DETAILED ACTION
Claims 1-8 are pending. This is in response to the application filed on May 15, 2024 which claims priority to the foreign application CHINA 202410263587.9 filed on March 7, 2024.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claim 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 1-8 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) without significantly more.
The rationale for this determination is explained below:
First – following Step 1 of the guidance, claims 1-8 are directed to a method comprising a series of functional steps applied in deep neural learning, or a system with storage device/memory, a physical device. Therefore, the claimed invention falls into one of the four statutory categories.
Secondly – following Step 2 of the guidance, claims 1-8 are analyzed for its underlying inventive concept with a new two-prong inquiry (1) does the claim recite an abstract idea, law of nature, or natural phenomenon, and/or judicial exceptions? And (2) does the claim recite additional elements that integrate the judicial exception into a practical application?
It is determined that claimed invention is directed to an abstract idea using deep neural learning to produce a model for adversarial attack detection, because the concept of the invention is constructing parameters using adversarial samples to fine-tune the detecting out-of-distribution (OOD) samples in order to distinguish the samples so disjointed from the in-distribution (ID) samples. As such, the claimed invention is directed to a series of steps similar to a mental process wherein concepts are performed in the human mind or by a human analyst (including an inputting, evaluation, judgment, opinion, etc.) with or without aid of computer. When giving broadest reasonable interpretations, a human analyst can produce deep learning model, although they may do so by using a computer or certain software, which are left out in the claims.
Regarding the second prone, the identified additional elements – a storage device and a device to perform the above steps – fail to integrate the idea of “calculating a plurality of distances between the plurality of in-distribution embeddings and the test embedding and selecting one of the plurality of distances as an out-of-distribution score of the test embedding, when the out-of-distribution score exceeds a threshold, classifying the test sample as out-of-distribution” into a practical application (e.g. any action is taken to prevent attacks). The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because the claim merely recites a storage device and a device to carry out the steps for calculation for an out-of-distribution score. These elements only perform functions of a general computer such as inputting and calculating data. Further, the claim does not recite an improvement to another technology or technical field, an improvement to the functioning of the computer itself, or meaningful limitations beyond generally linking the use of an abstract idea to a particular technological environment. Therefore, the claim is abstract without significantly more.
Dependent claims as presented, when analyzed individually or as a whole, are held to be patent ineligible under 35 U.S.C. 101 because, the additional recited limitation(s) fail(s) to amount to “significantly more” than the judicial exception, and thereby non-statutory.
Please see “The 2019 Revised Patent Subject Matter Eligibility Guidance (or “2019 PEG” for short) published in January 2019 at USPTO Website. Note that the groupings of abstract ideas in the 2019 PEG are not the same as those on the Abstract Ideas QRS or in the MPEP. The groupings in the 2019 PEG should be FOLLOWED for identifying abstract ideas. The 2019 PEG does not change the analysis at Step 2B which pertains to an improvement to conventional functioning of a computer or to technological processes; see also MPEP 2106.05(a).
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.
Claims 1-2 and 5-6 are rejected under 35 U.S.C. 103 as being unpatentable over Pub 20210034965 (hereinafter Tan) in view of Pub 20210150366 (hereinafter Ramachandran) and further in view of Pub CN-117436075-A (hereinafter Li)
Regarding to claim 1, Tan discloses using out-of-distribution detection method performed by a computing device comprising:
a training phase configured to train a detection model according to a plurality of in-distribution samples, wherein the training phase comprises a plurality of epochs (Fig. 2 related text discloses different datasets 210A-C with each dataset comprises different feature vectors 212A-C, etc.), and one of the plurality of epochs comprises:
Tan does not disclose adding a perturbation to each of the plurality of in-distribution samples to generate a plurality of adversarial samples. Ramachandran and Li disclose this feature. Ramachandran discloses adding noise to the in-domain training data to improve the detection of in-distribution and out-of-distributation (par. [0021] and [0095]). Although neither Tan nor Ramachandran discloses the samples are malware sample. Li discloses using unknown malicious software based on hypersphere embedding and distributed external sample detection (Fig. 1 and related text. Note that Li performs reversely assembling a malicious software PE file, collecting and splicing the obtained assembly ASM file API information to obtain the API characteristic vector; collecting and splicing the ASM file operation code sequence to obtain the operation code feature vector. Hence, one can think of the malicious PE file is a dataset 210 of Tan and Li teaches the API characteristic vector and obtain the operation code feature vector as the equivalent 212 feature vectors in Tan as branches). Therefore, it would have been obvious before the effective filing date of the claimed invention to modify Tan with Ramachandran and Li to further teach the aforementioned feature by using same neural network analysis for improvement of detecting in-distribution (ID) and out-of-distribution (OOD) as taught in Ramachandran and Li to arrive at the claimed invention with reasonable expectation of success.
inputting each of the plurality of adversarial samples into the detection model, wherein the detection model includes a plurality of branches; and calculating a loss function of each of the plurality of branches to optimize the detection model (Tan, Fig. 2 discloses different datasets as epochs and different feature vectors as branches while Li teaches using the API characteristic vector and obtain the operation code feature vector as the equivalent 212 feature vectors in Tan as branches with the improvement of ID and OOD detection); and
a testing phase comprising: inputting the plurality of in-distribution samples into the detection model to generate a plurality of in-distribution embeddings (again, both Tan and Li disclose using in-distribution embeddings);
inputting a test sample into the detection model to generate a test embedding; calculating a plurality of distances between the plurality of in-distribution embeddings and the test embedding and selecting one of the plurality of distances as an out-of-distribution score of the test embedding, when the out-of-distribution score exceeds a threshold, classifying the test sample as out-of-distribution (Tan, Figs. 1-3 and par. [0031]-[0054] disclose calculating a distance considered as an OOD. Note, Tan discloses modifying one or more parameter of the embedding network to improve the ID and OOD detection) .
Regarding to claim 2, Ramachandran discloses wherein adding the perturbation to each of the plurality of in-distribution samples to generate the plurality of adversarial samples comprises: adjusting a magnitude of the perturbation with jitter adversarial attack until the detection model misclassifies the adversarial samples (par. [0047] and [0051] discloses adding noise as a variance value for OOD identification. This means magnitude of noise introduced can degrade the detection).
Claims 5-6 are rejected in view of claims 1-2 rejections.
Claims 3-4 and 7-8 are rejected under 35 U.S.C. 103 as being unpatentable over Pub 20210034965 (hereinafter Tan) in view Ramachandran and Li and further in view of Pub 20220270155 (hereinafter Volkovs) and Pub WO-2022167774-A1 (hereinafter Sim)
Regarding to claim 3, Tan discloses using hyperparameter for tuning the loss calculation (par. [0021]-[0022]). However, Tan does not disclose wherein calculating the loss function of each of the plurality of branches to optimize the detection model comprises: reducing a sharpness of an overall loss function by Riemannian sharpness-aware minimization, wherein the overall loss function is a sum of the loss function of each of the plurality of branches and a cross-entropy loss. Volkovs discloses using hyperbolic with a Riemannian stochastic gradient in embeddings to produce a recommended score for a content item by training iterations until a reduction in the loss function across iterations is below a threshold or has reached a local minima (Fig. 6 and par. [0063]-[0068]). Volkovs does not discloses the loss function comprises a cross-entropy loss. Sim discloses this feature (par. [0027]-[0028]). Therefore, it would have been obvious before the effective filing date of the claimed invention to modify Tan, Ramachandran and Li with Volkovs and Sim to further teach the aforementioned feature by using hyperparameters to calculate the loss function comprising a cross-entropy loss for better false positive detection as taught in Volkovs and Sim to arrive at the claimed invention with reasonable expectation of success.
Regarding to claim 4, Li and Volkovs discloses wherein the plurality of branches comprises a hypersphere manifold (see Li for hypersphere embedding) and a hyperbolic manifold (Volkovs).
Claims 7-8 are rejected in view of claims 3-4 rejections.
Inquiry communication
Any inquiry concerning this communication or earlier communications from the examiner should be directed to TRI M TRAN whose telephone number is (571)270-1994. The examiner can normally be reached Mon-Fri: 9am-5pm.
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, Jeffrey Nickerson can be reached at (469)295-9235. 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.
/TRI M TRAN/Primary Examiner, Art Unit 2432