data to a single sample and a characteristic spectrum for each of three-dimensional spectral data with respect to a plurality of samples for each characteristic spectrum.”)
…based on the calculated similarity between the feature spectrum…and the known feature spectrum (Nakakimura, ¶30 “Normally, since a plurality of characteristic spectrums are obtained, the spectrum similarity calculation unit calculates the similarity between each spectrum at each measurement time extracted from the three-dimensional spectral data to a single sample and a characteristic spectrum for each of three-dimensional spectral data with respect to a plurality of samples for each characteristic spectrum.”);
Yeh and Nakakimura are analogous art because both references concern methods for data processing. Accordingly, it would have been obvious to a person of ordinary skill in the art, before the effective filing date of the claimed invention, to modify Yeh’s image analysis method to incorporate the spectrums taught by Nakakimura. The motivation for doing so would have been to improve the throughput of the whole difference analysis, as stated in Nakakimura, ¶38 “Further, in the three-dimensional spectral data processing device and processing method according to the present invention, since a second parameter such as a retention time, etc., is not taken into account when determining the characteristic spectrum, no alignment processing is required for aligning the retention time among a plurality of samples which are normally required when obtaining a two-dimensional characteristic data table including characteristic data for a plurality of samples, and the time and effort required for such processing can be saved. As a result, the throughput of the whole difference analysis can be improved.”
Yeh in view of Nakakimura does not teach “creating an explanatory text…
(b4) output, for display on a user device, the class discrimination result with the explanatory text indicating a reason for the class determined for the object.”
However, CONTEXT teaches creating an explanatory text… (CONTEXT, page 16, section 3.2 “Prediction file (text output) Optionally, predict writes prediction values in the text format, one data point per line.” The prediction values in text format can be considered an explanatory text)…
(b4) output, for display on a user device, the class discrimination result (CONTEXT, page 16, section 3.2 “Prediction file (text output) Optionally, predict writes prediction values in the text format, one data point per line.” The text output of CONTEXT can be considered outputting the explanatory text and as a data file will be read on a user device) with the explanatory text indicating a reason for the class determined for the object (CONTEXT, page 16, section 3.2 “reNet predictapplies a model saved during training to new data and write prediction values to a file.” Here, the prediction values can be considered a reason).
Yeh in view of Nakakimura and CONTEXT are analogous art because both references concern classification using convolution neural networks. Accordingly, it would have been obvious to a person of ordinary skill in the art, before the effective filing date of the claimed invention, to have combined the text output of CONTEXT to the teachings of Yeh in view of Nakakimura. The motivation for doing so would have been to have a structured method of outputting the discrimination result, as stated in CONTEXT, col 22, lines 41-48, “The application server provides access control services in cooperation with the data store and is able to generate content such as text, graphics, audio and/or video to be transferred to the user, which may be served to the user by the Web server in the form of HTML, XML or another appropriate structured language in this example.”.
Regarding claim 10:
Yeh in view of Nakakimura in view of CONTEXT teaches [t]he method according to claim 1, wherein the at least one component is at least one preset wavelength band of the spectral reflectance data (Nakakimura, ¶3 “ Further, in a liquid chromatograph using a PDA detector as a detector, it is possible to obtain an absorption spectrum indicating a relationship between a wave number, a wavelength, etc., and a signal intensity (absorbance) from moment to moment.”).
It would have been obvious to combine the teachings of Yeh in view of Nakakimura in view of CONTEXT for the reasons set forth in connection with claim 1 above.
Allowable Subject Matter
Claim 2 is objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
The following is a statement of reasons for the indication of allowable subject matter:
After detailed search, the cited arts, neither alone nor in combination, teach the claimed subject matter of claim 2, “wherein the specific layer has a configuration in which vector neurons arranged on a plane defined by two axes including a first axis and a second axis are arranged as a plurality of channels along a third axis that is in a direction different from those of the two axes,
in the specific layer, when a region which is specified by a plane position defined by a position in the first axis and a position in the second axis and which includes the plurality of channels along the third axis is referred to as a partial region, for each partial region of a plurality of partial regions included in the specific layer, the feature spectrum is obtained as any one of:
(i) a feature spectrum of a first type in which a plurality of element values of an output vector of each of the vector neurons included in the partial region are arranged over the plurality of channels along the third axis;
(ii) a feature spectrum of a second type obtained by multiplying each of the element values of the feature spectrum of the first type by a normalization coefficient corresponding to a vector length of the output vector, and
(iii) a feature spectrum of a third type in which the normalization coefficient is arranged over the plurality of channels along the third axis.”
Pertinent art Ma et al. (“Fine-Grained Vehicle Classification With Channel Max Pooling Modified CNNs”, Ma et al., 4 April 2019) discloses a partial region is on each plane which is defined along a first and second axis and is repeated along a third axis but does not specifically disclose the claimed subject matter of claim 2.
Claims 3-6 are objected to as being dependent upon a rejected base claim.
Response to Arguments
Applicant's arguments filed June 5th, 2026, have been fully considered but they are not persuasive.
Applicant’s arguments regarding the 35 U.S.C. § 112(b) rejections of the previous office action have been fully considered, and are persuasive. The rejections have been withdrawn due to claim amendments. However, the amendments have required additional indefiniteness rejections to be made in this action.
Regarding the rejection of claims under 35 U.S.C. § 103, Applicant's arguments have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
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 JACOB Z SUSSMAN MOSS whose telephone number is (571) 272-1579. The examiner can normally be reached Monday - Friday, 9 a.m. - 5 p.m. ET.
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.
/J.S.M./Examiner, Art Unit 2122
/KAKALI CHAKI/Supervisory Patent Examiner, Art Unit 2122
Prosecution Insights