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 .
Claim Status
Claims 1-10 are pending.
Priority
The instant application has the effective filing date of 29 January 2022, with no claims to prior applications.
Information Disclosure Statement
No information disclosure statements (IDS) have been filed in the instant application. Applicants are reminded of their duty to disclose all information known to them to be material to the patentability as defined in 37 CFR 1.56.
Drawings
The drawings, submitted on 01/29/2022, are accepted by the examiner.
Claim Objections
Claim 9 is objected to because of the following informalities: “hyderparameters” likely is a typographical error, intended to recite “hyperparameters”. Appropriate correction is required.
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 8 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 for the following reasons.
Claim 8 recites that “b1 and b are bias added for the hidden layer and output layer, respectively”, wherein there is no “b” variable without an accompanying subscript. To overcome this rejection, please amend to “b1 and b2 are bias…”, as the claim is instantly interpreted, or provide further clarification for the variables.
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 1-10 are rejected under U.S.C 101 because the claimed invention is directed to abstract ideas without significantly more, as detailed in the analysis below.
Eligibility Step 1: Subject matter eligibility evaluation in accordance with MPEP § 2106:
Claims 1-10 are directed to a statutory category (method).
Therefore, in accordance with MPEP § 2106.03, all claims have patent eligible subject matter.
[Eligibility Step 1: YES]
Eligibility Step 2A: This step determines whether a claim is directed to a judicial exception in accordance with MPEP § 2106.
Eligibility Step 2A -- Prong One: Limitations are analyzed to determine if the claims recite any concepts that could equate to a judicial exception (i.e. abstract idea, law of nature, or natural phenomenon).
Recitations of Judicial Exceptions:
Claim 1: A method comprising
(2) establishing a model for predicting biodegradability of organic molecules in sewage through machine learning; (mathematical concept)
(4) predicting, according to the model established in (2), the biodegradability of the organic molecules in the sewage from the target sewage plant. (mathematical concept)
Claim 3:
a) calculating a molecular parameter of the organic molecules, and performing data standardization by using the molecular parameter as a feature value; (mathematical concept)
(b) calculating a Pearson correlation coefficient between the feature value and the biodegradability of the organic molecules; extracting, according to the absolute value of the Pearson correlation coefficient, desired feature values as input features in a neural network; (mathematical concept)
(c) splitting a dataset into a training set and a test set, determining topology of the neural network, which comprises a number of hidden layers and a number of neurons in each hidden layer; (mental process)
(d) optimizing hyperparameters of the model, training the neural network with the training set, and evaluating the performance of the neural network by using the test set. (mathematical concept, mental process)
Claim 4: The method of claim 3 wherein in (a):
The method of claim 3, wherein in (a), the molecular parameter as the feature value comprises: molecular parameters of all organic molecules, and molecular parameters of seven classes of organic molecules; the molecular parameters of all organic molecules comprise: a mass-to-charge ratio m/z, a number C of carbon atoms, a number H of hydrogen atoms, a number O of oxygen atoms, a number N of nitrogen atoms, a ratio O/C of the number of oxygen atoms to the number of carbon atoms, a ratio H/C of the number of hydrogen atoms to the number of carbon atoms, a number DBE of double bond equivalents, a ratio DBE/H of the number of double bond equivalents to the number of hydrogen atoms, a ratio DBE/O of the number of double bond equivalents to the number of oxygen atoms, a ratio (DBE-0)/C of a difference between the number of double bond equivalents and the number of oxygen atoms to the number of carbon atoms, an average value of a nominal oxidation state of carbon (NOSC) of all organic molecules, and strength weighted average values of molecular parameters, which are equal to a sum of products of respectively multiplying corresponding relative peak strength of molecules by m/z, C, H, O, N, O/C, H/C, DBE, DBE/H, DBE/O, (DBE-0)/C and NOSC; the seven classes of organic molecules are: lipids, proteins/amino sugars, carbohydrates, unsaturated hydrocarbons, lignin, tannins and condensed aromatics; screening conditions for lipids are as follows: O/C < 0.2 and 1.7 < H/C < 2.2;
screening conditions for proteins/amino sugars are as follows: 0.2 <O/C < 0.6, 1.5 < H/C < 2.2 and N/C > 0.05; screening conditions for carbohydrates are as follows: 0.6 <O/C < 1.0 and 1.5 < H/C < 2.2; screening conditions for unsaturated hydrocarbons are as follows: O/C<0.1, 0.7<H/C<1.5; screening conditions for lignin are as follows: 0.1 <O/C < 0.6, 0.6 < H/C < 1.7, and a modified aromaticity index Almod < 0.67; screening conditions for tannins are as follows: 0.6 <O/C < 1.0, 0.5 < H/C < 1.5 and a modified aromaticity index Almod < 0.67;
and, screening conditions for condensed aromatics are as follows: O/C < 1.0, 0.3 < H/C < 0.7 and the modified aromaticity index Almod >0.67; and the molecular parameters of seven classes of organic molecules comprise: a mass-to-charge ratio m/zi, a number DBEi of double bond equivalents, and an average value of the nominal oxidation state of carbon NOSCI of seven classes of organic molecules, a proportion Numi of the number of molecules in each class, and strength weighted average values of molecular parameters, which are equal to a sum of products of respectively multiplying corresponding relative peak strength of molecules by m/zi, DBEi and NOSCI, which i represents the molecule classes. (mathematical concept)
The limitations of claim 4 serve as additional information within the calculation step of claim 3 (a), which falls under the mathematical concepts grouping of abstract ideas.
Claim 5: The method of claim 3, wherein in (a), the data standardization is performed on the feature value using the formula:
z
=
(
x
-
u
)
s
where z is a standardized feature value, x is an original feature value, u is an average value of the features, and s is a standard deviation of the feature value. (mathematical concept)
Claim 6: The method of claim 3, wherein in b), the Pearson correlation coefficient between the feature value and the biodegradability of the organic molecules is calculated using the formula:
r
=
∑
i
=
1
n
(
x
i
-
x
)
(
y
i
-
y
)
∑
i
=
1
n
(
x
i
-
x
)
2
∑
i
=
1
n
(
y
i
-
y
)
2
where xi is a feature value, yj is a measured value of the biodegradability of the organic molecules in the sewage, x=n1xi,y=nLyi, n is a total number of sewage samples; a correlation matrix of the feature value and the biodegradability of the organic molecules in the sewage is obtained using the formula; and according to the absolute value of the Pearson correlation coefficient, the feature value highly correlated with the biodegradability of the organic molecules in the sewage is selected as input values of the neural network. (mathematical concept)
Claim 7: The method of claim 3, wherein in (c),
the dataset is randomly split into a training set and a test set with a 7: 3 ratio;
a model for biodegradability forecasting is established based on the multi-layer perceptron;
the input value of the input layer is connected to the neurons in the hidden layer and the neurons in the hidden layer are connected to the neurons of an output layer;
each neuron in one layer is connected to all neurons in a next layer; the topology of the multi-layer perceptron is determined as follows: a range for the numbers of the neurons in each hidden layer is determined according to the number of input variables;
a range for the number of the hidden layers is determined according to the characteristics of data structure, and the same number of neurons is used for all hidden layers. (mental process)
The limitations of claim 7 serve as additional information within the limitation of claim 3 (c) that recites “splitting” and “determining” data, which fall under the mental process grouping of abstract ideas.
Claim 8: The method of claim 3, wherein in (d), the topology of the multi-layer perceptron comprises m input neurons, n hidden neurons, and one output neuron; an output of the neural network, a predicted value for biodegradability of sewage, is expressed within the following equation:
PNG
media_image1.png
72
357
media_image1.png
Greyscale
where, f is a predicted value, Wand w are weights of a hidden layer and an input layer, respectively; bi and b are bias added for the hidden layer and the output layer, respectively; and theta is a an activation function; training the neural network is to minimize a loss function; the loss function is expressed as below:
PNG
media_image2.png
77
336
media_image2.png
Greyscale
where, alpha/2 II W II is a L2 regularization term for penalizing a complex model; parameters in an opposite direction of a gradient of an objective function are updated at each iteration through gradient descent; an example formula of gradient descent is as follows where the weights are updated:
PNG
media_image3.png
89
212
media_image3.png
Greyscale
where, i is a number of iterations, /E (0,1) is a learning rate, and VLoss' is a gradient of the loss function with respect to the weights. (mathematical concept)
Claim 9: The method of claim 8, wherein the hyperparameters that need to be optimized comprises:
an algorithm used to minimize the loss function, comprising stochastic gradient descent (SGD), adaptive moment estimation (Adam) and Limited-memory BFGS (L-BFGS); (mathematical concept)
an activation function, comprising Sigmoid, tanh and ReLU; (mathematical concept)
a parameter a for L2 regularization term; (mathematical concept)
and a maximum number iter of iterations; (mathematical concept, mental process)
the training set is used to fit the neural network, and the test set is used to measure the performance of the neural work; (mathematical concept)
a coefficient of determination (R2) and root mean squared error (RMSE) are used as indicators for evaluating the accuracy of the model; and R2 is calculated using the formula:
R
2
y
,
y
=
1
-
∑
i
=
1
n
(
y
i
-
y
i
)
2
∑
i
=
1
n
(
y
i
-
y
i
)
2
RMSE is calculated using the formula:
RMSE =
1
2
∑
i
=
1
n
(
y
i
-
y
i
)
2
where, yi is a measured value, yi is a predicted value,ny=-Ly1, n is a total number of the sewage sample. (mathematical concept)
The limitations of this claim recite mathematical calculations and provide additional information regarding the limitations of claim 8 that recite formulas, thus falling under the mathematical concepts grouping of abstract ideas.
Claim 10: The method of claim 1, wherein in (3) and (4), using the model to predict the biodegradability of the sewage comprises:
(b) extracting a desired feature value; and performing data normalization on the feature value; (mental process, mathematical concept)
Step 2A – Prong One Analysis:
The recitation of mathematical formulas and functions such as regularization, standardization, coefficient of determination, and RMSE represent transformations of data via mathematical calculations. Limitations of this nature fall under the mathematical concepts grouping of abstract ideas.
Techniques such as extracting data from a dataset, splitting data, simple calculations, and making determinations from data analysis that require no more than the human mind and pen/paper, fall under the mental process grouping of abstract ideas.
Eligibility Step 2A – Prong Two: A claim that integrates a judicial exception into a practical application will apply, rely on, or use the judicial exception in a manner that imposes a meaningful limit on the judicial exception. If the claim contains no additional claim elements beyond the abstract idea, the claim fails to integrate the abstract idea into a practical application (MPEP 2106.04(d)).
Eligibility Step 2B: Claim elements are probed for inventive concept equating to significantly more than the judicial exception (MPEP 2106.04(II)).
Additional Elements within the claimed invention include:
Claim 1: A method comprising
(2)establishing a model for predicting biodegradability of organic molecules in sewage through machine learning;
(4) predicting, according to the model established in (2), the biodegradability of the organic molecules in the sewage from the target sewage plant.
Claim 2: The method of claim 1, wherein
the biodegradability of the organic molecules in the sewage is represented by BOD5/COD
These limitations equate to mere data gathering activities necessary to obtain data for downstream analysis and output them. As such, they are categorized as insignificant extra-solution activity and do not integrate the judicial exceptions into practical application per MPEP 2106.05(g).
[Eligibility Step 2A – Prong Two: YES]
Furthermore, predicting biodegradability parameters such as BOD5 and COD via machine learning is well-understood, routine and conventional per Hur et al. (Sensors; Vol. 10; p. 2460-2471; 2010), Saleh (Journal of Ecological Engineering; Vol. 22(7); p. 35-45; 2021), and Abyaneh (Journal of Environmental Health Science & Engineering; Vol. 12:40; 2014) which each independently perform the modelling.
[Eligibility Step 2B: NO]
Additional Elements that may be categorized differently include:
Claim 1: A method comprising
(1) collecting molecular composition information and biodegradability data of organic molecules in a sewage sample;
(3) measuring the molecular composition information of organic molecules in sewage from a target sewage plant; and
Claim 2: The method of claim 1, wherein
(1), the molecular composition information of organic molecules in the sewage sample comes from data measured by a Fourier transform ion cyclotron resonance mass spectrometer,
Claim 10: The method of claim 1, wherein in (3) and (4), using the model to predict the biodegradability of the sewage comprises:
(c) feeding the feature value in (b) into the model, running the model to obtain an output value for the biodegradability of organic molecules in sewage.
(a) measuring the molecular composition information of organic molecules in the sewage sample by a Fourier transform ion cyclotron resonance mass spectrometer;
These limitations equate to mere data gathering activities necessary to obtain data for downstream analysis and output them. As such, they are categorized as insignificant extra-solution activity and do not integrate the judicial exceptions into practical application per MPEP 2106.05(g).
[Eligibility Step 2A – Prong Two: YES]
The necessary data gathering and output step, recited at such high level of generality (claim 10, limitation (c)), is well-understood, routine and conventional technique per Mayo, 566 U.S. at 79, 101 USPQ2d at 1968; OIP Techs., Inc. v. Amazon.com, Inc., 788 F.3d 1359, 1363, 115 USPQ2d 1090, 1092-93 (Fed. Cir. 2015). Measuring molecular composition of with a Fourier Transform Ion Cyclotron Resonance Mass Spectrometer is also deemed well-understood, routine, and conventional within the art per Marshall et al. (Mass Spectrometry Reviews; Vol. 17; 1998), which reviews the technique.
[Eligibility Step 2B: NO]
Additional Elements that may be categorized differently include:
Claim 3: The method of claim 1, wherein in (2), the model for predicting the biodegradability of organic molecules in the sewage is established by a multi-layer perceptron that is a neural network model used in machine learning.
The multi-layer perceptron neural network, when viewed separately, and in the context of the whole claimed invention, merely acts as a tool to perform the abstract ideas of predicting biodegradability by training a machine learning model with an analyzed set of data (mental processes) and optimization algorithms (mathematical concepts). As such, the element does not integrate the judicial exceptions into practical application.
[Eligibility Step 2A – Prong Two: YES]
Furthermore, neural networks for predicting biodegradability are found well-understood, routine, and conventional within the art per Saleh (Journal of Ecological Engineering; Vol. 22(7); p. 35-45; 2021), Abyaneh (Journal of Environmental Health Science & Engineering; Vol. 12:40; 2014), and Abba et al. (Procedia Computer Science; Vol. 120; 9th Internation Conference; p. 156-163; 2018), which each independently perform neural network modelling of biodegradability parameters.
[Eligibility Step 2B: NO]
As such, claims 1-10 are directed to judicial exceptions without significantly more and are rejected under 35 U.S.C 101.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claim 1 is rejected under 35 U.S.C. 103 as being unpatentable over Hur et al. (Sensors; Vol. 10; 2010).
Claim 1 is directed to a method that collects data pertaining the molecular composition and biodegradability of organic molecules in a sewage sample; establishes a model; measures molecular composition of organic molecules from a particular sewage plant; and uses the model to predict the biodegradability of organic molecules in sewage from the particular sewage plant.
Hur et al. describes a method of estimating biological oxygen demand (BOD) and chemical oxygen demand (COD) for combined sewer systems.
Hur et al. teaches collecting wastewater samples from four different sewer pipes; conducting BOD and COD experiments (page 2462, column 1) to determine biodegradability (page 2463, column 1), and measuring synchronous fluorescence spectra on the unfiltered samples (page 2462, column 1) to gain information on the composition of organic matters (page 2461, column 1).
Hur et al. further teaches establishing the correlation between measured BOD, COD, and predicted values by a multiple regression model (page 2468, column 1).
Hur et al. does not teach that the sewage comes from a particular sewage plant.
However, Hur et al. does teach that the high time resolution about water quality data for combined sewer overflows is also important for the proper evaluation of the influence of storm water flows on wastewater treatment plants (page 2461, column 1). As such, Hur et al. teaches techniques that predict the biodegradability of sewage, a species of wastewater, and wastewater plants. Therefore, the method of Hur et al. is also applicable to sewage samples coming from particular sewage plants.
Claims 2 and 10 are rejected under 35 U.S.C. 103 as being unpatentable over Hur et al. (Sensors; Vol. 10, 2460-2471; 2010), as applied to claim 1 above, in view of Lu et al. (PLOS One; Vol. 10 (12); 2015).
Claim 2 is directed to molecular composition data being measured by a Fourier Transform Ion Cyclotron Resonance Mass Spectrometer, and biodegradability data including BOD5/COD data. Claim 10 is directed to measuring molecular composition via Fourier transform ion cyclotron resonance mass spectrometer; extracting a desired feature value, normalizing it; providing the value to the model, and using it to obtain an output value pertaining the biodegradability of organic molecules in sewage.
Hur et al. teaches that at least five days are required to acquire BOD data from the experiment (page 2461, column 1). Therefore, the BOD/COD ratio in Hur et al. is BOD5/COD.
Hur et al. further teaches that measured fluorescence intensities were normalized (page 2462, column 1); the intensity of one peak in the PLF region (Index I) and another peak in the PLF region area (Index II) were selected as the potential estimation indices for BOD, as PLF is reported to be associated with microbial activities and/or biodegradable organic matter(page 2466, column 1); and multiple regression method based on Index I and SS concentrations was employed to compensate for the SS contribution and to improve the estimation capability for BOD and COD (page 2468, column 1).
Therefore Hur et al. teaches measuring molecular composition via fluorescence spectra; extracting and normalizing feature values; and using the data to predict BOD5/COD, which estimates the biodegradability of sewage samples.
Hur et al. does not teach measuring molecular composition via Fourier transform ion cyclotron resonance mass spectrometer (FT-ICR-MS).
However, Hur et al. does teach that there are limitations in using fluorescence spectra for estimating BOD and COD in sewage (page 2469, column 1).
Lu et al. describes use of FT-ICR-MS to characterize dissolved organic matter (DOM) in headwater streams.
Lu et al. teaches that the application of ESI-FTICR-MS technique offers additional insights into compound composition and reactivity unrevealed by fluorescence and stable carbon isotopic measurements (page 2, column 1); and that a molecular formula calculator developed at the FTICR-MS Facility at the National High Magnetic Field Laboratory of Florida State University (Molecular Formula Calc v.1.0 NHMFL, 1998) was used to generate empirical formulas (page 5, column 1) of the matter.
Lu et al. further teaches analyzing DOM with FT-ICR-MS during 15-day biodegradation experiments (page 1, column 1) to assess how DOM composition influences the biodegradability (page 2, column 1).
Therefore Lu et al. provides one of ordinary skill in the art sufficient teachings, suggestions, and motivation to substitute fluorescence spectra with FT-ICR-MS for measuring parameters regarding the molecular composition of dissolved organic matter within sewage, in order to gain more information about its biodegradability.
Claims 3, 5, and 6 are rejected under 35 U.S.C. 103 as being unpatentable over Hur et al. (Sensors; Vol. 10; 2010) as applied to claim 1 previously, in view of Saleh (Journal of Ecological Engineering; Vol. 22 (7); 2021) and Hou et al. (Environmental International; Vol. 135; 2020).
Hur et al. teaches a biodegradability prediction model for sewage samples.
Claim 3 is directed to the establishing a multi-layer perceptron neural network by performing the following steps:
Calculating and standardizing a molecular parameter of organic molecules as a feature;
Calculating the absolute value of a Pearson correlation coefficient between the organic molecule parameter feature and biodegradability data to get input features;
Splitting the dataset into a training and testing set and determining a number of hidden layers and neurons within each layer; and
Optimizing hyperparameters of the model; training the neural network with the training set; and evaluating the neural network performance with test set.
Hur et al. does not teach that the model is a multi-layer perceptron neural network; with the parameters described above.
Saleh describes an artificial neural network wastewater pollution prediction model.
Saleh teaches that the simplest and most widely employed neural network architectures are the multilayered perceptron artificial neural networks (page 37, column 1).
Regarding limitation 3a, Saleh teaches standardizing the input and output data in the range 0-1 (page 38, column 1). The input data includes molecular parameters of organic molecules such as Total nitrogen, NO2 , NO3 , NH3 and PO4 (page 26, column 2).
Regarding limitation 3b, Saleh teaches that the Pearson coefficient of correlation (r) was used for evaluating the optimal model configuration for the outcomes of the model that meet the minimum errors values (page 39, column 1).
Regarding claim 3c, Saleh teaches allocating 70% of the data to training and 15% each to validation and test sets (page 38, column 2); selecting the number of hidden layers based on task difficulty, and determining the number of neurons in the layer based on a trial-and-error approach (page 37, column 1).
Regarding claims 3d, Saleh teaches optimizing properties of the ANN model (page 29, column 1) such a transfer functions and iteration counts (page 29, column 2); using the training data to modify weights connecting the neurons (page 38, column 2); and utilizing the test data to evaluate its optimality and generalization capabilities (page 38, column 2).
Though Saleh teaches calculating Pearson correlation coefficient between the feature value and biodegradability values, Saleh does not teach using the absolute value of the correlation coefficient (r) in order to derive input values for the neural network (limitation 3b).
Hou et al. describes estimating ecotoxicity characterization factors for chemicals in life cycle assessment using machine learning models.
Hou et al. teaches using the absolute values of the calculated Pearson correlation coefficient between variables to show the correlation between any two variables, including input and output variables (page 6, column 2).
Hou et al. further teaches that Life cycle assessment is a widely used analytical tool that examines how the environmental impacts of products along its whole life cycle, are associated with the usage and release of chemicals such as those from waste treatment plants (page 1, column 1). Therefore, it is applicable to modelling the task at hand.
Hou et al. further teaches that the exploratory analysis of some of the 13 physical-chemical property input variables (page 6, column 1), including biodegradation half-life (page 3, column 1), are highly skewed before transformation and filtering (page 6, column 1). As such, one of ordinary skill in the art has sufficient motivation to use the techniques taught by Hou et al. to model ecological waste treatment variables in a manner that would result in predictable improvements to the machine learning system.
Claim 5 is directed to standardizing data by subtracting the average value for the features from the original feature value and dividing by the standard deviation of the feature value from the mean.
Saleh teaches adjusting the data around its mean value, according to the standard deviation (page 38, column 1), as follows:
x
i
=
x
i
-
μ
σ
where: µ and σ are the mean and standard deviation values of the data (page 38, column 2).
Claim 6 is directed to calculating the Pearson correlation coefficient via the standard equation.
Saleh teaches the Pearson coefficient of correlation (r) calculated according to the following formula (page 39, column 1):
PNG
media_image4.png
75
255
media_image4.png
Greyscale
Therefore, Saleh provides an artificial neural network that predicts a parameter of biodegradation in wastewater treatment plants. As such, the known techniques within can be combined or substituted by one of ordinary skill in the art building an analogous model with a reasonable expectation of success.
Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Hur et al. (Sensors; Vol. 10; p. 2460-2471; 2010) in view of Saleh (Journal of Ecological Engineering; Vol. 22(7); p. 35-45; 2021) and Hou et al. (Environmental International; Vol. 135; 2020), as applied to claims 3, 5 and 6 above, and in further view of Boehmke (University of Cincinnati; Analytics; 2018).
Claim 7 is directed to splitting the training and testing data with a 7:3 ratio; establishing the multi-layer perceptron in the order input value, hidden layer of neurons, output value; where each neuron in one layer is connected to all the neurons in the next layer; the number of neurons in each hidden layer is determined by the number of input variables; each hidden layer has the same number of neurons; and the range of hidden layers is determined based on data structure characteristics.
Hur et al. in view of Saleh and Hou et al. teach a multilayer perceptron ANN for predicting biodegradability.
Hou et al. further teaches splitting the data into 70% for training and validation, and 30% for testing (page 3, column 2).
Saleh further teaches that within the multilayered perceptron ANN (page 37, column 1) that forecasts the quality of treated effluent (page 40, column 12), the structure of the developed neural networks consists of one input layer, variable number of hidden layers and one output layer (page 39, column 2), wherein the neurons of the hidden layer connect the layers (page 37, fig. 1); and every neuron is bound to all the neurons in the next layer (page 36, column 2). Saleh further teaches that one, two, and three hidden layers with 5 and 6 neurons in each layer, respectively, were used (page 39, column 2), which establishes that each hidden layer had the same number of neurons.
Saleh further teaches that the number of neurons in the layer is determined based on trial-and-error approach, beginning with the lowest value and gradually increasing, according to the nature of the problem (page 37, column 1); and the size of the ANN is determined by the number of hidden layers in the ANN (page 36; column 2).
Saleh and Hou et al. do not explicitly teach that the number of hidden neurons is determined according the number of input variables or the range of hidden layers is determined according to the data structure.
Boehmke is a programming guide regarding feedforward deep learning models.
Boehmke teaches that typically 2-5 hidden layers is sufficient for regular rectangular data, for example normal data frames in R (page 9, column 1); and that the number of nodes you incorporate in these hidden layers of a multilayer perceptron is largely determined by the number of features in your data, and that often the number of nodes in each layer is equal to or less than the number of features (page 9, column 1).
Therefore, Saleh teaches that the number of hidden neurons and layers varies according to the task; and Boehmke provides sufficient motivation for one of ordinary skill in the art to use the number of input features and data structure as a general guide to the determining the number of hidden neurons and layers of a feedforward ANN.
Claim 8 is rejected under 35 U.S.C. 103 as being unpatentable over Hur et al. (Sensors; Vol. 10; p. 2460-2471; 2010) in view of Saleh (Journal of Ecological Engineering; Vol. 22(7); 2021) and Hou et al. (Environmental International; Vol. 135; 2020), as applied to claims 3, 5, and 6 above, and in further view of Razavi et al. (IEEE Transactions on Neural Networks; Vol. 22 (10); 2011) and Roughgarden et al. (CS168; Lecture 6; 2016).
Claim 8 is directed to the MLP ANN including m input neurons, n hidden neurons, and at least one output neuron. The neural network output is the predicted value for biodegradability of sewage, expressed within the equation:
PNG
media_image1.png
72
357
media_image1.png
Greyscale
where, f is a predicted value, Wand w are weights of a hidden layer and an input layer, respectively; bi and b2 are bias added for the hidden layer and the output layer, respectively; and theta is an activation function.
The claim further trains the neural network is to minimize loss, according to the equation:
PNG
media_image2.png
77
336
media_image2.png
Greyscale
where, alpha/2 II W II22 is a L2 regularization term for penalizing a complex model; and parameters in an opposite direction of a gradient of an objective function are updated at each iteration through gradient descent.
Saleh teaches that within a feed forward back propagation ANN (page 36, column 1), the weighted summation of all inputs is computed as the first step in processing the elements, according to the equation:
PNG
media_image5.png
87
283
media_image5.png
Greyscale
where W is the weight factor, x in the input (page 36, column 2), and f is the activation function (page 37, fig. 1).
Saleh does not teach including a variable representing bias.
Razavi et al. describes a formulation of feedforward neural networks.
Razavi et al. teaches expressing single-hidden layer neural networks through the following equations (page 1588, column 1):
PNG
media_image6.png
141
412
media_image6.png
Greyscale
where R and n are the numbers of input and hidden nodes, respectively; Iw and Hw are the input and hidden weights matrices, respectively; Hb is the bias vector of the hidden layer and Ob is the bias value of the output layer; x is the input vector of the network; Ho is the output vector of the hidden layer, and y is the network output (page 1588, column 1).
Therefore Razavi et al. teaches a back propagation algorithm that generates output (y) as a function of the summed, weighted input (wijxj) and bias of the hidden (Hb) and outer layers (Ob). Razavi teaches further multiplying the result by an activation function (f). As the equations are also applicable to a feedforward artificial neural network, one of ordinary skill in the art could reasonably apply this function to generate a prediction value of the claimed ANN, with a reasonable expectation of success.
Razavi et al. does not teach including a L2 regularization penalty in the loss equation.
Roughgarden et al. is an excerpt from a computer science lecture regarding model overfitting and loss.
Roughgarden et al. teaches that one can reduce overfitting by calculating a mean-squared error that shows the relationship between input(x) and prediction (y) (page 9, column 1) and using regularization, a concrete method for implementing the principle of Occam’s razor which adds a “penalty term” to the optimization problem, such that more complex models incur a larger penalty (page 10, column 1) with the equation:
MSE (w) + penalty (w)
where penalty(w) is increasing with the “complexity” of w (page 10, column 1).
Roughgarden et al. teaches that this ensures a complex solution will be chosen over a simple solution, only if it leads to a large decrease in the mean-squared error (page 10, column 1).
Roughgarden et al. further teaches that although there are many ways to define the penalty term, the most widely used is L2 regularization, represented as:
PNG
media_image7.png
37
181
media_image7.png
Greyscale
where λ is a positive “hyperparameter,” a knob that allows you to trade-off smaller MSE versus smaller model complexity (page 10, column 1).
Roughgarden et al. further teaches the gradient moves in the opposite direction of its algorithm (page 3, column 1); and that for many machine learning problems, replacing the basic gradient descent method by stochastic gradient descent is crucial for accommodating large data sets (page 12, column 1), is the dominant paradigm in modern machine learning and deep learning work (page 12 column 1), and adding regularization imposes no extra computational demands on it (page 12, column 1).
Therefore Roughgarden et al. provides sufficient motivation for one of ordinary skill in the art to apply a stochastic gradient descent algorithm to minimize loss with a L2 regularization penalty term.
Claim 9 is rejected under 35 U.S.C. 103 as being unpatentable over Hur et al. (Sensors; Vol. 10; p. 2460-2471; 2010) and Saleh (Journal of Ecological Engineering; Vol. 22(7); 2021) in view of Razavi et al. (IEEE Transactions on Neural Networks; Vol. 22 (10); 2011) and Roughgarden et al. (CS168; Lecture 6; 2016) as applied to claim 8 above, and in further view of Hou et al. (Environmental International; Vol. 135; 2020).
Razavi et al. in view of Roughgarden et al. teach a feed forward backpropagated neural network that minimizes loss through many parameters, including stochastic gradient descent (SGD) and a L2 regularization penalty.
Razavi et al. further teaches using sigmoid, as the activation function (page 1588, column 1).
Roughgarden et al. further teaches that another stopping rule to prevent overfitting is to run a gradient descent model for a fixed number of iterations (page 2, column 1).
Razavi et al. and Roughgarden et al. do not explicitly teach using an Adam, optimizer, nor measuring neural network performance with coefficient of determination (
R
2
) and root mean squared error (RSME).
Hou et al. describes modelling ecotoxicity factors via machine learning as, described previously.
Hou et al. further teaches using ReLu as an activation function (page 6, column 2) and optimizing neural network models with rmsprop, adam, sgd (stochastic gradient descent), adagrad, adadelta, adamax, and nadam algorithms (page 5, table 1).
Hou et al. further teaches using the test set to evaluate the predictive performance of the selected model; specifically, coefficient of determination (
R
2
) and root mean squared error (RMSE) as model evaluation metrics (page 4, column 2).
Hou et al. teaches
R
2
is calculated using the equation (page 4, column 1):
PNG
media_image8.png
137
366
media_image8.png
Greyscale
Hou et al. teaches RMSE is calculated using the equation (page 4, column 1):
PNG
media_image9.png
120
392
media_image9.png
Greyscale
Hou et al. further teaches evaluating and choosing these parameters based on trial and error (page 4, column 2), similar to Saleh. Therefore, Hou et al. provides a model that has made improvements according to other known techniques taught by the prior art. As such, there is sufficient motivation for one of ordinary skill in the art to include further neural network training and validation components within an analogous system and field of use.
Claims 4 is rejected under 35 U.S.C. 103 as being unpatentable over Hur et al. (Sensors; Vol. 10; 2010) in view of Saleh (Journal of Ecological Engineering; Vol. 22 (7); 2021) and Hou et al. (Enviornmental International; Vol. 135; 2020), as applied to claims 1 and 3 previously, and in further view of Lu et al. (PLOS One; Vol. 10 (12); 2015.).
Claim 4 is directed to the of organic molecules parameters including one of the following: a mass-to-charge ratio m/z, a number C of carbon atoms, a number H of hydrogen atoms, a number of O of oxygen atoms, a number N of nitrogen atoms, a ratio O/C of the number of oxygen atoms to the number of carbon atoms, a ratio H/C of the number of hydrogen atoms to the number of carbon atoms, a number DBE of double bond equivalents, a ratio DBE/H of the number of double bond equivalents to the number of hydrogen atoms, a ratio DBE/O of the number of double bond equivalents to the number of oxygen atoms, a ratio (DBE-0)/C of a difference between the number of double bond equivalents and the number of oxygen atoms to the number of carbon atoms, an average value of a nominal oxidation state of carbon (NOSC) of all organic molecules, and strength weighted average values of molecular parameters, which are equal to a sum of products of respectively multiplying corresponding relative peak strength of molecules by m/z, C, H, O, N, O/C, H/C, DBE, DBE/H, DBE/O, (DBE-0)/C or NOSC;
Lu et al. teaches that a molecular formula calculator (page 5, column 1) considers m/z values, the number of different atoms Carbon, Hydrogen, Oxygen, and Nitrogen atoms, the number of double bond equivalents (DBE) (page 5, column 1), and H/C and O/C atomic ratios (page 6, column 1) as molecular parameters.
Lu et al. further teaches that the molecular parameters can be classified into six regions: 1) lipid-like region (H/C = 1.5–2.0, O/C = 0–0.3); 2) protein-like region (H/C = 1.5–2.2, O/ C=0.3–0.67); 3) lignin-like region (H/C = 0.7–1.5, O/C = 0.1–0.67); 4) carbohydrate-like region (H/C = 1.5–2.4, O/C = 0.67–1.2); 5) unsaturated hydrocarbon-like region (H/C = 0.7 1.5, O/C = 0–0.1); and 6) condensed aromatic ring structure (CAS) region (H/C = 0.2–0.7,O/ C=0–0.67) (page 6, column 1).
Lu et al. further teaches analyzing DOM with FT-ICR-MS during 15-day biodegradation experiments (page 1, column 1) to assess how DOM composition influences the biodegradability (page 2, column 1).
Therefore Lu et al. provides one of ordinary skill in the art sufficient teachings, suggestions, and motivation to measure these molecular parameters of the dissolved organic matter in sewage, in order to thoroughly model and asses its impact on biodegradability.
Conclusion
No claims are currently allowed.
Correspondence
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Milana Thompson whose telephone number is (571)272-8740. The examiner can normally be reached Monday - Friday, 9:00-6:00 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, Karlheinz Skowronek can be reached at (571) 272-1113. 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.
/M.K.T./Examiner, Art Unit 1687
/Karlheinz R. Skowronek/Supervisory Patent Examiner, Art Unit 1687