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 .
DETAILED ACTION
This action is in response to the amendments filed 17 December 2025. Claim 8 is canceled. Claims 1-4, 6-7, and 10 are amended. Claims 1-7, 9, and 10 are pending and have been examined.
Priority
Acknowledgment is made of applicant's claim for foreign priority based on an application filed in Germany on 13 January 2022. It is noted, however, that applicant has not filed a certified copy of the DE 10 2022 200 287.3 application as required by 37 CFR 1.55.
Response to Arguments
Applicant' s arguments, see pages 6-7, filed 17 December 2025, with respect to the rejections of Claim 7 under 35 U.S.C. 112(a) and 35 U.S.C. 112(b) have been fully considered and are persuasive. The rejections of Claim 7 under 35 U.S.C. 112(a) and 35 U.S.C. 112(b) have been withdrawn.
APPLICANT'S ARGUMENT: Applicant argues (page 6, paragraph 1) that "Applicants have amended claim 7 to more concretely recite that a computer having a processor that carries out the method. ¶ The terms 'computer' and ''processor' should not be interpreted to invoke 35 U.S.C. §112(f)."
Applicant argues (page 7, paragraph 1) that "a person having ordinary skill in the art would understand it to be clearly implied or inherent that training a deep neural network must be performed by a computing device having a processor and memory. Thus, the specification provides implied or inherent support computing device having a processor and memory."
Applicant argues (page 7, paragraph 2) that "A person having ordinary skill in the art would understand terms 'computer' and 'processor' to have sufficiently definite meanings as the names for the structures that perform the recited functions. Accordingly, the rejections on §112(b) grounds should also be withdrawn."
EXAMINER'S RESPONSE: Examiner agrees. The rejections of Claim 7 have been withdrawn in light of arguments and/or amendments.
Applicant's arguments, see pages 7-10, filed 17 December 2025, with respect to the rejections of Claims 1-10 under 35 U.S.C. 101 have been fully considered but they are not persuasive.
APPLICANT'S ARGUMENT: Applicant argues (page 9, paragraph 2) that "The claimed invention provides a solution to this problem in the technical field of training a neural network to process sensor values. ... [T]he claimed invention advantageously includes an initialization of the weighting matrices and bias vectors of the individual layers of the neural network (par. 0013). ... As a result, the training method is able to converge faster (par. 0013; 0025)."
Applicant argues (page 10, paragraph 2) that "even if the processes for determining the distribution interval and initializing model parameters are understood to be abstract (as alleged by the Office), these features are integrated in a practical application by the non-abstract process of using training a neural network to generate regression outputs or classification outputs based on sensor values."
Applicant argues (page 10, paragraph 3) that "the claimed invention provides a clear improvement to a technology or to a technical field because the claimed invention solves a technological problem."
EXAMINER'S RESPONSE: Examiner respectfully disagrees. The claims of the invention do not currently appear to reflect the purported improvement in neural network training. While the specification describes "faster" ([0013]) "rapid" ([0025]) convergence of training, the recited steps of amended Claim 1 do not appear to recite or reflect the purported improvement. As indicated below with respect to the Alice/Mayo framework, the additional elements recited by the claims (i.e., providing training data, sensor network data, and the step of training the neural network) appear to be incidental to the invention rather than providing a practical application or significantly more.
In particular, as currently recited, the sensor values of the claim do not appear to restrict or limit the step of processing values by a neural network during training in a manner that amounts to more than generally linking the neural network values to the particular field of sensor networks.
Applicant's arguments, see pages 11-13, filed 17 December 2025, with respect to the rejections of Claims 1-10 under 35 U.S.C. 102(a)(1) have been fully 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.
APPLICANT'S ARGUMENT: Applicant argues (page 12, paragraph 1) that "Khellal does not disclose 'determining a distribution interval of the respective sensor values of all of the training data sets' and 'initializing model parameters of the neural network as a function of the distribution interval.' Particularly, Khellal discloses that, instead of reconstructing the input, the ELM-CNN algorithm reconstructs the normalized input (§2.2). Therefore, first, the data matrix is normalized to have O mean, and 1 as standard deviation referred as XN (§2.2). Second, to handle the convolution bias, the ELM-CNN algorithm adds the intercept term. ... ¶ However, as best understood, Khellal fails to disclose determining a distribution interval of the input data sets (e.g., of the MNIST dataset), nor does Khellal disclose initializing model parameters depending on such a distribution interval. Instead, Khellal clearly discloses that the ELM algorithm, the input weights are randomly generated according to a normal distribution (§3.1)."
Applicant argues (page 13, paragraph 1) that "Claims 2-10 depend from and incorporate the limitations of claim 1. Therefore, for at least the reasons presented above with respect to claim 1, Khellal fails to anticipate the limitations of the claims."
EXAMINER'S RESPONSE: Examiner notes that Applicant's arguments are moot. Amended Claim 1 is now rejected under 35 U.S.C. 103 as obvious in view of Khellal in view of Wu. In the 35 U.S.C. 103 rejection below, Khellal is relied on to teach features pertaining to training a neural network by determining a distribution interval of input training data values, initializing the neural network, and generating classification outputs based on network input values. Wu is relied on to teach features related to training a neural network to process values from a sensor in a sensor system. Amended Claims 2-7, 9, and 10 are now rejected in view of the Khellal/Wu combination as indicated below.
Claim Objections
Claim 1 is objected to because of the following informalities. Claim 1 includes a grammatical error in the limitation: "a sensor of in a system." For the purposes of examination, the limitation has been interpreted to read: "sensor values from a sensor in a system" (emphasis added). Appropriate correction is required.
Claim Interpretation
Claim 7 is not being interpreted under 35 U.S.C. 112(f) in light of arguments and/or amendments.
Claim Rejections - 35 USC § 112(a)
The rejection of Claim 7 under 35 U.S.C. 112(a) is withdrawn in light of arguments and/or amendments.
Claim Rejections - 35 USC § 112(b)
The rejection of Claim 7 under 35 U.S.C. 112(b) is withdrawn in light of arguments and/or amendments.
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.
The rejection of Claim 8 under 35 U.S.C. 101 for non-statutory subject matter is withdrawn in light of arguments and/or amendments.
Claims 1-7, 9, and 10 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding Claim 1
Step 1
Claim 1 recites a method for training a neural network to process sensor values from a sensor of in a system, and thus the claimed process falls within a statutory category of invention.
Step 2A Prong 1
The claim recites determining a distribution interval of the respective ... values of all of the training data sets, which is a mental process. The claim recites initializing model parameters of the neural network as a function of the distribution interval, which is a mental process. The claim recites to generate regression outputs or classification outputs based on ... values, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2
The additional element providing training data sets, each training data set including respective ... values and one or more labels assigned to the respective ... values amounts to insignificant extra-solution activity (see MPEP 2106.05(g), "mere data gathering"). The additional element sensor values does not amount to more than generally linking the use of a judicial exception to a particular field of use (see MPEP 2106.05(h), "limit the use of the abstract idea to a particular technological environment"). The additional element training the neural network ... using the training data sets by further adaptation of the model parameters simply recites a judicial exception with the words "apply it" (see MPEP 2106.05(f)(1), "the claim recites only the idea of a solution or outcome, i.e., the claim fails to recite details of how a solution to a problem is accomplished," as the limitations do not provide meaningful restriction with respect to how it is accomplished or provide steps by which an improvement in technology may be achieved).
Step 2B
The additional element providing training data sets, each training data set including respective ... values and one or more labels assigned to the respective ... values is well-understood, routine, conventional activity (see MPEP 2106.05(d), "receiving or transmitting data over a network"). The additional element sensor values does not amount to more than generally linking the use of a judicial exception to a particular field of use (see MPEP 2106.05(h), "limit the use of the abstract idea to a particular technological environment"). The additional element training the neural network ... using the training data sets by further adaptation of the model parameters simply recites a judicial exception with the words "apply it" (see MPEP 2106.05(f)(1), "the claim recites only the idea of a solution or outcome, i.e., the claim fails to recite details of how a solution to a problem is accomplished," as the limitations do not provide meaningful restriction with respect to how it is accomplished or provide steps by which an improvement in technology may be achieved).
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 2
Step 1
Regarding Claim 2, the rejection of Claim 1 is incorporated.
Step 2A Prong 1
The claim recites the determination of the distribution interval is determined as a function of a minimum value and a maximum value of all of the respective ... values of the training data sets, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The additional element sensor values does not amount to more than generally linking the use of a judicial exception to a particular field of use (see MPEP 2106.05(h), "limit the use of the abstract idea to a particular technological environment").
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 3
Step 1
Regarding Claim 3, the rejection of Claim 1 is incorporated.
Step 2A Prong 1
The claim recites the model parameters for each layer of artificial neurons of the neural network are provided as elements of a weighting matrix and of a bias vector, which is a mental process and mathematical concept.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 4
Step 1
Regarding Claim 4, the rejection of Claim 3 is incorporated.
Step 2A Prong 1
The claim recites wherein, initializing the model parameters of the neural network comprises: determining a transformation function for mapping an assumed normalized input data set with a predetermined normalized distribution onto a distribution of values of elements of the input data sets, which is a mental process. The claim recites specifying preliminary model parameters, which is a mental process. The claim recites applying the transformation function to the preliminary model parameters in order to obtain transformed model parameters, which is a mathematical concept. The claim recites initializing the neural network with the transformed model parameters, which is a mathematical concept.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 5
Step 1
Regarding Claim 5, the rejection of Claim 4 is incorporated.
Step 2A Prong 1
The claim recites wherein the predetermined normalized distribution of the input data sets has a distribution interval between -1 and 1, which is a mental process and a mathematical concept.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 6
Step 1
Regarding Claim 6, the rejection of Claim 4 is incorporated.
Step 2A Prong 1
The claim recites applying the transformation function to the preliminary model parameters in order to obtain transformed model parameters (as recited by Claim 4), wherein neuron functions of a layer of the layers of artificial neurons are subjected to an inverted version of the transformation function to obtain the transformed model parameters, which is a mathematical concept and mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 7
Step 1
Claim 7 recites a computer comprising: a processor configured to carry out the method according to claim 1, and thus the claimed machine falls within a statutory category of invention.
Step 2A Prong 1
Claim 7 recites the abstract ideas recited by Claim 1.
Step 2A Prong 2, Step 2B
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 9
Step 1
Claim 9 recites a non-transitory machine-readable storage medium comprising instructions, and thus the claimed manufacture falls within a statutory category of invention.
Step 2A Prong 1
Claim 7 recites the abstract ideas recited by Claim 1.
Step 2A Prong 2, Step 2B
The additional element instructions which, when executed by a computer, cause the computer to execute the method according to claim 1 invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it").
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 10
Step 1
Regarding Claim 10, the rejection of Claim 2 is incorporated.
Step 2A Prong 1
The claim recites wherein the distribution interval is determined as a function of an average value and a standard deviation of the values of all of the respective ... values of the training data sets, which is a mental process and mathematical concept.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The additional element sensor values does not amount to more than generally linking the use of a judicial exception to a particular field of use (see MPEP 2106.05(h), "limit the use of the abstract idea to a particular technological environment").
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
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.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
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.
Claims 1-7, 9, and 10 are rejected under 35 U.S.C. 103 as being unpatentable over Khellal, et al., "Convolutional Neural Network Features Comparison Between Back-Propagation and Extreme Learning Machine" (hereinafter "Khellal") in view of Wu, et al. (US 2018/0357542 A1, hereinafter "Wu").
Regarding Claim 1, Khellal teaches:
A method for training a neural network to process ... values (Khellal, p. 9629, Abstract: "The developed framework (ELM-CNN) is based on the concept of autoencoding to learn the convolutional filters with biases, by reconstructing the normalized input and the intercept term. ... The experimental results on the popular MNIST dataset show that the ELM-CNN algorithm achieves competitive results in terms of generalization performance") ... of in a system (Khellal, p. 9631, 3.1 Experimental settings, Simulation environment: "We used MATLAB 2016a running on a desktop computer with Intel i7 CPU and 32GB of memory (RAM). As for the training of CNN based on backpropagation, we employed the MatConvNet library"), the method comprising:
providing training data sets, each training data set including respective ... values and one or more labels assigned to the respective ... values (Khellal, p. 9631, 3.1 Experimental settings: "MNIST dataset Is a publicly available standard classification benchmark, composed of a set of
28
×
28
gray-scale images of handwritten digits. It contains a training set of 60,000 examples and a test set of 10,000 examples. Each sample is labeled with one correct digit class
{
0
,
1
,
…
,
9
}
");
determining a distribution interval of the respective ... values of all of the training data sets (Khellal, p. 9631, 2.2 ELM-CNN, Filters learning: "instead of reconstructing the input, we reconstruct the normalized input. Therefore, first, the data matrix is normalized to have 0 mean, and 1 as standard deviation referred as
X
N
" where Khellal's normalized input corresponds to the instant training set interval), and p. 9631, Algorithm 2 Training procedure of CONV layer using ELM, line 3: "Generate normalized training data
X
N
");
initializing model parameters of the neural network as a function of the distribution interval (Khellal, p. 9631, 2.2 ELM-CNN, Filters learning, Algorithm 2 Training procedure of CONV layer using ELM, lines 3-7:
Generate normalized training data
X
N
Compute desired target
T
=
X
N
1
Generate randomly the input weights and biases
W
and
b
.
Compute the hidden matrix
H
=
G
X
W
+
b
Compute the output weights
β
=
H
†
T
Compute filters and bias
F
m
a
t
T
B
T
=
β
Reshape the filters matrix
F
=
r
e
s
h
a
p
e
F
m
a
t
.
Return CONV parameters
F
and
B
," where Khellal's CONV parameters
F
and
B
correspond to the instant model parameters, which are initialized based on the normalized training data); and
training the neural network to generate regression outputs or classification outputs based on ... values ...(Khellal, p. 9631, 2.2 ELM-CNN, Filters learning: "if we desire a classification model, we just need to replace the last layer (classification scores) with any classifier like SVM, ELM, and many others" and p. 9631, 3.2 Quantitative comparison, Methodology: "For a given model (ELM-CNN or BP-CNN), and features of interest (CONV 2, POOL 2,...), we compute the features codes of the whole dataset. Subsequently, we train an independent classifier using only the features codes of the training dataset, and then we report the performance on the test set").
Khellal teaches a method for training a neural network using training data sets to process values to generate classification outputs based on the values.
Khellal does not explicitly teach a neural network to process sensor values from a sensor of in a system, training data set including respective sensor values and one or more labels assigned to the respective sensor values, regression outputs or classification outputs based on sensor values, and using the training data sets by further adaptation of the model parameters.
However, Wu teaches
a neural network to process sensor values from a sensor of in a system (Wu, [0046]: "when the event object to be analyzed is ... sensing signals of distributed optical fiber sensors, the differences between the present invention and other CNN-based methods lie in that: ... the one-dimensional convolution neural network (1D-CNN) structure suitable for the time sequence structure of distributed optical fiber sensing signals is specially designed in the present invention, so that the computational complexity is reduced and the learning effect is better"),
training data set including respective sensor values and one or more labels assigned to the respective sensor values (Wu, [0083]: "A predicted label of training data raw train is obtained through the one-dimensional convolution neural network (1 D-CNN) with parameters setting and is compared with the true label of the sample to obtain a loss value, so that a gradient is calculated thereby to update a network parameter 8" and [0079]: "A ID-CNN deep learning network structure is constructed and trained based on the typical event dataset"),
regression outputs or classification outputs based on sensor values (Wu, [0095]: "The loss function is calculated according to the classification output obtained in the step (2), so as to continue to update, adjust and optimize the constructed CNN network" and [0079]: "A ID-CNN deep learning network structure is constructed and trained based on the typical event dataset"),
using the training data sets by further adaptation of the model parameters (Wu, [0098]: "2) Calculating the update gradient according to the cross entropy loss function C, updating the network weight obtained through training"and [0088]: "For convolution layers, the first convolution layer C1 is taken for instance. The number of convolution kernels in C1 is M, the size is assumed as m, and each convolution kernel needs K convolutions. One input training data is assumed as X ..., wherein X, belongs to raw train; the initialized weight matrix of the jth convolution kernel is assumed as W1 ... and the bias vector is Bias1").
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Khellal regarding a method for training a neural network using training data sets to process values to generate classification outputs based on the values with those of Wu regarding a neural network to process sensor values from a sensor of in a system, training data set including respective sensor values and one or more labels assigned to the respective sensor values, regression outputs or classification outputs based on sensor values, and using the training data sets by further adaptation of the model parameters.
The motivation to do so would be to facilitate training of more efficient and more accurate neural networks (Wu, [0046]: "the differences between the present invention and other CNN-based methods lie in that: ... the one-dimensional convolution neural network (1D-CNN) structure suitable for the time sequence structure of distributed optical fiber sensing signals is specially designed in the present invention, so that the computational complexity is reduced and the learning effect is better" and [0066]: "The method has the advantages of effectively improving the adaptive ability and the identification accuracy of the system in complex noise environment, updating the algorithm easily, omitting the time-consuming and arduous process of manual feature extraction and classification, having identification rate better than the classification results of manually extracted features, being conducive to the largescale application of optical fiber sensing and having a huge potential application value").
Regarding Claim 2, the rejection of Claim 1 is incorporated. The Khellal/Wu combination teaches:
the determination of the distribution interval is determined as a function of a minimum value and a maximum value of all of the respective sensor values of the training data sets (Khellal, p. 9631, 2.2 ELM-CNN, Filters learning: "instead of reconstructing the input, we reconstruct the normalized input. Therefore, first, the data matrix is normalized to have 0 mean, and 1 as standard deviation referred as
X
N
," where the distribution of Khellal's input data is determined as a function of a zero-mean normal distribution with minimum and maximum values of the first standard deviation being -1 and 1, respectively, similarly to [0016] of the instant specification).
Regarding Claim 3, the rejection of Claim 1 is incorporated. The Khellal/Wu combination teaches:
wherein the model parameters for each layer of artificial neurons of the neural network are provided as elements of a weighting matrix and of a bias vector (Khellal, p. 9631, 2.2 ELM-CNN, Filters learning, Algorithm 2 Training procedure of CONV layer using ELM, line 10, "Return CONV parameters
F
and
B
," where p. 9631, 2.2 ELM-CNN, Filters learning: "the CONV-layer parameters can be computed as
F
m
a
t
T
B
T
=
β
, where
B
is the bias vector, which is defined as the transpose of the last column of β. Finally, we just need to reshape the
F
m
a
t
matrix to obtain the filters F," where Khellal's
F
and
B
correspond to the instant weighting matrix and bias vector).
Regarding Claim 4, the rejection of Claim 3 is incorporated. The Khellal/Wu combination teaches:
wherein, initializing the model parameters of the neural network comprises: determining a transformation function for mapping an assumed normalized input data set with a predetermined normalized distribution onto a distribution of values of elements of the input data sets (Khellal, p. 9631, 2.2 ELM-CNN, Filters learning, Algorithm 2 Training procedure of CONV layer using ELM, where function
G
at line 6 of Khellal's algorithm corresponds to the instant transformation function, and the random input weights of line 5 correspond to the instant second distribution of values, as in Khallal, p. 9631, 3.1 Experimental settings, Training procedure: "In the ELM algorithm, additive sigmoid activation function is used, and the input weights are randomly generated according to the normal distribution");
specifying preliminary model parameters (Khellal, p. 9631, 2.2 ELM-CNN, Filters learning, Algorithm 2 Training procedure of CONV layer using ELM, line 5, "Generate randomly the input weights and biases
W
and
b
");
applying the transformation function to the preliminary model parameters (Khellal, p. 9631, 2.2 ELM-CNN, Filters learning, Algorithm 2 Training procedure of CONV layer using ELM, line 6: "Compute the hidden matrix
H
=
G
X
W
+
b
") in order to obtain transformed model parameters (Khellal, p. 9631, 2.2 ELM-CNN, Filters learning, Algorithm 2 Training procedure of CONV layer using ELM, line 10: "Return CONV parameters F and B"); and
initializing the neural network with the transformed model parameters (Khellal, p. 9631, 2.2 ELM-CNN, Filters learning, Algorithm 2 Training procedure of CONV layer using ELM, line 10: "Return CONV parameters F and B").
Regarding Claim 5, the rejection of Claim 4 is incorporated. The Khellal/Wu combination teaches:
wherein the predetermined normalized distribution of the input data sets has a distribution interval between -1 and 1 (Khellal, p. 9631, 2.2 ELM-CNN, Filters learning: "instead of reconstructing the input, we reconstruct the normalized input. Therefore, first, the data matrix is normalized to have 0 mean, and 1 as standard deviation referred as
X
N
," where Khellal's normalization to given mean and standard deviation corresponds to the instant determined interval, and where Khellal's data matrix is a preprocessed input).
Regarding Claim 6, the rejection of Claim 4 is incorporated. The Khellal/Wu combination teaches:
wherein neuron functions of a layer of the layers of artificial neurons are subjected to an inverted version of the transformation function to obtain the transformed model parameters (Khellal, p. 9631, 2.2 ELM-CNN, Filters learning, Algorithm 2 Training procedure of CONV layer using ELM, line 7: "Compute the output weights
β
=
H
†
T
," where Khellal's
H
†
corresponds to the instant inverse transformation, previously defined at p. 9630, 2.1 ELM-AE overview: "
H
†
stands for the Moore-Penrose generalized inverse of matrix
H
");
Regarding Claim 7, Khellal teaches:
A computer comprising: a processor configured to carry out the method according to claim 1 (Khellal, p. 9631, 3.1 Experimental settings: "To verify the proposed algorithm ELM-CNN, we carried out several experiments on the public dataset MNIST. ... ¶ Simulation environment We used MATLAB 2016a running on a desktop computer with Intel i7 CPU and 32GB of memory (RAM)"). Claim 7 is rejected under the same rationale as Claim 1.
Regarding Claim 9. Khellal teaches:
A non-transitory machine-readable storage medium comprising instructions which, when executed by a computer, cause the computer to execute the method according to claim 1 (Khellal, p. 9631, 3.1 Experimental settings: "To verify the proposed algorithm ELM-CNN, we carried out several experiments on the public dataset MNIST. ... ¶ Simulation environment. We used MATLAB 2016a running on a desktop computer with Intel i7 CPU and 32GB of memory (RAM)," where a non-transitory storage medium is inherent in verifying ELM-CNN using a dataset). Claim 9 is rejected under the same rationale as Claim 1.
Regarding Claim 10, the rejection of Claim 2 is incorporated. The Khellal/Wu combination teaches:
wherein the distribution interval is determined as a function of an average value and a standard deviation of the values of all of the respective sensor values of the training data sets (Khellal, p. 9631, 2.2 ELM-CNN, Filters learning: "instead of reconstructing the input, we reconstruct the normalized input. Therefore, first, the data matrix is normalized to have 0 mean, and 1 as standard deviation referred as
X
N
," where Khellal's normalization to given mean and standard deviation corresponds to the instant determined interval, and where Khellal's data matrix is a preprocessed input).
Conclusion
THIS ACTION IS MADE FINAL. 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 ROBERT N DAY whose telephone number is (703)756-1519. The examiner can normally be reached M-F 9-5.
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.
/R.N.D./Examiner, Art Unit 2122
/KAKALI CHAKI/Supervisory Patent Examiner, Art Unit 2122