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 .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 8/4/2024 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Priority
Acknowledgment is made of applicant’s claim for foreign priority under 35 U.S.C. 119 (a)-(d). The certified copy has been filed in the instant application, filed on 6/3/2022. Receipt is acknowledged of certified copies of papers required by 37 CFR 1.55. Effective filing date for claims 1-20 is 6/10/2021.
Claim Status
Claims 1-20 are pending.
Claims 1-20 are rejected.
Drawings
Color photographs and color drawings are not accepted in utility applications unless a petition filed under 37 CFR 1.84(a)(2) is granted. Figures 2, and 7-10 are included in color. Any such petition must be accompanied by the appropriate fee set forth in 37 CFR 1.17(h), one set of color drawings or color photographs, as appropriate, if submitted via the USPTO patent electronic filing system or three sets of color drawings or color photographs, as appropriate, if not submitted via the via USPTO patent electronic filing system, and, unless already present, an amendment to include the following language as the first paragraph of the brief description of the drawings section of the specification:
The patent or application file contains at least one drawing executed in color. Copies of this patent or patent application publication with color drawing(s) will be provided by the Office upon request and payment of the necessary fee.
Color photographs will be accepted if the conditions for accepting color drawings and black and white photographs have been satisfied. See 37 CFR 1.84(b)(2).
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(d):
(d) REFERENCE IN DEPENDENT FORMS.—Subject to subsection (e), a claim in dependent form shall contain a reference to a claim previously set forth and then specify a further limitation of the subject matter claimed. A claim in dependent form shall be construed to incorporate by reference all the limitations of the claim to which it refers.
The following is a quotation of pre-AIA 35 U.S.C. 112, fourth paragraph:
Subject to the following paragraph [i.e., the fifth paragraph of pre-AIA 35 U.S.C. 112], a claim in dependent form shall contain a reference to a claim previously set forth and then specify a further limitation of the subject matter claimed. A claim in dependent form shall be construed to incorporate by reference all the limitations of the claim to which it refers.
Claims 2 and 16 are rejected under 35 U.S.C. 112(d) or pre-AIA 35 U.S.C. 112, 4th paragraph, as being of improper dependent form for failing to further limit the subject matter of the claim upon which it depends, or for failing to include all the limitations of the claim upon which it depends. Specifically, claims 1 and 15 require the inputting of the genetic data into both a genetic feature generation network for feature extraction and both the genetic data and features into a genetic identification network layer to obtain a testing result. Conversely claims 2 and 16 require the performing of genetic testing using the genetic data and extracted features to obtain a reference. However these steps cannot be separated as such, as inputting any information into a network model will automatically cause it to perform the steps of processing and calculating the information provided, i.e. if the process of claim 1 is followed it will automatically produce the results of claim 2 (here there is no difference between a testing result and a reference). Therefore, as the steps cannot be separated, claims 2 and 16 do not further limit claims 1 and 15 as they are already performing the steps of claims 2 and 16 inherently. Applicant may cancel the claim(s), amend the claim(s) to place the claim(s) in proper dependent form, rewrite the claim(s) in independent form, or present a sufficient showing that the dependent claim(s) complies with the statutory requirements.
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 1-20 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, CRM, and apparatus for genetic training and testing. The judicial exception is not integrated into a practical application because while claims 1-20 attempt to integrate the exception into a practical application, said application is either generically recited computer elements that do not add a meaningful limitation to the abstract idea or it is insignificant extra solution activity and merely 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 stator 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 1-6), a CRM (Claims 7-14), and an apparatus (Claims 15-20)
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)]
The claims herein recite abstract ideas, specifically mental processes and mathematical concepts.
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.
Claims 1 and 15: While not directing stating the use of a judicial exception, claims specify the inputting of genetic data into both a feature generation network layer and a genetic identification network, there is no difference between inputting and running and so are directed to the feature extraction and identification results, which are processes of comparing/contrasting and calculating information that can be done via pen and paper or within the human mind and are therefore abstract ideas, specifically mental processes.
Claims 2 and 16: Performing genetic testing processing is a process of comparing/contrasting and 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.
Claims 3 and 17: The testing reference information including the specified data is merely further limiting the data itself which is an abstract idea, specifically a mental process.
Claims 5 and 19: Determining a loss function, and optimizing the feature generation network layer using the loss function are verbal articulations of mathematical processes and are therefore abstract ideas, specifically mathematical concepts.
Claims 6 and 20: Optimizing the data identification network based on the loss function and determining the optimized feature generation network layer are verbal articulations of mathematical processes and are therefore abstract ideas, specifically mathematical concepts.
Claim 7: Determining genetic features corresponding to genetic samples and performing learning based on reference genetic results are processes of comparing/contrasting and calculating information that can be done via pen and paper or within the human mind and are therefore abstract ideas, specifically mental processes.
Claim 8: Performing learning based on the genetic samples and performing learning based on the enhanced features are processes of comparing/contrasting and calculating information that can be done via pen and paper or within the human mind and are therefore abstract ideas, specifically mental processes.
Claim 9: Performing learning based on the genetic features and the reference and optimizing the feature generation sub-model of comparing/contrasting and calculating information that can be done via pen and paper or within the human mind and are therefore abstract ideas, specifically mental processes.
Claim 10: Obtaining a loss function and determining the optimized feature generation sub-model of comparing/contrasting and calculating information that can be done via pen and paper or within the human mind and are therefore abstract ideas, specifically mental processes. Optimizing the data identification model based on the loss function is a verbal articulation of a mathematical process and is therefore an abstract idea, specifically a mathematical concept.
Claim 11: Analyzing and processing the genetic features and determining the loss function are processes of comparing/contrasting and calculating information that can be done via pen and paper or within the human mind and are therefore abstract ideas, specifically mental processes.
Claim 12: Performing learning based on the reference features and enhanced features, and optimizing the feature generation sub-model using the adversarial discriminative model are processes of comparing/contrasting and calculating information that can be done via pen and paper or within the human mind and are therefore abstract ideas, specifically mental processes.
Claim 13: Optimizing the feature generation sub-model based on the judgement and identification result is a process of comparing/contrasting and 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 14: Determining the genetic testing model and analyzing/processing the genetic data are processes of comparing/contrasting and calculating information that can be done via pen and paper or within the human mind and are therefore abstract ideas, specifically mental processes.
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) and MPEP § 2106.05(a)-(c) & (e)-(h)]
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:
Claims 1 and 15: Obtaining genetic data to be processed, inputting the genetic data into a feature generation network layer, and inputting the genetic data into a genetic identification network layer are insignificant extra solution activities, specifically necessary data gathering (See 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) (presenting offers and gathering statistics amounted to mere data gathering), 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 Claims derived from an ultrasound scan, to be used in a diagnosis)) [See MPEP § 2106.05(g)]. An apparatus, processors, memory, and instructions 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)].
Claims 4 and 18: Obtaining the testing result is an insignificant extra solution activity, specifically necessary data outputting (See 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) (presenting offers and gathering statistics amounted to mere data gathering), 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 Claims derived from an ultrasound scan, to be used in a diagnosis)) [See MPEP § 2106.05(g)].
Claims 5 and 19: Obtaining a standard data type and inputting the genetic features to a data identification network layer are insignificant extra solution activities, specifically necessary data gathering (See 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) (presenting offers and gathering statistics amounted to mere data gathering), 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 Claims derived from an ultrasound scan, to be used in a diagnosis)) [See MPEP § 2106.05(g)].
Claim 7: Computer readable storage media, instructions, and processors 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)]. Obtaining genetic samples, and performing training based on reference results are insignificant extra solution activities, specifically necessary data gathering (See 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) (presenting offers and gathering statistics amounted to mere data gathering), 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 Claims derived from an ultrasound scan, to be used in a diagnosis)) [See MPEP § 2106.05(g)].
Claim 8: Performing training based on the genetic samples, features, and enhanced features, and performing training based on the enhanced features and reference results are insignificant extra solution activities, specifically necessary data gathering (See 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) (presenting offers and gathering statistics amounted to mere data gathering), 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 Claims derived from an ultrasound scan, to be used in a diagnosis)) [See MPEP § 2106.05(g)].
Claim 9: Performing training based on the genetic features and the reference results is an insignificant extra solution activity, specifically necessary data gathering (See 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) (presenting offers and gathering statistics amounted to mere data gathering), 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 Claims derived from an ultrasound scan, to be used in a diagnosis)) [See MPEP § 2106.05(g)].
Claim 12: Obtaining reference features and performing training based on the reference features are insignificant extra solution activities, specifically necessary data gathering (See 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) (presenting offers and gathering statistics amounted to mere data gathering), 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 Claims derived from an ultrasound scan, to be used in a diagnosis)) [See MPEP § 2106.05(g)].
Claim 13: Obtaining a judgement and identification result is an insignificant extra solution activities, specifically necessary data outputting (See 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) (presenting offers and gathering statistics amounted to mere data gathering), 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 Claims derived from an ultrasound scan, to be used in a diagnosis)) [See MPEP § 2106.05(g)].
Claim 14: Obtaining genetic data is an insignificant extra solution activity, specifically necessary data gathering (See 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) (presenting offers and gathering statistics amounted to mere data gathering), 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 Claims 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 idea 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 exception because the claims recite additional elements that are generic, conventional or nonspecific. These additional elements include:
The additional elements of an apparatus, processors, memory, instructions, and computer readable storage media 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 (See 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 elements of obtaining genetic data to be processed (Conventional: Specification Paragraph [0083] – transmitting information over a network), inputting the genetic data into a feature generation network layer, inputting the genetic data into a genetic identification network layer, obtaining the testing result, obtaining a standard data type, inputting the genetic features to a data identification network layer, obtaining genetic samples, performing training based on reference results (Conventional: Basheer et al. Page 7, column 2, paragraph 2 - “ANN learning is performed iteratively as the network is presented with training examples, similar to the way we learn from experience”), performing training based on the genetic samples, features, and enhanced features (Conventional: Basheer et al. Page 7, column 2, paragraph 2 - “ANN learning is performed iteratively as the network is presented with training examples, similar to the way we learn from experience”), performing training based on the enhanced features and reference results (Conventional: Basheer et al. Page 7, column 2, paragraph 2 - “ANN learning is performed iteratively as the network is presented with training examples, similar to the way we learn from experience”), performing training based on the genetic features and the reference results (Conventional: Basheer et al. Page 7, column 2, paragraph 2 - “ANN learning is performed iteratively as the network is presented with training examples, similar to the way we learn from experience”), obtaining reference features, performing training based on the reference features (Conventional: Basheer et al. Page 7, column 2, paragraph 2 - “ANN learning is performed iteratively as the network is presented with training examples, similar to the way we learn from experience”), obtaining a judgement and identification result, and obtaining genetic data, are insignificant extra solution activities, specifically necessary data gathering (See 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) (presenting offers and gathering statistics amounted to mere data gathering), 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)]. 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 1-20, when the limitations are considered individually and as a whole, are rejected under 35 USC § 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 1-2, 4, 7-8, 14-16, and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Wisesty et al. (IOP Conference Series: Materials Science and Engineering (2020) 1-11) and Liu et al. (Trends in genetics (2019) 852-867).
Claim 1 is directed to a method for using genetic data to extract features and use the features and genetic information to perform genetic testing.
Claim 15 is directed to an apparatus for using genetic data to extract features and use the features and genetic information to perform genetic testing.
Wisesty et al. teaches in the abstract “Early detection of cancer can be done through the analysis of DNA abnormalities from blood cell samples, where the sampling does not require surgery, non-invasive and painless, and can reduce the sampling cost. DNA abnormalities can occur due to heredity or gene mutation. This paper presents a systematic review that includes an explanation of DNA sequences, gene mutations that occur in breast cancer, and bioinformatics techniques for detecting breast cancer”, on page 3, paragraph 3 “This paper presents a review of methods for detecting gene mutations, specifically those related to breast cancer in the field of bioinformatics. The review included gene mutations related to breast cancer, as well as techniques in bioinformatics for detecting breast cancer which included the method of DNA mapping, feature extraction, and classification”, on page 7, paragraph 1 “The feature extraction phase aims to extract important information from DNA sequence data. The feature extraction method is needed because it can speed up computing time compared to conventional pattern matching methods”. and on page 9, paragraph 1 “The methods that can be used are starting from regression approaches, probability models (Bayesian and Hidden Markov Models), Support Vector Machines, Neural Networks and Deep Learning”, reading on a method implemented by one or more computing devices, the method comprising: obtaining genetic data to be processed; inputting the genetic data to be processed into a feature generation network layer for performing a feature extraction operation to obtain genetic features corresponding to the genetic data to be processed and enhanced features corresponding to the genetic features; and inputting the genetic data to be processed and the enhanced features into a genetic identification network layer for performing a genetic testing operation to obtain a testing result.
Liu et al. teaches on page 856, paragraph 1 “Moreover, the strategy adopted for decision making is also profound. The hard cut-off strategy for variant selection is mainly based on allele frequency or read depth, which could omit rare genetic variants below the predefined threshold representing pathogenic meaning. Thus, the variant prioritization strategy could be utilized for ranking in order the variants regarding their potential clinical impact”, reading on wherein an average number of genetic segments corresponding to each position in the genetic data to be processed is less than or equal to a preset threshold.
It would have been obvious at the time of first filing to have modified the teachings of Wisesty et al. for the method of claim 1 and apparatus of claim 15 with the specifics of a feature generation layer and a genetic identification layer as the former is performing the latter using a neural network, and would therefore be implied to be using such layers. One would have had a reasonable expectation of success given that the methods are the same, the data is the same, and the output is the same, only the architecture of the network is more specified within the claims. Furthermore, it would have been obvious at the time of first filing to have modified the teachings of Wisesty et al. for the CRM of claim 7, with the teachings of Liu et al. for the use of genetic position thresholds as the latter teaches in the same paragraph “the variant prioritization strategy could be utilized for ranking in order the variants regarding their potential clinical impact”. One would have had a reasonable expectation of success given that this is merely a preprocessing step not affecting the implementation of the neural network itself. Therefore, it would have been obvious at the time of first filing to have modified the teachings of each and to be successful.
Claim 2 is directed to the method of claim 1 but further specifies the use of the genetic data and features to generate testing reference data.
Claim 16 is directed to the apparatus of claim 15 but further specifies the use of the genetic data and features to generate testing reference data.
Wisesty et al. teaches in Table 1 a review of gene prediction/classification research in bioinformatics including “The data used is 131 data for training and 369 data to test system performance obtained from NCBI Genbank”, for cancerous gene detection using a neural network, reading on performing genetic testing processing on the genetic data to be processed and the enhanced features using the genetic identification network layer to obtain testing reference information corresponding to the genetic data to be processed.
Claim 4 is directed to the method of claim 2 but further specifies obtaining testing result according to the testing reference information.
Claim 18 is directed to the apparatus of claim 15 but further specifies obtaining testing result according to the testing reference information.
Wisesty et al. teaches on page 2, paragraph 4 “This paper presents a systematic review of breast cancer detection based on DNA sequence data”, and in Table 1 “DNA sequence data is first converted into numerical vectors by replacing A, T, G, and C with 1, -1, j, and -j. Then, the vector is processed using DWT, statistical feature extraction which includes mean, median, standard deviation, range, mean absolute deviation, and mean absolute deviation. By using SVM as a classification method of characteristic statistic features, the study can achieve 100% accuracy for lung, ovarian and breast cancer data” and “The research aims to classify normal genes and cancer genes. The data used is 131 data for training and 369 data to test system performance obtained from NCBI Genbank. The data is converted using EIIP mapping where nucleotide A is converted to 0.1260, C becomes 0.1340, T becomes 0.1335, G becomes 0.0806. The FLANN architecture used uses the trigonometry expansion technique where the input data is expanded with the sin and cos functions. The output signal from FLANN is then calculated Normalized Mean Square Error (NMSE) and classified into three classes namely Healthy (NMSE <0.1), Equiprobable (0.1 <= NMSE <= 0.15), and Cancer (NMSE> 0.15)”, reading on obtaining the testing result corresponding to the genetic data to be processed according to the testing reference information.
Claim 7 is directed to a CRM that obtains genetics, determines features and performs learning and training.
Wisesty et al. teaches on page 2, paragraph 4 “This paper presents a systematic review of breast cancer detection based on DNA sequence data”, and in Table 1 “DNA sequence data is first converted into numerical vectors by replacing A, T, G, and C with 1, -1, j, and -j. Then, the vector is processed using DWT, statistical feature extraction which includes mean, median, standard deviation, range, mean absolute deviation, and mean absolute deviation. By using SVM as a classification method of characteristic statistic features, the study can achieve 100% accuracy for lung, ovarian and breast cancer data” and “The research aims to classify normal genes and cancer genes. The data used is 131 data for training and 369 data to test system performance obtained from NCBI Genbank. The data is converted using EIIP mapping where nucleotide A is converted to 0.1260, C becomes 0.1340, T becomes 0.1335, G becomes 0.0806. The FLANN architecture used uses the trigonometry expansion technique where the input data is expanded with the sin and cos functions. The output signal from FLANN is then calculated Normalized Mean Square Error (NMSE) and classified into three classes namely Healthy (NMSE <0.1), Equiprobable (0.1 <= NMSE <= 0.15), and Cancer (NMSE> 0.15)”, reading on obtaining genetic samples, where the genetic samples correspond to sample mutation results; determining genetic features corresponding to the genetic samples and enhanced features corresponding to the genetic features; and performing learning and training based on reference genetic results, the genetic features, and the enhanced features corresponding to the genetic samples to obtain a genetic testing model, wherein the genetic testing model is configured to perform a feature extraction operation on genetic data and perform a testing operation on the genetic data based on extracted features.
Liu et al. teaches on page 856, paragraph 1 “Moreover, the strategy adopted for decision making is also profound. The hard cut-off strategy for variant selection is mainly based on allele frequency or read depth, which could omit rare genetic variants below the predefined threshold representing pathogenic meaning. Thus, the variant prioritization strategy could be utilized for ranking in order the variants regarding their potential clinical impact”, reading on and an average number of genetic segments corresponding to each position in the genetic samples is less than or equal to a preset threshold.
Claim 8 is directed to the CRM of claim 7 but further specifies performing learning and training on a feature generation sub-model and on the features and reference to obtain a variant identification model.
Wisesty et al. teaches on page 2, paragraph 4 “This paper presents a systematic review of breast cancer detection based on DNA sequence data”, and in Table 1 “DNA sequence data is first converted into numerical vectors by replacing A, T, G, and C with 1, -1, j, and -j. Then, the vector is processed using DWT, statistical feature extraction which includes mean, median, standard deviation, range, mean absolute deviation, and mean absolute deviation. By using SVM as a classification method of characteristic statistic features, the study can achieve 100% accuracy for lung, ovarian and breast cancer data” and “The research aims to classify normal genes and cancer genes. The data used is 131 data for training and 369 data to test system performance obtained from NCBI Genbank. The data is converted using EIIP mapping where nucleotide A is converted to 0.1260, C becomes 0.1340, T becomes 0.1335, G becomes 0.0806. The FLANN architecture used uses the trigonometry expansion technique where the input data is expanded with the sin and cos functions. The output signal from FLANN is then calculated Normalized Mean Square Error (NMSE) and classified into three classes namely Healthy (NMSE <0.1), Equiprobable (0.1 <= NMSE <= 0.15), and Cancer (NMSE> 0.15)”, and “The research aims to detect Single Nucleotide Variants (SNVs) which can show gene mutations in cancer cells. The study uses data with low sequencing depth, where if the detection process is carried out with the previous algorithm it still has low performance. The author also conducted experiments using SNVMix, and it was proven that SNVHMM had a higher performance with an f-score of 0.85. SNVHMM can reduce false positives and increase true negatives and requires less training data compared to SNVMix”, reading on performing learning and training based on the genetic samples, the genetic features and the enhanced features to obtain a feature generation sub-model, wherein the feature generation sub-model is used for performing feature extraction and enhancing extracted genetic features; performing learning and training based on the enhanced features and the reference genetic results corresponding to the genetic samples to obtain a variant identification model, wherein the variant identification model is used for testing genetic data based on feature information; and generating the genetic testing model based on the feature generation sub-model and the variant identification model.
Claim 14 is directed to the CRM of claim 7 but further specifies the average number of genetic segments corresponding to each position being less than or equal to a specified threshold.
Liu et al. teaches on page 856, paragraph 1 “Moreover, the strategy adopted for decision making is also profound. The hard cut-off strategy for variant selection is mainly based on allele frequency or read depth, which could omit rare genetic variants below the predefined threshold representing pathogenic meaning. Thus, the variant prioritization strategy could be utilized for ranking in order the variants regarding their potential clinical impact”, reading on obtaining genetic data to be processed, wherein an average number of genetic segments corresponding to each position in the genetic data to be processed is less than or equal to a preset threshold; determining the genetic testing model for analyzing and processing the genetic data to be processed; and analyzing and processing the genetic data to be processed using the genetic testing model to obtain a testing result.
Claims 3 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Wisesty et al. (IOP Conference Series: Materials Science and Engineering (2020) 1-11) and Liu et al. (Trends in genetics (2019) 852-867) as applied to claims 1-2, 4, 15-16, and 18 above, and further in view of Catic et al. (BMC Medical Genomics (2018) 1-12).
Claim 3 is directed to the method of claim 2 but further specifies the information includes one of the specified data types.
Claim 17 is directed to the apparatus of claim 15 but further specifies the information includes one of the specified data types.
Wisesty et al. and Lui et al. teach the method of claim 1 and the apparatus of claim 15 as previously described.
Wisesty et al. and Lui et al. do not teach that the information includes one of the specified data types.
Catic et al. teaches in the abstract “The usage of Artificial Neural Networks (ANNs) for genome-enabled classifications and establishing genome-phenotype correlations have been investigated more extensively over the past few years. The reason for this is that ANNs are good approximates of complex functions, so classification can be performed without the need for explicitly defined input-output model. This engineering tool can be applied for optimization of existing methods for disease/syndrome classification. Cytogenetic and molecular analyses are the most frequent tests used in prenatal diagnostic for the early detection of Turner, Klinefelter, Patau, Edwards and Down syndrome. Feedback neural network architecture out performed feedforward neural network architecture in predictive ability for all five aneuploidy prenatal syndrome classes. Feedforward neural network with 15 neurons in hidden layer achieved classification sensitivity of 92.00%. Classification sensitivity of feedback (Elman’s) neural network was 99.00%. Average accuracy of feedforward neural network was 89.6% and for feedback was 98.8%”, reading on wherein the testing reference information includes at least one of: 21-type genotype prediction information, zygotic prediction information, first allelic mutation length information, and second allelic mutation length information.
It would have been obvious at the time of first filing to have modified the teachings of Wisesty et al. and Lui et al. for the method and apparatus of claim 15 with the teachings of Catic et al. for the application of networks to the classification of trisomy 21 diseases as the latter teaches in the abstract “This engineering tool can be applied for optimization of existing methods for disease/syndrome classification” and “Feedforward neural network with 15 neurons in hidden layer achieved classification sensitivity of 92.00%. Classification sensitivity of feedback (Elman’s) neural network was 99.00%. Average accuracy of feedforward neural network was 89.6% and for feedback was 98.8%”. One would have had a reasonable expectation of success given that it is merely the predicted outcome that is changing not the data itself nor the method, and the latter citation specifically says that their method can be used with any existing methods. Therefore, it would have been obvious at the time of first filing to have modified the teachings of each and to be successful.
Claims 5-6, 9-11, and 19-20 are rejected under 35 U.S.C. 103 as being unpatentable over Wisesty et al. (IOP Conference Series: Materials Science and Engineering (2020) 1-11) and Liu et al. (Trends in genetics (2019) 852-867) as applied to claims 1-2, 4, 7-8, 15-16, and 18 above, and further in view of Janocha et al. (arXiv preprint (2017) 1-10).
Claim 5 is directed to the method of claim 1 but further specifies selecting a loss function and using said function to optimize the network.
Claim 19 is directed to the apparatus of claim 15 but further specifies selecting a loss function and using said function to optimize the network.
Wisesty et al. and Lui et al. teach the method of claim 1 and the apparatus of claim 15 as previously described.
Wisesty et al. and Lui et al. do not teach the selection and use of a loss function to optimize the neural network.
Janocha et al. teaches in the abstract “Deep neural networks are currently among the most commonly used classifiers. Despite easily achieving very good performance, one of the best-selling points of these models is their modular design {one can conveniently adapt their architecture to specific needs, change connectivity patterns, attach specialized layers, experiment with a large amount of activation functions, normalization schemes and many others… In this paper we try to investigate how particular choices of loss functions affect deep models and their learning dynamics, as well as resulting classifiers robustness to various effects…we show that L1 and L2 losses are, quite surprisingly, justified classification objectives for deep nets, by providing probabilistic interpretation in terms of expected misclassification”, and on page 8, paragraph 2 “This paper provides basic analysis of effects the choice of the classification loss function has on deep neural networks training as well as their final characteristics. We believe the obtained results will lead to a wider adoption of various losses in DL work {where up till now log loss is unquestionable favorite”, reading on obtaining a standard data type corresponding to the genetic data to be processed; inputting the genetic features to a data identification network layer for performing a data type identification operation to obtain a genetic data type; determining a loss function used for the feature generation network layer based on the genetic data type and the standard data type; and optimizing the feature generation network layer using the loss function to obtain an optimized feature generation network layer.
It would have been obvious at the time of first filing to have modified the teachings of Wisesty et al. and Lui et al. for the method of claim 1 and apparatus of claim 15 with the teachings of Janocha et al. for the use of loss functions within a neural network architecture to optimize the network as the latter points out in the abstract “we show that L1 and L2 losses are, quite surprisingly, justified classification objectives for deep nets, by providing probabilistic interpretation in terms of expected misclassification” and on page 8, paragraph 2 “We believe the obtained results will lead to a wider adoption of various losses in DL work {where up till now log loss is unquestionable favorite”. One would have had a reasonable expectation of success given that optimization is an inherent part of neural network design and the latter reference provides an alternative means for optimization within network architectures. Therefore, it would have been obvious at the time of first filing to have modified the teachings of each and to be successful.
Claim 6 is directed to the method of claim 5 but further specifies optimizing the data identification layer using the loss function and using this optimization in the feature generation network layer.
Claim 20 is directed to the apparatus of claim 15 but further specifies optimizing the data identification layer using the loss function and using this optimization in the feature generation network layer.
Wisesty et al. teaches the method of claim 1 and the apparatus of claim 15 as previously described.
Wisesty et al. does not teach optimizing the data identification layer using the loss function and using this optimization in the feature generation network layer.
Janocha et al. teaches in the abstract “Deep neural networks are currently among the most commonly used classifiers. Despite easily achieving very good performance, one of the best-selling points of these models is their modular design {one can conveniently adapt their architecture to specific needs, change connectivity patterns, attach specialized layers, experiment with a large amount of activation functions, normalization schemes and many others… In this paper we try to investigate how particular choices of loss functions affect deep models and their learning dynamics, as well as resulting classifiers robustness to various effects. we show that L1 and L2 losses are, quite surprisingly, justified classification objectives for deep nets, by providing probabilistic interpretation in terms of expected misclassification”, and on page 8, paragraph 2 “This paper provides basic analysis of effects the choice of the classification loss function has on deep neural networks training as well as their final characteristics. We believe the obtained results will lead to a wider adoption of various losses in DL work {where up till now log loss is unquestionable favorite”, reading on wherein the feature generation network layer comprises a part of the data identification network layer, and optimizing the feature generation network layer using the loss function to obtain the optimized feature generation network layer comprises: optimizing the data identification network layer based on the loss function to obtain an optimized data identification network layer; and determining the optimized feature generation network layer based on the optimized data identification network layer.
Claim 9 is directed to the CRM of claim 8 and thus claim 7, but further specifies the obtaining of a data identification model based on reference and features using an optimized feature generation model.
Wisesty et al. teaches the method of claim 1, CRM of claim 7, and the apparatus of claim 15 as previously described.
Wisesty et al. does not teach the use of an optimized feature generation model.
Janocha et al. teaches in the abstract “Deep neural networks are currently among the most commonly used classifiers. Despite easily achieving very good performance, one of the best-selling points of these models is their modular design {one can conveniently adapt their architecture to specific needs, change connectivity patterns, attach specialized layers, experiment with a large amount of activation functions, normalization schemes and many others… In this paper we try to investigate how particular choices of loss functions affect deep models and their learning dynamics, as well as resulting classifiers robustness to various effects. we show that L1 and L2 losses are, quite surprisingly, justified classification objectives for deep nets, by providing probabilistic interpretation in terms of expected misclassification”, and on page 8, paragraph 2 “This paper provides basic analysis of effects the choice of the classification loss function has on deep neural networks training as well as their final characteristics. We believe the obtained results will lead to a wider adoption of various losses in DL work {where up till now log loss is unquestionable favorite”, reading on performing learning and training based on the genetic features and the reference genetic results corresponding to the genetic samples to obtain a data identification model, wherein the data identification model is used for performing a variant identification operation on genetic data based on genetic features; and optimizing the feature generation sub-model using the data identification model to obtain an optimized feature generation sub-model.
Claim 10 is directed to the CRM of claim 9 and thus claim 7, but further specifies obtaining a loss function for optimizing various models, and optimizing them.
Wisesty et al. teaches the method of claim 1, CRM of claim 7, and the apparatus of claim 15 as previously described.
Wisesty et al. does not teach the obtaining a loss function for optimizing various models, and optimizing them.
Janocha et al. teaches in the abstract “Deep neural networks are currently among the most commonly used classifiers. Despite easily achieving very good performance, one of the best-selling points of these models is their modular design {one can conveniently adapt their architecture to specific needs, change connectivity patterns, attach specialized layers, experiment with a large amount of activation functions, normalization schemes and many others… In this paper we try to investigate how particular choices of loss functions affect deep models and their learning dynamics, as well as resulting classifiers robustness to various effects. we show that L1 and L2 losses are, quite surprisingly, justified classification objectives for deep nets, by providing probabilistic interpretation in terms of expected misclassification”, and on page 8, paragraph 2 “This paper provides basic analysis of effects the choice of the classification loss function has on deep neural networks training as well as their final characteristics. We believe the obtained results will lead to a wider adoption of various losses in DL work {where up till now log loss is unquestionable favorite”, reading on wherein the feature generation sub-model comprises a part of the data identification model, and optimizing the feature generation sub-model using the data identification model to obtain the optimized feature generation sub-model comprises: obtaining a loss function used for optimizing the data identification model; optimizing the data identification model based on the loss function to obtain an optimized data identification model; and determining the optimized feature generation sub-model based on the optimized data identification model.
Claim 11 is directed to the CRM of claim 10 and thus claim 7, but further specifies using features to predict results corresponding to features and using the loss function to optimize the model based on the features.
Wisesty et al. teaches the method of claim 1, CRM of claim 7, and the apparatus of claim 15 as previously described.
Wisesty et al. does not teach the use of a loss function to optimize the model based on the features.
Janocha et al. teaches in the abstract “Deep neural networks are currently among the most commonly used classifiers. Despite easily achieving very good performance, one of the best-selling points of these models is their modular design {one can conveniently adapt their architecture to specific needs, change connectivity patterns, attach specialized layers, experiment with a large amount of activation functions, normalization schemes and many others… In this paper we try to investigate how particular choices of loss functions affect deep models and their learning dynamics, as well as resulting classifiers robustness to various effects. we show that L1 and L2 losses are, quite surprisingly, justified classification objectives for deep nets, by providing probabilistic interpretation in terms of expected misclassification”, and on page 8, paragraph 2 “This paper provides basic analysis of effects the choice of the classification loss function has on deep neural networks training as well as their final characteristics. We believe the obtained results will lead to a wider adoption of various losses in DL work {where up till now log loss is unquestionable favorite”, reading on analyzing and processing the genetic features using the data identification model to obtain predicted genetic results corresponding to the genetic features; and determining the loss function used for optimizing the data identification model based on the genetic features, the predicted genetic results, and the reference genetic results.
Claims 12 and 13 are rejected under 35 U.S.C. 103 as being unpatentable over Wisesty et al. (IOP Conference Series: Materials Science and Engineering (2020) 1-11) and Liu et al. (Trends in genetics (2019) 852-867) as applied to claims 1-2, 4, 7-8, 15-16, and 18 above, and further in view of Tzeng et al. (Proceedings of the IEEE conference on computer vision and pattern recognition (2017) 1-10), and Janocha et al. (arXiv preprint (2017) 1-10).
Claim 12 is directed to the CRM of claim 8 and thus claim 7, but further specifies use of an average number of genetic segments corresponding to each position being less than or equal to a specified threshold and use of an adversarial discriminative model that is used in optimizing the sub-model.
Wisesty et al. and Lui et al. teach the method of claim 1, CRM of claim 7, and the apparatus of claim 15 as previously described.
Wisesty et al. and Lui et al. do not teach the use of an adversarial discriminative model that is used in optimizing the sub-model.
Liu et al. teaches on page 856, paragraph 1 “Moreover, the strategy adopted for decision making is also profound. The hard cut-off strategy for variant selection is mainly based on allele frequency or read depth, which could omit rare genetic variants below the predefined threshold representing pathogenic meaning. Thus, the variant prioritization strategy could be utilized for ranking in order the variants regarding their potential clinical impact”, reading on obtaining reference features for performing analysis processing on the enhanced features, wherein an average number of genetic segments corresponding to each position in the reference features is greater than the preset threshold.
Tzeng et al. teaches in the abstract “Adversarial learning methods are a promising approach to training robust deep networks, and can generate complex samples across diverse domains. They also can improve recognition despite the presence of domain shift or dataset bias…We first outline a novel generalized framework for adversarial adaptation, which subsumes recent state-of-the-art approaches as special cases, and we use this generalized view to better relate the prior approaches. We propose a previously unexplored instance of our general framework which combines discriminative modeling, untied weight sharing, and a GAN loss, which we call Adversarial Discriminative Domain Adaptation (ADDA). We show that ADDA is more effective yet considerably simpler than competing domain-adversarial methods, and demonstrate the promise of our approach byexceeding state-of-the-art unsupervised adaptation results on standard cross-domain digit classification tasks and a new more difficult cross-modality object classification task”, and on page 4, column 2, paragraph 4 “Once we have decided on a parametrization of Mt, we employ an adversarial loss to learn the actual mapping”, reading on performing learning and training based on the reference features and the enhanced features to obtain an adversarial discriminative model, wherein the adversarial discriminative model is configured to perform a discriminative operation on the genetic features.
Janocha et al. teaches in the abstract “Deep neural networks are currently among the most commonly used classifiers. Despite easily achieving very good performance, one of the best-selling points of these models is their modular design {one can conveniently adapt their architecture to specific needs, change connectivity patterns, attach specialized layers, experiment with a large amount of activation functions, normalization schemes and many others… In this paper we try to investigate how particular choices of loss functions affect deep models and their learning dynamics, as well as resulting classifiers robustness to various effects. we show that L1 and L2 losses are, quite surprisingly, justified classification objectives for deep nets, by providing probabilistic interpretation in terms of expected misclassification”, and on page 8, paragraph 2 “This paper provides basic analysis of effects the choice of the classification loss function has on deep neural networks training as well as their final characteristics. We believe the obtained results will lead to a wider adoption of various losses in DL work {where up till now log loss is unquestionable favorite”, reading on optimizing the feature generation sub-model using the adversarial discriminative model to obtain an optimized feature generation sub-model.
It would have been obvious at the time of first filing to have modified the teachings of Wisesty et al. and Lui et al. for the CRM of claims 7 and 8 with the teachings of Janocha et al. for the use of loss functions within a neural network architecture to optimize the network as the latter points out in the abstract “we show that L1 and L2 losses are, quite surprisingly, justified classification objectives for deep nets, by providing probabilistic interpretation in terms of expected misclassification” and on page 8, paragraph 2 “We believe the obtained results will lead to a wider adoption of various losses in DL work {where up till now log loss is unquestionable favorite”. One would have had a reasonable expectation of success given that optimization is an inherent part of neural network design and the latter reference provides an alternative means for optimization within network architectures. Finially, it would have been obvious at the time of first filing to have modified the previous teachings with those of Tzeng et al. for the use of adversarial discriminative models as the latter teaches in the abstract “We show that ADDA is more effective yet considerably simpler than competing domain-adversarial methods, and demonstrate the promise of our approach by exceeding state-of-the-art unsupervised adaptation results on standard cross-domain digit classification tasks and a new more difficult cross-modality object classification task”. One would have had a reasonable expectation of success given that the latter teaches the use of such models as an approach to the training of robust networks. Therefore, it would have been obvious at the time of first filing to have modified the teachings of each and to be successful.
Claim 13 is directed to the CRM of claim 12 and thus claim 7, but further specifies optimizing the adversarial model based on the results.
Wisesty et al. teaches the method of claim 1, CRM of claim 7, and the apparatus of claim 15 as previously described.
Wisesty et al. does not teach optimizing the adversarial model based on the results.
Tzeng et al. teaches on page 8, paragraph 2 “This paper provides basic analysis of effects the choice of the classification loss function has on deep neural networks training as well as their final characteristics. We believe the obtained results will lead to a wider adoption of various losses in DL work {where up till now log loss is unquestionable favorite”.
Janocha et al. teaches in the abstract “Deep neural networks are currently among the most commonly used classifiers. Despite easily achieving very good performance, one of the best-selling points of these models is their modular design {one can conveniently adapt their architecture to specific needs, change connectivity patterns, attach specialized layers, experiment with a large amount of activation functions, normalization schemes and many others… In this paper we try to investigate how particular choices of loss functions affect deep models and their learning dynamics, as well as resulting classifiers robustness to various effects. we show that L1 and L2 losses are, quite surprisingly, justified classification objectives for deep nets, by providing probabilistic interpretation in terms of expected misclassification”, and on page 8, paragraph 2 “This paper provides basic analysis of effects the choice of the classification loss function has on deep neural networks training as well as their final characteristics. We believe the obtained results will lead to a wider adoption of various losses in DL work {where up till now log loss is unquestionable favorite”, which in view of Tzeng et al. reads on obtaining a judgment and identification result of analyzing and processing the enhanced features using the adversarial discriminative model; and optimizing the feature generation sub-model based on the judgment and identification result to obtain the optimized feature generation sub-model.
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
/OLIVIA M. WISE/Supervisory Patent Examiner, Art Unit 1685