DETAILED ACTION
This action is in response to the original filing on 08/18/2023. Claims 1-9 are pending for examination.
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 Objections
Claims 1, 5, 7, and 9 are objected to because of the following informalities:
Claim 1:
Lines 3, 5, 6, 7, 9 and 12, remove S1, S2, S3, S4, S5, S6.
Line 12, add “and” after the phrase “an optimized machine learning model;”
Claim 5: replace “step S4 specifically” with “integrating and training data based on a neural network to obtain multi-dimensional spatial proxy layer data sets and training points”.
Claim 7: remove “step S5”.
Claim 9: “replace “step S6 specifically” with “applying the optimized machine learning model to complete machine learning result evaluation and target area delineation”.
Appropriate corrections are required.
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 1-6 and 8-9 are rejected under 35 U.S.C. 103 as being unpatentable over YE et al. (CN 104866630 B, hereinafter Ye) in view of Sun et al. (‘GIS-based mineral prospectivity mapping using machine learning methods: A case study from Tongling ore district, eastern China’, hereinafter Sun).
Regarding Claim 1, Ye teaches a method for metallogenic prediction by using multi-source heterogeneous information (Paragraph [0018] This invention addresses the problem that existing mineralization prediction modeling relies primarily on manual analysis and human-computer interaction to extract relevant information for modeling, without achieving intelligent processing. It provides a GIS and ES automatic spatial modeling system and method for mineralization prediction.) comprising the following steps:
SI, collecting geological, geochemical and remote sensing multi-source geoscience information data; (Paragraph [0040] The comprehensive information prospecting model is based on geological information, studies geological, geophysical, geochemical and remote sensing information, and studies the conversion law between information. It uses indirect information to find hidden mineral resource bodies and blind mineral resource bodies to achieve the purpose of prospecting. Paragraph [0074] Figure 2 shows the construction of GIS spatial information and map libraries, where geoscientific map libraries are provided by users or created by the system through the geoscientific map library construction module of map library management; Figure 3 shows the construction of ES database and knowledge base. Paragraph [0075] Geoscience map libraries involve multi-disciplinary geoscience data maps from geophysical exploration, geochemical exploration, geology, and remote sensing. Except for remote sensing which is saved as image files, the sheer number of these graphic files and the vast amount of spatial nformation they contain are generally stored in a hierarchical, categorized directory structure. (A knowledge base is built containing information (collecting information) related to geophysical prospecting))
S2, building a conceptual model of a metallogenic system; (Paragraph [0076] Combining Figures 11 and 12, the GIS modeling process for mineralization prediction includes a conceptual model and a logical model; Paragraph [0077] Conceptual Model: Based on the characteristics of geoscientific multivariate data modeling in comprehensive information mineral prediction and resource evaluation, the modeling process of the comprehensive information metallogenic prediction conceptual model is summarized as follows: ① Establish a geoscientific map library; ② Study the ore-controlling factors, metallogenic regularities and prospecting indicators of typical deposits; ③ Establish a comprehensive information prospecting model; ④ Establish a comprehensive information prediction model; ⑤ Locate and quantify the prediction. Among them, the geoscience map library refers to the map system established by geology, geophysics, geochemistry, remote sensing and other fields. (Building a conceptual model based on the geological information gathered prior))
S3, extracting geoscience multi-source spatial proxy mineralization indication information according to the conceptual model of the metallogenic system; (Paragraph [0284] For example, under the above conditions, establish the mineralization information integration process function "Mineral Exploration Space Library\Geology\Silver Anomaly wt"; using the graphic files in the eoscientific map library as the condition part and the process function as the conclusion part, automatically establish a knowledge rule i: Paragraph [0285] If (Geological Map Library\Geology\Silver Mine.wt) and (Geological Map Library\Geochemical Exploration\Ag.wt) Paragraph [0286] Then (Mineral exploration space reservoir\Geology\Silver anomaly wt); Paragraph [0287] If the spatial information integration result file "Exploration Space Library\Geology\Silver Anomaly.wt" already exists in the ES database, then proceed to the next search step to verify whether a spatial information integration result file matching "Exploration Space Library\Geology\Silver Anomaly.wt" exists in the ES database. If not, then establish the process function for the next rule to complete the establishment of the next knowledge rule. This process continues until all spatial modeling knowledge rules are established based on the initial ES database (Extraction of mineralization data based on a conceptual model))
Ye does not teach S4, integrating and training data based on a neural network to obtain multi-dimensional spatial proxy layer data sets and training points; S5, inputting the multi-dimensional spatial proxy layer data sets and the training points, and applying a machine learning algorithm for hyper-parameter optimization to obtain an optimized machine learning model; S6, applying the optimized machine learning model to complete machine learning result evaluation and target area delineation.
In the same field of endeavor, Sun teaches S4, integrating and training data based on a neural network to obtain multi-dimensional spatial proxy layer data sets and training points; (Section 4.1 Pg. 3.2 In some complicated cases that the origin feature spaces are not linearly separable (Fig. 3b), the input data have to be transformed into a higher n-dimensional feature space through various kernel functions, where a dichotomy classification can be performed by a (n-1) dimensional plane referred to as hyperplane. Section 5.1.1. Pg. 33 The generation of predictor maps was guided by the understanding of the skarn Cu mineral system (Table 2) and was based on publicly available datasets. In this study, various sources of exploration data, including geological, geophysical, geochemical, and remote sensing data related to Cu mineralization, were utilized as spatial proxies to represent the source, transport, trap, and deposition processes critical for ore formation (Figs. 6–9). (Obtaining proxies of n-dimensional geological data and then generating a predictive map based on these proxies))
S5, inputting the multi-dimensional spatial proxy layer data sets and the training points, and applying a machine learning algorithm for hyper-parameter optimization to obtain an optimized machine learning model; (Section 5.2 Pg. 39-40 Determining the parameters used for training models is the first step in the predictive modelling and is a key procedure to yield reliable predictions. However, it is difficult to specify a priori suitable configuration for making predictions with a desired precision, because there are no universal rules for determining the optimal parameters in a real-world application. Although some empirical terms may be helpful in this process, a highly subjective trial-and-error procedure is invariably needed to obtain an optimal configuration of parameters. In this study, a 10-fold cross-validation procedure was performed to assess the predictions derived from possible combinations of parameters. The input training dataset was randomly subdivided into 10 subsets with equal size, in which a single subset was retained as the validation dataset and the other 9 subsets were used for training models. This process was repeated 10 times until each subset had been used once as the validation dataset (Xiong and Zuo, 2017). The Mean Square Error (MSE) was employed to evaluate the results of cross-validation, which can be formulized as: [Formula 13] where Nv is the number of data in validation dataset; 𝑦𝑖 denotes the predicted class value of target data (i.e., 1 for deposit and 0 for non-deposit); and yi denotes the true class value of target data. The model configuration with the lowest MSE was determined as the best one. (Obtaining an optimized model configuration through usage of the input proxy parameters))
S6, applying the optimized machine learning model to complete machine learning result evaluation and target area delineation. (Section 6.3 Pg. 42 The ANN model achieves a 100% positive predict value, indicating that 100% of predicted “deposit” cells are true deposit locations. The RF model also yields a very high positive predictive rate (98.33%), with only one non-deposit sample incorrectly predicted as a deposit. The SVM model has a lower positive predictive value of 94.34% and also yields the lowest negative predictive value (82.19%), indicating that 82.19% of cells classified as “non-deposit” correspond to true non-deposit locations. The RF and ANN models have higher negative predictive values, which are 93.94% and 91.30%, respectively. RF yields the leading accuracy of 96.03%, which represents the highest correctly-classified rate of both deposit and non-deposit locations, compared to ANN and SVM with accuracies of 95.24% and 87.30%, respectively. (The optimized model is applied to a target area))
It would have been obvious to one having ordinary skill in the art before the filing date of the claimed invention to have combined the steps of integrating training data based on proxy data sets, inputting the proxy data to obtain an optimized model, and applying to optimized model to the target area as taught by Sun into Ye as both references are in the same field of predicting where metallogenic ore-bearing deposits are beneath the ground, and doing so would be desirable in order to easier detect these complex structures, as their formation and deposition is controlled by a variety of factors too complex to be addressed by conventional prospecting methods (Sun Section 1, Pg 26)
Regarding Claim 2, the combination of Ye and Sun teaches all of the limitations of Claim 1 above, including: wherein the geological, geochemical and remote sensing multi-source geoscience information data comprises geological data of a working area, (Ye Paragraph [0040] The comprehensive information prospecting model is based on geological information, studies geological, geophysical, geochemical and remote sensing information, and studies the conversion law between information. It uses indirect information to find hidden mineral resource bodies and blind mineral resource bodies to achieve the purpose of prospecting. Paragraph [0089] Geography (geographic base map of the survey area, topographic contour lines or digital elevation map) (The collection of geological data from a working area, including dispersal/spatial patterns of geological ores (halo patterns) and other geochemical factors.))
geochemical data of a primary halo or a secondary halo, (Ye Paragraph [0052] Provided that at least three graphic files are provided—a geological map (geological map.wp), a predicted ore deposit map (e.g., silver ore.wt), and a main ore-forming element anomaly map (e.g., Ag.wt)—the system can automatically build a database, perform reasoning, and complete a series of very complex automatic spatial modeling processes for ore-forming prediction, ultimately establishing a comprehensive information prediction model. Paragraph [0065] Figure 10 is a schematic diagram of the spatial distribution of geochemical anomalies in the predicted spatial reservoir geological unit; (Informational data comprising the distribution of ore anomalies within a given area. A halo in geology is a term meaning the dispersion of ores, and is also characterized by the term "anomalies", acting as a measure of either enrichment (positive) or depletion (negative) of ores in a given area))
multispectral or hyperspectral remote sensing image data, geophysical data, and genetic data of a known deposit. (Ye Paragraph [0020] The mineral exploration spatial database, including mineral exploration spatial information and map library, refers to the extraction, integration, and establishment of a comprehensive mineral exploration spatial information and map library that integrates geological, geophysical, and geochemical spatial information and map library with mineral maps, geochemical and heavy mineral anomalies to conduct spatial analysis of metallogenic regularities and ore-controlling factors. Paragraph [0040] The comprehensive information prospecting model is based on geological information, studies geological, geophysical, geochemical and remote sensing information, and studies the conversion law between information. It uses indirect information to find hidden mineral resource bodies and blind mineral resource bodies to achieve the purpose of prospecting. (Remote sensing is a term well known in the art and already comprises multi/hyper-spectral imaging when used in visible/infrared spectrums.))
Regarding Claim 3, the combination of Ye and Sun teaches all of the limitations of Claim 1 above, including: wherein the conceptual model of the metallogenic system comprises: a dynamic mechanism of a deposit of a working area, (Ye Paragraph [0052] Provided that at least three graphic files are provided—a geological map (geological map.wp), a predicted ore deposit map (e.g., silver ore.wt), and a main ore-forming element anomaly map (e.g., Ag.wt)—the system can automatically build a database, perform reasoning, and complete a series of very complex automatic spatial modeling processes for ore-forming prediction, ultimately establishing a comprehensive information prediction model. Paragraph [0077] Conceptual Model: Based on the characteristics of geoscientific multivariate data modeling in comprehensive information mineral prediction and resource evaluation, the modeling process of the comprehensive information metallogenic prediction conceptual model is summarized as follows: ① Establish a geoscientific map library; ② Study the ore-controlling factors, metallogenic regularities and prospecting indicators of typical deposits; ③ Establish a comprehensive information prospecting model; ④ Establish a comprehensive information prediction model; ⑤ Locate and quantify the prediction. Among them, the geoscience map library refers to the map system established by geology, geophysics, geochemistry, remote sensing and other fields. (The conceptual model comprises the ore-forming/ore-controlling factors, analogous to the dynamic mechanism of a deposit, as the dynamic mechanism is the label applied to the processes of ore formation and (natural) distribution))
sources of a metallogenic geological body and a metallogenic material, (Ye Paragraph [0077] Conceptual Model: Based on the characteristics of geoscientific multivariate data modeling in comprehensive information mineral prediction and resource evaluation, the modeling process of the comprehensive information metallogenic prediction conceptual model is summarized as follows: ① Establish a geoscientific map library; ② Study the ore-controlling factors, metallogenic regularities and prospecting indicators of typical deposits; (The model is based on sources of ore-controlling/metallogenic bodies))
ore-conducting and ore-bearing structures, (Ye Paragraph [0081] The mineral exploration spatial (information and map) library is a comprehensive database of geological, geophysical, and geochemical spatial information and maps that is extracted and integrated from geological maps, geochemical maps, and heavy mineral anomalies to analyze mineralization regularities and ore-controlling factors. (The model analyzes ore-controlling factors))
mineralization and denudation preservation, and an alteration type and distribution range of a surface caused by hydrothermal solution. (Sun Section 3 Pg. 29 The mineral systems approach provides a holistic view of the geological processes (including tectonic, physical, and chemical processes) critical for ore formation at a variety of scales, which commonly include (Wyborn et al., 1994; Kreuzer et al., 2008; Joly et al., 2012; Kreuzeret al., 2015; Hagemann et al., 2016): (i) generation of energy gradients to trigger mineralization events; (ii) extraction of necessary components required for ore formation (e.g., metals, fluids, and ligands) from mantle and/or crustal sources; (iii) migration of ore-forming materials through pathways that connect source regions and trap zones; (iv) focusing of metalliferous fluid and modification of fluid composition by physical and chemical processes in trap zones that lead to deposition of metals; and (v) preservation of deposits. Section 3.4 Pg. 31-32 Metal deposition in hydrothermal systems can result from a variety of processes that induce changes in physical and chemical conditions of fluids to reduce metal solubility, such as cooling of fluids, depressurization, fluid mixing, and fluid-rock reactions. In a skarn Cu mineralization system, interactions between metalliferous fluids and carbonate wall rocks play a major role in the metal deposition. Fluid boiling, which is commonly observed in the studies of fluid inclusions (Deng et al., 2011; Cao et al., 2017; Liu et al., 2019), is an important mechanism for mineral precipitation in the Tongling ore district. The fluid boiling may result from the abrupt pressure release and rapid temperature drop as metalliferous fluids were focused towards physical traps (i.e., multi-layered fracture zones). This process led to the loss of volatiles (e.g., CO2 and H2O) and an increase of pH, which in turn destabilized the metal complexes and promoted metal deposition (Liu et al., 2019). Geochemical anomalies and hydrothermal alteration are good indicators of chemical deposition. (The system offers a full analysis of geological processes critical to ore formation, including mineralization, distribution, and preservation. Furthermore, section 3.4 discusses alterations and distributions of minerals deposited through hydrothermic means))
Regarding Claim 4, the combination of Ye and Sun teaches all of the limitations of Claim 1 above, including: determining a grid size according to an area of a study area; (Sun Section 5.1.3. Pg. 36 Before implementation of prospectivity modelling, maps of evidential features should be transformed into raster maps in which each cell has a numerical representation of the evidential features. The cell size was objectively selected based on a methodology proposed by Carranza (2009). Firstly, a suitable cell size should ensure one deposit occurring in any cell. The range of suitable sizes can be inferred from the point pattern analysis (Fig. 10), which indicates the nearest neighboring distance of any two deposits is 378m, indicating that the cell in a grid with size larger than 378 m plausibly contains more than one deposit. Thus, the upper limit of the cell size is considered to be 378m. Secondly, the lower limit of cell size is considered based on the map scales of the spatial evidences. (Determination of a grid size of a chosen study area))
extracting a spatial gridding evidence layer of a metallogenic related geological body, a combined anomaly map and a comprehensive anomaly map, and (Sun Section 5.1.3 Pg. 36 Before implementation of prospectivity modelling, maps of evidential features should be transformed into raster maps in which each cell has a numerical representation of the evidential features. The cell size was objectively selected based on a methodology proposed by Carranza (2009). Firstly, a suitable cell size should ensure one deposit occurring in any cell. The range of suitable sizes can be inferred from the point pattern analysis (Fig. 10), which indicates the nearest neighboring distance of any two deposits is 378m, indicating that the cell in a grid with size larger than 378 m plausibly contains more than one deposit. Thus, the upper limit of the cell size is considered to be 378m. Secondly, the lower limit of cell size is considered based on the map scales of the spatial evidences. (Extracting an evidentiary grid of geologic features of subterranean copper deposits, including detected geochemical anomalies (Fig. 9)))
a distribution range of a hydroxyl-containing altered mineral combination according to a type of a target deposit; (Sun Section 3.4 Pg. 31-32 Geochemical anomalies and hydrothermal alteration are good indicators of chemical deposition. The univariate Cu anomalies and multivariate anomalies related to Cu mineralization extracted from assay results of stream sediments, as well as iron-oxide and argillic alteration interpreted from ETM + images were utilized as proxies for the deposition process (Detection of altered (via hydrothermic processes, which produce iron-oxide and hydroxyl bearing minerals) geochemical mineral distributions in a targeted (copper) metal.))
computing a geological connotation value of the evidence layer through an inverse distance weighted average interpolation method and re-classification; and (Sun Section 5.1.3 Pg. 36-37 The finest legible spatial resolution Rf can be estimated by (Hengl, 2006) R(f) = MS x 0.00025 where MS represents the map scale. In this study, the largest map scale is 1: 200,000, and thus the lower limit of cell size is considered to be 50m. Within the suitable size range, we used a cell size of 200m to generate raster maps containing 20,250 pixels. It should be noted that the original spatial data were sampled at a coarser resolution in various surveys, meaning that the resulting raster maps cannot ensure one sample/station/observation per cell. Spatial interpolation techniques, such as inverse distance weighted and Kriging, were necessarily applied in the rasterizing processes, which would influence the accuracy of feature delineation. (Usage of inverse distance weighted spatial interpolation techniques to acquire a legible (i.e. correlated) resolution of an evidentiary grid map))
determining an optimal buffer zone range of the evidence layer according to statistics of adjacent points. (Sun Section 5.1.2 Pg. 35 shown in Fig. 10, the distances between each deposit and its nearest neighbor deposit were calculated and statistically plotted. It can be observed that all the deposit pairs are located within 4284m, which indicates there is 100% probability that another deposit occurs within this distance for any given deposit. In other words, deposits are unlikely to be located beyond 4284 m apart, where non-deposit locations should be located. However, few locations can be selected in this range. In stead, we selected 1838m as the buffer distance within which there is an 86% probability of finding a neighboring deposit next to any given deposit (Fig. 10). (Determination of an optimal buffer zone range based on statistical probability of adjacent points))
Regarding Claim 5, the combination of Ye and Sun teaches all of the limitations of Claim 1 above, including: wherein the data integration in step S4 specifically comprises integration of multi-dimensional geological, geochemical and remote sensing mineralized spatial proxy indication information on the same grid through the neural network to obtain different mineralization information spatial proxy data sets and (Sun Section 5.1.1 Pg. 34 The absence of these advanced techniques in this contribution is because our study mainly focuses on the integration of all available predictor maps and seeks to provide intuitive representations of relevant geo information that can be more efficiently utilized and processed by machine learning methods Section 7 Pg. 44 These data were used as fresh data to test the generalizing performance of the trained models. In terms of reasonable input, the understanding of ore-forming processes of the deposit-type sought is crucial in guiding the selection of evidential features to be utilized in the MPM. Therefore, the mineral systems approach was introduced to help in translating understanding of the skarn Cu mineral system into mappable spatial proxies, generating 12 predictor maps that represent spatial proxies of source, transport, trap, and deposition processes critical for ore formation. (Using proxy information to derive different information sets of the same locale. See also Table 2, Pg. 31 for different mineralization information derived from proxies))
training point files having geological connotation. (Sun Section 1 Pg. 27 The Tongling ore district, one of the most important Cu producers in China, was selected as a case study area for the GIS-based data-driven MPM in this study because it is a mature mining area with an exploration history of more than a century. Section 5.1.2. Pg. 35 shown in Fig. 10, the distances between each deposit and its nearest neighbor deposit were calculated and statistically plotted. It can be observed that all the deposit pairs are located within 4284m, which indicates there is 100% probability that another deposit occurs within this distance for any given deposit. In other words, deposits are unlikely to be located beyond 4284 m apart, where non-deposit locations should be located. However, few locations can be selected in this range. In stead, we selected 1838m as the buffer distance within which there is an 86% probability of finding a neighboring deposit next to any given deposit (Fig. 10). (The training points are all in the same geographic are (and thus, geologically connotated), the area being the Tongling ore district of China))
Regarding Claim 6, the combination of Ye and Sun teaches all of the limitations of Claim 5 above, including: wherein the training points in a training point file comprise known deposit points and an equal number of non-deposit points randomly distributed in a blank area of the study area. (Sun Section 4.1 Pg. 32 Nonlinear SVM for binary classification is employed in this study for the pattern recognition of “mineral deposit” and “non-deposit”. Given a training data set x with n feature vectors, a labelled target dataset y is associated with each vector xi. In our case, y=1 indicates deposit and y=−1 represents non-deposit occurrence. As the input data cannot be separated linearly in the original feature space, they are firstly mapped into a higher-dimensional space H by a mapping function Φ(Burges,1998): (Training data having both positive (ore-bearing) and negative (non-ore-bearing) examples))
Regarding Claim 8, the combination of Ye and Sun teaches all of the limitations of Claim 1 above, including: wherein the hyper-parameter optimization specifically comprises steps of determining a number of radial basis functions and the number of times of iterations of machine learning in a hidden layer of the neural network, and determining accuracy of machine learning by means of parameters of a mean variance error (MSE) and the sum of squared errors (SSE) to obtain an optimal prediction result. (Sun Section 4.1 Pg. 32 An optimal separation of different classes is achieved by maximum margin of the hyperplane (Fig. 3a), which allows the generation of the lowest errors for the classifier (Asadi and Hale, 2001, Mahvash Mohammadi and Hezarkhani, 2018). The vectors used for determining the optimal separating hyperplane are called support vectors. Section 5.2 Pg 39-40 Although some empirical terms may be helpful in this process, a highly subjective trial-and-error procedure is invariably needed to obtain an optimal configuration of parameters. In this study, a 10-fold cross-validation procedure was performed to assess the predictions derived from possible combinations of parameters. The input training dataset was randomly subdivided into 10 subsets with equal size, in which a single subset was retained as the validation dataset and the other 9 subsets were used for training models. This process was repeated 10 times until each subset had been used once as the validation dataset (Xiong andZuo, 2017). The Mean Square Error (MSE) was employed to evaluate the results of cross-validation, which can be formulized as: [Formula 13] (Parameter optimization in a hidden layer (see Sec. 4.2 Pg 32) of a neural network through use of a radial basis function and mean squared error))
Regarding Claim 9, the combination of Ye and Sun teaches all of the limitations of Claim 1 above, including: wherein step S6 specifically comprises: obtaining different exploration potential areas through a C-A fractal theory for prediction results of the optimized machine learning model. (Sun Section 5.1.1 Pg 34 For example, fractal and multifractal analyses are widely used to delineate the complexity of fault distribution through various fractal indices, such as box-counting fractal dimension, multifractal spectrum, and singularity index (Agterberg et al., 1996; Zhao et al., 2011; Wang et al., 2012, 2013a). The fractal/multifractal-based singularity index mapping technique can be utilized to characterize geophysical anomalies from gravity and magnetic data (Wang et al., 2013b), and the correlation analysis method can be used to separate positively and negatively correlated gravity and magnetic anomalies (Xiao and Wang, 2017).)
Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Ye in view of Sun in further view of Li et al. (‘Mineral Prospectivity Prediction via Convolutional Neural Networks Based on Geological Big Data’, hereinafter Li).
Regarding Claim 7, the combination of Ye and Sun teaches all of the limitations as applied in Claim 1 above, including: wherein the machine learning algorithm in step S5 comprises: a radial basis function neural network (Sun Section 4.1 Pg 32 In this case, we would only need to employ K in the training process and never need to know the explicit form of Φ (Burges,1998). Four popular kernel functions commonly used in SVM include linear, radial basis, polynomial, and sigmoid function, of which the radial basis function (RBF) is chosen in this study because of its low errors and simplicity of parameters in the application of geoscience data (Zuo and Carranza,2011). (A radial basis function neural network))
The combination of Ye and Sun does not teach: wherein the machine learning algorithm in step S5 comprises: fuzzy clustering, and a feasibility neural network.
In the same field of endeavor, Li teaches: fuzzy clustering, (Section 2.1 Pg. 336 In other words, the subordination degree of an event in the fuzzy evidence weight model can be linearly assigned a value ranging from 0 to 1. Since the evidence weight can be calculated by the fuzzy evidence weight method to the classification of the assigned values of Xi, the numerous assigned values of Xi will lead to results containing noise, consequently affecting the reliability of the weighted value. Therefore, it is necessary to reclassify the assigned values. However, the reclassification will require the linearly assigned values in the original interval [0, 1] to be reduced to several classes again, resulting in information loss. Therefore, this study adopts the double fuzzy method, which does not require reclassifying the evidence layer, but requires considering the evidence layer as the fuzzy set related to event i, and the calculation model of the fuzzy probability, conditional fuzzy probability, and fuzzy evidence weight is defined. (Classification of elements/weights through fuzzy means))
a feasibility neural network. (Section 3.1 Pg. 339 Based on the difference in the input ore-controlling factor combinations, we divide the comparative experiments into six groups, and the two prediction methods, namely the 3D CNN and WofE models, are applied to each group (Table 5). The objective is to determine the optimal ore-controlling factor combination and combination number by comparing the training loss, training accuracy, validation accuracy, and validation loss of the 3D CNN model trained on different factor layer combinations, and to analyze the feasibility and superiority of the 3D CNN model applied to 3D MPM by comparing the prediction results of the 3D CNN and WofE models. (Determination that the model is sufficient (feasible) to perform its task))
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention to have combined a fuzzy clustering and feasibility neural network as taught by Li into the combination of Ye and Sun as all three references are in the same field of geological detection and metallogenic prediction, and doing so would be desirable in order to facilitate the application of advancing machine learning technologies when dealing with the enormous datasets typical in the field of geological modeling (Li Section 1)
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
CAI et al. (CN 110442666 A) discusses methods for predicting mineral resources based on neural network modelling
HE et al. (CN 115759816 A) discusses methods for remote sensing detection of gold deposits and application of mineral models.
Yin et al. (CN 103942841 B) discusses usage of conceptual models to process and predict mineral deposits.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JUSTIN A CARDOSO whose telephone number is (571)272-8512. The examiner can normally be reached M-F 7:30 - 5:00, alternate Friday's off.
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, Jennifer Welch can be reached at (571) 272-7212. 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.
/JUSTIN CARDOSO/
Patent Examiner, Art Unit 2143
/JENNIFER N WELCH/Supervisory Patent Examiner, Art Unit 2143
Prosecution Insights