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 .
Priority
Acknowledgment is made of applicant’s claim for priority. Application is a 371 of PCT/US2021/060362 and claims the benefit of U.S. Provisional Application 63/117,554 filed 11/24/2020. As such, the effective filing date of claims 1-12 is 11/24/2020.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 5/22/2023 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Claim Status
Claims 1-12 are pending.
Claims 1-12 are rejected.
Specification
The use of the term NVIDIA, which is a trade name or a mark used in commerce, has been noted in this application. The term should be accompanied by the generic terminology; furthermore the term should be capitalized wherever it appears or, where appropriate, include a proper symbol indicating use in commerce such as ™, SM , or ® following the term.
Although the use of trade names and marks used in commerce (i.e., trademarks, service marks, certification marks, and collective marks) are permissible in patent applications, the proprietary nature of the marks should be respected and every effort made to prevent their use in any manner which might adversely affect their validity as commercial marks.
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 1-12 are rejected under 35 U.S.C. 101 because the claimed invention is directed to abstract ideas without significantly more. The claims recite a method and device for predicting elasticity of biological tissues. This judicial exception is not integrated into a practical application because while claims 1-12 attempt to integrate the exception into a practical application, said practical application is a generically recited computer element that does not add meaningful limitations to the abstract idea as it is simply implementing the abstract idea on a computer. The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because the computer elements only store and retrieve information in memory as well as perform basic calculations that are known to be well-understood, routine and conventional computer functions as recognized by the decisions listed in MPEP § 2106.05(d).
Framework with which to Analyze Subject Matter Eligibility:
Step 1: Are the claims directed to a category of statutory subject matter (a process, machine manufacture, or composition of matter)? [see MPEP § 2106.03]
Claims are directed to statutory subject matter, specifically a method (claims 1-11), and a device (claim 12).
Step 2A Prong One: Do the claims recite a judicially recognized exception, i.e., an abstract idea, a law of nature, or a natural phenomenon? [see MPEP § 2106.04(a)]
With respect to the Step 2A Prong One evaluation, the instant claims are found herein to recite abstract ideas that fall into the grouping of mental processes and mathematical concepts.
The following claims recite abstract ideas (mental processes and mathematical concepts):
Claims 1 and 12: Producing a predicted stress distribution, applying convolutional filters, and predicting an elasticity distribution are processes of calculating information that can be done via pen and paper or within the human mind and are therefore abstract ideas, specifically mental processes. The data set comprising the specified data is merely further limiting the data itself which is an abstract idea, specifically a mental process.
Claim 2: Updating weights at each iteration is a process of calculating information that can be done via pen and paper or within the human mind and is therefore an abstract idea, specifically a mental process.
Claim 3: Applying filters encoded with an equilibrium condition in x and y directions is a verbal articulation of a mathematical process and is therefore an abstract idea, specifically a mathematical concept.
Claim 4: Applying an elastic constitutive relation encoded into the DNN is process of calculating information that can be done via pen and paper or within the human mind and is therefore an abstract idea, specifically a mental process.
Claim 5: Encoding the equilibrium conditions into the DNN is a process of calculating information that can be done via pen and paper or within the human mind and is therefore an abstract idea, specifically a mental process.
Claim 6: The DNN operating using an elastic constitutive relation and equilibrium conditions with no labeled data is merely further limiting the data itself which is an abstract idea, specifically a mental process.
Claim 7: The strain data comprising the specified data is merely further limiting the data itself which is an abstract idea, specifically a mental process.
Claim 8: The final elasticity distribution having a higher resolution than the strain data is merely further limiting the data itself which is an abstract idea, specifically a mental process.
Claim 9: The solid comprising human tissue is merely further limiting the data itself which is an abstract idea, specifically a mental process.
Claim 10: Applying an adaptive moment optimizer in the DNN is a process of calculating information that can be done via pen and paper or within the human mind and is therefore an abstract idea, specifically a mental process.
Claim 11: The DNN performing full-batch learning is merely further limiting the data itself which is an abstract idea, specifically a mental process.
Step 2A Prong Two: If the claims recite a judicial exception under prong one, then is the judicial exception integrated into a practical application? [see MPEP § 2106.04(d)]
Because the claims do recite judicial exceptions, direction under Step 2A Prong Two provides that the claims must be examined further to determine whether they integrate the abstract ideas into a practical application.
The following claims recite the following additional elements in the form of non-abstract elements:
Claim 1: A computer is a generic and nonspecific element of a computer that does not improve the functioning of any computer or technology described herein [See MPEP § 2106.04(d)(1) and MPEP § 2106.05(d)]. Receiving a data set is an insignificant extra solution activity, specifically mere data gathering and outputting (See Performing clinical tests on individuals to obtain input for an equation, In re Grams, 888 F.2d 835, 839-40; 12 USPQ2d 1824, 1827-28 (Fed. Cir. 1989) and Determining the level of a biomarker in blood, Mayo, 566 U.S. at 79, 101 USPQ2d at 1968. See also PerkinElmer, Inc. v. Intema Ltd., 496 Fed. App'x 65, 73, 105 USPQ2d 1960, 1966 (Fed. Cir. 2012) (assessing or measuring data derived from an ultrasound scan, to be used in a diagnosis)) [See MPEP § 2106.05(g)].
Claim 12: A computing device, processors, and code are generic and nonspecific elements of a computer that do not improve the functioning of any computer or technology described herein [See MPEP § 2106.04(d)(1) and MPEP § 2106.05(d)]. Receiving a data set is an insignificant extra solution activity, specifically mere data gathering and outputting (See Performing clinical tests on individuals to obtain input for an equation, In re Grams, 888 F.2d 835, 839-40; 12 USPQ2d 1824, 1827-28 (Fed. Cir. 1989) and Determining the level of a biomarker in blood, Mayo, 566 U.S. at 79, 101 USPQ2d at 1968. See also PerkinElmer, Inc. v. Intema Ltd., 496 Fed. App'x 65, 73, 105 USPQ2d 1960, 1966 (Fed. Cir. 2012) (assessing or measuring data derived from an ultrasound scan, to be used in a diagnosis)) [See MPEP § 2106.05(g)].
Step 2B: If the claims do not integrate the judicial exception, do the claims provide an inventive concept? [see MPEP § 2106.05]
Because the additional claim elements do not integrate the abstract ideas into a practical application, the claims are further examined under Step 2B, which evaluates whether the additional elements, individually and in combination, amount to significantly more than the judicial exception itself by providing an inventive concept.
The claims do not recite additional elements that are sufficient to amount to significantly more than the judicial exceptions because the claims recite additional elements that are generic, conventional, nonspecific, or insignificant extra solution activity. These additional elements include:
The additional elements of a computer, computing device, processors, and code are generic and nonspecific elements of a computer that are well-understood, routine and conventional within the art and therefore do not improve the functioning of any computer or technology described therein (Receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information), Performing repetitive calculations, Flook, 437 U.S. at 594, 198 USPQ2d at 199 (recomputing or readjusting alarm limit values), and Storing and retrieving information in memory, Versata Dev. Group, Inc. v. SAP Am., Inc., 793 F.3d 1306, 1334, 115 USPQ2d 1681, 1701 (Fed. Cir. 2015)) [See § MPEP 2106.05(d)(II)]. Therefore, taken both individually and as a whole, the additional elements do not amount to significantly more than the judicial exception by providing an inventive concept.
The additional element of receiving a data set is an insignificant extra solutional activity, specifically mere data gathering, that are recognized as well understood, routine and conventional by the courts (See Analyzing DNA to provide sequence information or detect allelic variants, Genetic Techs. Ltd., 818 F.3d at 1377; 118 USPQ2d at 1546, and Receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information)) [See MPEP § 2106.05(g)]. Therefore, taken both individually and as whole, the additional elements do not amount to significantly more than the judicial exception by providing an inventive concept.
Therefore, claims 1-12, when the limitations are considered individually and as a whole, are rejected under 35 U.S.C. § 101 as being directed to non-statutory subject matter.
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.
Claims 1-3, 5, and 7-9, and 12 are rejected under 35 U.S.C. 103 as being unpatentable over Won et al. (US 20200205665 A1) and Eskandari et al. (US 20130253318 A1).
Claim 1 is directed to a method of predicting elasticity of a solid using a deep neural network.
Claim 12 is directed to a computing device for predicting elasticity of a solid using a deep neural network.
Won et al. teaches in paragraph [0122] “Here we present a method to obtain the absolute elasticity of the target. Here the embodiment is for a human tumor application, but other applications are possible. In order to compute the absolute Young's modulus (elasticity) of the tumor, we utilize deep learning. We will model the target's mechanical properties such as size of the deformation due to elasticity”, in paragraph [0127] “The objective is to investigate the effects of the skin in a biological tissue associated with the application of loading on the tissue”, in paragraph [0128] “Absolute Young's Modulus Determination using inverse Approach…One technique for solving inverse problems is using a deep learning of the forward results. The first part is the data acquisition and construction of tactile maps using the FEM. In the second part, we require an inverse model that takes a CIS map as input and produces the size and elasticity of the tissue and tumor that has been imaged as output. An inverse algorithm estimates the elastic modulus, depth, and size of the tumor from maximum deformation, total deformation, and deformation area, which are obtained from CIS. We use a deep neural network method”, reading on a computer-implemented method of predicting elasticity of a solid, comprising, receiving a data set comprised of position data and corresponding strain data for points on a solid at a deep neural network (DNN), and producing a predicted stress distribution.
Won et al. does not teach the application of convolution filters to the stress distribution or iteratively using the information to predict an elasticity distribution.
Eskandari et al. teaches in paragraph [0073] “spatial filter can be applied to the elasticity and viscosity distributions at every iteration to make the solution of the problem smooth and less sensitive to the displacement noise. As a result, the elasticity or viscosity parameter of each element 1102 will be a weighted sum of its own value and the values of that parameters in the adjacent elements 1102. A linear filter can thus be constructed in the form of a sparse matrix that contains the required weights for each element 1102. The filter can be convolved with the updated distribution of the parameter at every iteration to conduct the optimization toward a smooth solution”, reading on applying convolutional filters to the predicted stress distribution to produce residual force maps. Eskandari et al. teaches in paragraph [0026] “The second approach involves assuming that tissue deformation can be modeled by a linear viscoelastic finite element model…Such a model can be linear or non-linear in the remaining unknown parameters. The linear equations can be solved iteratively or non-iteratively; the equations that are not linear can be solved iteratively”, reading on predicting an elasticity distribution of the solid by iteratively using the residual force maps and an equilibrium condition until the predicted stress distribution satisfies the equilibrium condition, producing the final elasticity distribution.
It would have been obvious at the time of first to have modified the teachings of Won et al. for a method of predicting elasticity of a solid using a deep neural network and strain data, with the teachings of Eskandari et al. for the application of convolutional filters and iterative use of the data in the prediction of the elasticity as both are directed to the prediction of elasticity within human tissues, specifically tumors using strain information. One would have had a reasonable expectation of success given that it is merely a substitution of one known method with another known method, specifically the addition of a convolution filter and iterative calculations, and such methods allow for the iterative optimization in prediction accuracy as described in the portions of Eskandari et al. cited above (paragraphs [0026] and [0073]). Therefore, it would have been obvious at the time of first filing and to a person skilled in the art, to have modified the teachings of each and to be successful.
Claim 2 is directed to the method of claim 1 but further specifies updating weights in determining the elasticity distribution.
Eskandari et al. teaches in paragraph [0067] “In this equation, p=[E.sup.T .eta..sup.T].sup.T is a 2m parameter vector composed of elasticity and viscosity parameters, considering that there are m elements in the FEM model. The parameter .alpha. is a weight determining the relative importance of the imaginary and real components of the displacement in the inverse problem. .alpha. can simply be equal to one…A number of cost functions similar to (8) can be utilized to solve for model parameters, and the function in (8) has been used only as an example of a particular embodiment”, reading on further comprising updating weights used in determining the elasticity distribution based upon the strain data at each iteration.
Claim 3 is directed to the method of claim 1 but further specifies the application of filters encoded with an equilibrium condition in both an x and y direction.
Eskandari et al. teaches in paragraph [0031] “If ultrasound is used as the imaging modality for viscoelasticity measurement of the tissue 28, the ultrasound probe 26 is placed on the surface of the tissue 28 and acquires radio-frequency (RF) data along several A-lines (straight lines extending from the surface of the probe 26) in one, two or three dimensions (1D, 2D or 3D). Therefore, each point in the tissue will have a displacement in 3D which can be projected into axial (along an A-line), lateral (perpendicular to axial in the imaging plane) and elevational (perpendicular to the imaging plane) directions; these directions are depicted in FIG. 3. Axial, lateral and elevational strains are the spatial derivative of such displacements. By applying a Fourier transform or a similar frequency transform to the time-domain displacement or strains, the displacement or strain phasors can be calculated at a desired frequency (.omega.). For example, the phasor of the axial displacement of a certain location at frequency w can be obtained by calculating the Fourier transform of that displacement signal at .omega..”, reading on wherein using the equilibrium data comprises applying filters encoded with an equilibrium condition in an x-direction and a y- direction.
Claim 5 is directed to the method of claim 1 but further specifies encoding the equilibrium conditions into the DNN prior to receiving the data.
Eskandari et al. teaches in paragraph [0099] “In one embodiment of this approach, a convex optimization problem can be set up in which the minimization functional is zero or any constant scalar. In such a linear programming problem, (21) can be applied as a constraint to the problem. Other linear equality and inequality constraints can be added to the optimization problem while maintaining its "linear program" formulation. For example, the elasticity and viscosity parameters of the model can be constrained within a range that is known for soft tissue. Furthermore, a limit can be imposed on the spatial variation of the parameters. For example, an inequality constraint can be assumed on a linear combination of the parameters”, reading on further comprising encoded the equilibrium conditions into the DNN prior to receiving the data set.
Claim 7 is directed to the method of claim 1 but further specifies that the data comprises the specified data.
Won et al. teaches in paragraph [0070] “The applied force can be measured by one or multiple force sensors 38 in the frontend attachment 24. This force sensor 38 can be a pressure sensor, load cell, strain sensor, piezoelectric sensor, strain gauge, etc. The direction of the applied three is important. The frontend 24 or the mobile device 22 can have attitude and position sensors to determine the applied force direction in absolute terms. More accurate results will be obtained if the applied force direction is measured by various sensors 38, so that the direction of the applied force relative to the device 20, and in particular any unevenness in the applied pressure, can be detected”, reading on wherein the strain data comprises strain data for each point of the position data and results from one of either experiments or simulation.
Claim 8 is directed to the method of claim 1 but further specifies that the final elasticity distribution has a higher resolution than the data.
Won et al. teaches in paragraph [0128] “The first part is the data acquisition and construction of tactile maps using the FEM. In the second part, we require an inverse model that takes a CIS map as input and produces the size and elasticity of the tissue and tumor that has been imaged as output. An inverse algorithm estimates the elastic modulus, depth, and size of the tumor from maximum deformation, total deformation, and deformation area, which are obtained from CIS. We use a deep neural network method called “Deep Brief Net” (DBN) for this inverse modeling. This DUN method improves the accuracy of the inverse algorithm (LeCun et al., 2015). A conventional Artificial Neural Network (ANN) has overfitting issues, which results in poor generalization performance. DBN can be pre-trained by an unsupervised manner to avoid the overfitting and the DBN with latent variable space in its deep structure can represent complex nonlinear functions otherwise not efficiently representable by the ANN with shallow structure (i.e., only one or two hidden layers). We use the DBN to improve the accuracy of the inverse algorithm”, reading on wherein the final elasticity distribution has a higher resolution than the strain data.
Claim 9 is directed to the method of claim 1 but further specifies that the solid comprises human tissue.
Won et al. teaches in paragraph [0032] “One possible application of the Mobile-platform Compression-induced Imaging System (CIS) is detection of malignant human tumors. An embodiment of the new mobile system is therefore geared towards medicine for its potential to aid physicians in prescreening of tumors and for training in the technique of palpation. The mechanical properties this system could provide are a valuable resource in the tumor screening process”, reading on wherein the solid comprises human tissue.
Claim 4 is rejected under 35 U.S.C. 103 as being unpatentable over Won et al. (US 20200205665 A1) and Eskandari et al. (US 20130253318 A1) as applied to claims 1-3, 5, 7-9, and 12 above, and further in view of Pai et al. (WO 2020014781 A1).
Claim 4 is directed to the method of claim 1 but further specifies the application of elastic constitutive relation encoded into the DNN.
Won et al. and Eskandari et al. teach the method of claim 1 as previously described.
Won et al. and Eskandari et al. do not teach the application of elastic constitutive relation encoded into the DNN.
Pai et al. teaches in paragraph [0076] “Model parameters 30 may include constitutive properties (e.g. elasticity parameters) of the body tissues of test subject / that are probed and, in some embodiments, other properties of the tissue of test subject /.sup.', such as mass density, effective thickness and/or the like. Non-limiting examples of elasticity parameters include the parameters of non-linear hyperelastic models, such as the Mooney-Rivlin model, Gasser-Ogden-Holzapfel model, polynomial hyperelastic models, related models and/or the like. Such constitutive models may account for the anisotropic behavior of tissue deformation in different material directions and may include orthotropic models”, reading on applying an elastic constitutive relation encoded into the DNN prior to receiving the data set.
It would have been obvious at the time of first filing to have modified the teachings of Won et al. and Eskandari et al. for the method of claim 1, with the teachings of Pai et al. for the application an elastic constitutive relation encoded into the DNN as all of the references are directed to the prediction of medical imaging and the through it the examination of elasticity, as well as other features of the tissues. Additionally, users would benefit from such as the methods would allow for performing calculations within the network based on independent elastic constitutive relation as described in paragraph [0076]. One would have had a reasonable expectation of success given that each of the references is working with similar image data, and it would merely be a substitution of one known method for another. Therefore, it would have been obvious at the time of first filing to a person skilled in the art to have modified the teachings of each and to be successful.
Claim 6 is rejected under 35 U.S.C. 103 as being unpatentable over Won et al. (US 20200205665 A1) and Eskandari et al. (US 20130253318 A1) as applied to claims 1-3, 5, 7-9, and 12 above, and further in view of Pai et al. (WO 2020014781 A1) and Sun et al. (Computerized Medical Imaging and Graphics (2017) 4-9).
Claim 6 is directed to the method of claim 1 but further specifies the DNN using elastic constitutive relation and equilibrium conditions with no labeled data.
Won et al. and Eskandari et al. teach the method of claim 1 as previously described.
Won et al. and Eskandari et al. do not teach the DNN using elastic constitutive relation and equilibrium conditions with no labeled data.
Pai et al. teaches in paragraph [0076] “Model parameters 30 may include constitutive properties (e.g. elasticity parameters) of the body tissues of test subject / that are probed and, in some embodiments, other properties of the tissue of test subject /.sup.', such as mass density, effective thickness and/or the like. Non-limiting examples of elasticity parameters include the parameters of non-linear hyperelastic models, such as the Mooney-Rivlin model, Gasser-Ogden-Holzapfel model, polynomial hyperelastic models, related models and/or the like. Such constitutive models may account for the anisotropic behavior of tissue deformation in different material directions and may include orthotropic models”.
Sun et al. teaches in the abstract “In this study we developed a graph based semi-supervised learning (SSL) scheme using deep convolutional neural network (CNN) for breast cancer diagnosis. CNN usually needs a large amount of labeled data for training and fine tuning the parameters, and our proposed scheme only requires a small portion of labeled data in training set. Four modules were included in the diagnosis system: data weighing, feature selection, dividing co-training data labeling, and CNN. 3158 region of interests (ROIs) with each containing a mass extracted from 1874 pairs of mammogram images were used for this study. Among them 100 ROIs were treated as labeled data while the rest were treated as unlabeled. The area under the curve (AUC) observed in our study was 0.8818, and the accuracy of CNN is 0.8243 using the mixed labeled and unlabeled data”, reading on wherein the DNN operates using an elastic constitutive relation and equilibrium conditions with no labeled data.
It would have been obvious at the time of first filing to have modified the teachings of Won et al. and Eskandari et al. for the method of claim 1, with the teachings of Pai et al. for the application an elastic constitutive relation encoded into the DNN as all of the references are directed to the prediction of medical imaging and the through it the examination of elasticity, as well as other features of the tissues. Additionally, users would benefit from such as the methods would allow for performing calculations within the network based on independent elastic constitutive relation as described in paragraph [0076]. Furthermore, it would have been obvious to combine the teachings of the others with the teachings of Sun et al. for the use of unlabeled data as Sun et al. teaches in the abstract “area under the curve (AUC) observed in our study was 0.8818, and the accuracy of CNN is 0.8243 using the mixed labeled and unlabeled data”. One would have had a reasonable expectation of success given that each of the references is working with similar image data, and it would merely be a substitution of known methods for others. Therefore, it would have been obvious at the time of first filing to a person skilled in the art to have modified the teachings of each and to be successful.
Claim 10 is rejected under 35 U.S.C. 103 as being unpatentable over Won et al. (US 20200205665 A1) and Eskandari et al. (US 20130253318 A1) as applied to claims 1-3, 5, 7-9, and 12 above, and further in view of Sataswathi et al. (International Conference on Computer Vision and Image Processing (2019) 123-133).
Claim 10 is directed to the method of claim 1 but further specifies applying an adaptive moment optimizer in the DNN.
Won et al. and Eskandari et al. teach the method of claim 1 as previously described.
Eskandari et al. teaches in paragraph [0073] “a spatial filter can be applied to the elasticity and viscosity distributions at every iteration to make the solution of the problem smooth and less sensitive to the displacement noise. As a result, the elasticity or viscosity parameter of each element 1102 will be a weighted sum of its own value and the values of that parameters in the adjacent elements 1102. A linear filter can thus be constructed in the form of a sparse matrix that contains the required weights for each element 1102. The filter can be convolved with the updated distribution of the parameter at every iteration to conduct the optimization toward a smooth solution”.
Won et al. and Eskandari et al. do not teach the application of an adaptive moment optimizer in the DNN.
Sataswathi et al. teaches on page 127, paragraph 1 “CNN is used in which firstly region of interest is selected using bounding box and then three different optimizers namely, adaptive moment optimizer (Adam), stochastic gradient descent with momentum (SGDM) and root mean square propagation (RMSProp) are used to classify three different classes of tumors”, reading on further comprising applying an adaptive moment optimizer in the DNN.
It would have been obvious to a person skilled in the art to have modified the teachings of Won et al. and Eskandari et al. for the method of claim 1 with the teachings of Sataswathi et al. for the use of an adaptive moment optimizer as the latter is using CNNs to identify brain images as being cancerous using the specified method. One would have had a reasonable expectation of success given that it is merely a substitution of one known method, a standard optimizer, for a more specific known method, an adaptive moment optimizer. Therefore, it would have been obvious at the time of first filing to a person skilled in the art to have modified the teachings of each and to be successful.
Claim 11 is rejected under 35 U.S.C. 103 as being unpatentable over Won et al. (US 20200205665 A1) and Eskandari et al. (US 20130253318 A1) as applied to claims 1-3, 5, 7-9, and 12 above, and further in view of You et al. (arXiv preprint (2017) 1-8).
Claim 11 is directed to the method of claim 1 but further specifies the DNN as performing full-batch learning.
Won et al. and Eskandari et al. teach the method of claim 1 as previously described.
Eskandari et al. teaches in paragraph [0073] “a spatial filter can be applied to the elasticity and viscosity distributions at every iteration to make the solution of the problem smooth and less sensitive to the displacement noise. As a result, the elasticity or viscosity parameter of each element 1102 will be a weighted sum of its own value and the values of that parameters in the adjacent elements 1102. A linear filter can thus be constructed in the form of a sparse matrix that contains the required weights for each element 1102. The filter can be convolved with the updated distribution of the parameter at every iteration to conduct the optimization toward a smooth solution”.
Won et al. and Eskandari et al. do not teach the DNN as performing full-batch learning.
You et al. teaches in the abstract “A common way to speed up training of large convolutional networks is to add computational units. Training is then performed using data-parallel synchronous Stochastic Gradient Descent (SGD) with mini-batch divided between computational units. With an increase in the number of nodes, the batch size grows. But training with large batch size often results in the lower model accuracy. We argue that the current recipe for large batch training (linear learning rate scaling with warm-up) is not general enough and training may diverge. To overcome this optimization difficulty, we propose a new training algorithm based on Layer-wise Adaptive Rate Scaling (LARS). Using LARS, we scaled Alexnet up to a batch size of 8K, and Resnet-50 to a batch size of 32K”, reading on wherein the DNN performs full-batch learning.
It would have been obvious to a person skilled in the art to have modified the teachings of Won et al. and Eskandari et al. for the method of claim 1 with the teachings of You et al. for the use of full-batch learning as the latter specifically states “we scaled Alexnet up to a batch size of 8K, and Resnet-50 to a batch size of 32K without loss in accuracy”. One would have had a reasonable expectation of success given that this is merely a substitution of one known method, batch learning, with another, full batch learning, that the latter citation shows to be equally as accurate, but quicker than the former method. Therefore, it would have been obvious at the time of first filing to a person skilled in the art to have modified the teachings of each and to be successful.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to KEENAN NEIL ANDERSON-FEARS whose telephone number is (571)272-0108. The examiner can normally be reached M-Th, alternate F, 8-5.
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, Karlheinz Skowronek can be reached at 571-272-9047. 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.
/K.N.A./ Examiner, Art Unit 1687
/LARRY D RIGGS II/ Supervisory Patent Examiner, Art Unit 1686
Prosecution Insights