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
Applicant’s claim for the benefit of a prior-filed application under 35 U.S.C. 119(e) or under 35 U.S.C. 120, 121, 365(c), or 386(c) is partially acknowledged. Applicant has not complied with one or more conditions for receiving the benefit of an earlier filing date under 35 U.S.C. 119(e). Applicant’s priority claim to provisional application US 63/376837 is incorrectly listed in the foreign priority information section of the application data sheet (ADS) filed on 3/21/25; it should be listed in the domestic benefit section. Applicant can correct the issue by:
(a) Submitting a corrected ADS. The corrected ADS must be relative to the information of record using underline and/or strikethrough. See MPEP 211.02.I and MPEP 601.05(a); AND
(b) Submitting a request to correct the filing receipt (on a separate paper) and index that request using Doc code: CFILE which as a document type\description “Request for Corrected Filing Receipt.” See MPEP 601.05(a)(II).
Specification
The substitute specification is currently objected to as it states the present application claims priority to US Provisional Patent Application No. 63/376,837, which is inconsistent with current status of the application due to the defect of the ADS filed on 3/21/25 as noted above. Correcting the defect in the priority claim to the provisional application, as outlined above, will resolve this objection to the specification.
Note: all references to the specification in the instant office action is directed to the clean version of the substitute specification filed on 3/21/25.
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 20-39 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Claim 20 recites:
A computer-implemented method for anonymizing and transforming input data into signatures, the method comprising:
obtaining the input data;
transforming the input data into a subset D of an ambient distance space U;
generating a reference object, described by a dataset T, wherein the dataset T is a subset of the ambient distance space U;
generating, for each point x in the dataset D, a filter function fx on the reference object T;
selecting at least one probability distribution that defines geometrical aspects of the reference object T;
sampling, for each point x in the dataset D, a predetermined number of sample points from the dataset T according to probabilities given by the selected probability distribution(s) evaluated on the filter function fx;
selecting a clustering method as a base parameter for generating zero-degree homology stable ranks, wherein the clustering method groups the sampled points based on geometric relationships;
selecting a homology degree and, for each point x of the dataset D and for each probability distribution and clustering method, computing an average associated homology stable rank, thereby generating a sequence of stable ranks for each data point x; and
outputting the sequence of stable ranks as one or more anonymized signatures of the input data, wherein the one or more anonymized signatures retain geometric properties of the dataset D while being non-invertible and preventing reconstruction of the original data.
All of the steps, as drafted, is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of a generic computer. That is, other than reciting “A computer-implemented method,” nothing in the claim precludes the steps from practically being performed in the mind. For example, but for the limitation “computer-implemented method,” the steps “obtaining … transforming … generating … generating … selecting … selecting … outputting …” in the context of this claim encompass a user, with the aid of pen and paper, manually transforming input data to generate a sequence of stable ranks. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of a generic computer, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
This judicial exception is not integrated into a practical application. In particular, the additional limitation “a computer-implemented method” amounts to no more than mere instructions to apply the exception using a generic computer. Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea. The claim is directed to an abstract idea.
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to integration of the abstract idea into a practical application, the additional element of “A computer-implemented method” amounts to no more than mere instructions to apply the exception using a generic computer. Mere instructions to apply an exception using a generic computer cannot provide an inventive concept. The claim is not patent eligible.
Dependent claims 21-32 merely recite additional computational characteristics to generate the sequence of stable ranks. Hence, these claims only recite further elements of the mental step recited in claim 20. The reasons set forth for claim 20 are applicable to claims 21-32, and these claims are ineligible.
Dependent claim 33 merely recite storing the anonymized signature in a generic database, which is insignificant post-solution activity. Hence, claim 33 is ineligible.
Dependent claims 34-36 recite the use of a machine learning model to the anonymized signatures to detect patterns or correlations. Pattern matching and detecting correlations in data are mental processes, which are abstract ideas. Moreover, the use of a generic machine learning model as a tool to perform an abstract idea amounts to no more than mere instructions to apply the exception using a generic computer component. Hence, claims 34-36 are ineligible.
Dependent claims 37-39 merely recite comparing the anonymized signatures to other data for a variety of purposes or using the anonymized signatures to construct a data set, which are mental steps. Hence, claims 37-39 are ineligible.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 20-39 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention.
Claim 20 recites “wherein the one or more anonymized signature retain geometric properties of the dataset D while being non-invertible ...” The specification describes generating anonymized signatures such that the original input data cannot be re-identified (pg. 9, lines 25-27). However, the specification does not disclose the anonymized signatures are “non-invertible.”
Claim 37 recites using the anonymized signatures for fraud detection by comparing newly generated signatures against a database of known fraudulent data patterns. The specification discloses generally using the anonymized signatures for fraud detection (see pg. 6, lines 29), but there is no disclosure of comparing new signatures against a database of known fraudulent data patterns.
Claim 38 recites the anonymized signatures are compared to a historical dataset to identify trends in anonymized data over time. The specification discloses using stable ranks from different input data to generally compare trends with the different data (see pg. 18, lines 26-29), but there is no disclosure of identifying trends over time. E.g., the disclosed trends could be between different populations, regions, etc. Pg. 8, lines 23-26 describes an embodiment whereby the input data distributed in an array/matrix may be stored in a time series of a patient’s measurements, but there is no additional disclosure of comparing one or more of these time series measurements with anonymized signatures to identify trends.
Claim 39 recites using the anonymized signatures to construct privacy-preserving collaborative datasets across multiple institutions without requiring data sharing. The most relevant passage in the specification is found on page 9, lines 6-12 ("Unlike most other methods for de-identification and/or anonymization, the present method enables downstream processing (9) of the anonymized data set, for example the comparison of multiple anonymized datasets by a machine learning model, e.g. in order to identify similarities and/or overlaps between the outputs of the method. This may enable companies to collaborate with other entities using data they consider to be valuable proprietary information, by sharing only de-identified and/or anonymized information."). However, this passage only describes a potential capability of the generated signatures. There is no further description about how the signatures are used to generate a collaborative dataset across multiple institutions; e.g., what necessary information (meta data) needs to be shared or how the anonymized signatures need to be organized for multiple institutions to be able to use such anonymized information. Hence, it does not illustrate the inventor was in possession of a "privacy-preserving collaborative dataset across multiple institutions." Furthermore, nothing in the specification suggests that such anonymized signatures can be shared without requiring data sharing. On the contrary, the claim implies the data collected within the signatures themselves are data to be shared. As such, there is no support of the negative claim limitation “without requiring data sharing” in the specification.
The dependent claims inherit the defects of their parent claims and are rejected for the same reasons.
Claim Rejections - 35 USC § 102
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 the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
Claim(s) 20-32 and 34-36 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Agerberg et al. “Data, Geometry and Homology.” (hereinafter Agerberg)
As per claim 20, Agerberg discloses A computer-implemented method for anonymizing and transforming input data into signatures, the method comprising:
obtaining the input data (pg. 4, 3. Global Stable Ranks, “In this section we illustrate examples of global stable ranks for the MNIST dataset”; pg. 9, 5. Relative Stable Ranks on MNIST);
transforming the input data into a subset D of an ambient distance space U (Abstract, “In our framework the dataset serves as a reference object; we then consider different points in the ambient space and endow them with a geometry defined in relation to the reference object”; pg. 9, 5. Relative Stable Ranks on MNIST; pg. 10, 5.1 Illustration of the pipeline and first example. MNIST training and test set);
generating a reference object, described by a dataset T, wherein the dataset T is a subset of the ambient distance space U (pg. 3, 2. Pipeline, “The initial input is a finite subset R ⊂ Rr called a reference object.”);
generating, for each point x in the dataset D, a filter function fx on the reference object T (pg. 3, 2. Pipeline Step A1: filter function. “The objective is to obtain a function filter: R → R called a filter”);
selecting at least one probability distribution that defines geometrical aspects of the reference object T (pg. 3, 2. Pipeline, Step A2: distribution and probabilities. “In this step a distribution D needs to be chosen.”; pg. 4, 3. Global Stable Ranks, “The results of the pipeline described in Section 2, when the outcome of step A is given by the uniform probability function, are called global stable ranks of the reference object. These global stable ranks encode aspects of the geometry of the reference object captured by homologies of its s-element subspaces.”; pg. 7, 4. Relative Stable Ranks in the Plane, “The results of the pipeline described in Section 2, when the outcome of step A is given by the probability function determined by a point p, are called relative stable ranks of the reference object. We think about relative stable ranks as encoding geometrical information about the position of the point p in the ambient space Rr in relation to the reference object.”);
sampling, for each point x in the dataset D, a predetermined number of sample points from the dataset T according to probabilities given by the selected probability distribution(s) evaluated on the filter function fx (pg. 3, 2. PIPELINE Step B1: sub-sampling. “The probability function prob: R → R, obtained in step A, is used to sample the reference object R.”);
selecting a clustering method as a base parameter for generating zero-degree homology stable ranks, wherein the clustering method groups the sampled points based on geometric relationships (pg. 3, 2. PIPELINE Step B1: sub-sampling.; pg. 7, 4. Relative Stable Ranks in the Plane “The outcome of our pipeline in this case, for every point p in X, is a constant function 0 or 1. In this way the initial dataset X is partitioned into two clusters: points leading to the stable rank 0 and points leading to the stable rank 1.”; pg. 12, 5.2. Inside and outside. “we now perform k-means clustering (k=10) on the reference object and select the center of mass of each cluster.”; pg. 12, fig. 10);
selecting a homology degree (pg. 3, 2. Pipeline Step B2: stable ranks. “In this step four choices need to be made. First, is a natural number l called homological degree.”) and, for each point x of the dataset D and for each probability distribution and clustering method, computing an average associated homology stable rank, thereby generating a sequence of stable ranks for each data point x (pg. 3, 2. PIPELINE Step B: averaged stable ranks; Step B2: stable ranks.; pg. 4, 3. Global Stable Ranks, “Since there is a probabilistic step in our pipeline, the whole process is repeated 10 times to demonstrate stability of the outcome… We compute the average stable ranks corresponding to the training and test set respectively and present the distance between them, for each digit”; pg. 12, 5.2 Inside and outside, “These stable ranks are displayed in Figure 13 together with the average stable rank corresponding to a uniform subsampling of the reference object.”; the claim only requires one probability distribution to be selected); and
outputting the sequence of stable ranks as one or more anonymized signatures of the input data, wherein the one or more anonymized signatures retain geometric properties of the dataset D while being non-invertible and preventing reconstruction of the original data (pg. 1, 1. Introduction, classifying handwritten numbers from MNIST [handwritten numbers are disassociated from any individual]; pg. 2, 1. Introduction, “Extracting stable ranks is a simplifying procedure whereby a large amount of information is discarded. The challenge is to be able to steer some of the choices of the parameters which control stable ranks in such a way that some of the aspects relevant to the problem at hand are retained.” [e.g., resulting stable ranks do not preserve all visual characteristics of a handwritten digit, only a limited set of geometrical information]; pg. 6, 3. Global Stable Ranks, “In Section 5, we discuss a strategy of how to use the geometry of the spaces Traind, encoded through our pipeline, to classify handwritten digits.”).
As per claim 21, Agerberg discloses the method of claim 20. In addition, Agerberg discloses wherein the input data comprises biological data, medical data, insurance data, financial data, banking data, industrial data, advertising data, or personal data (pg. 1, Introduction, “Our purpose is to illustrate cases where homology-based invariants describing some of these relative geometrical aspects of points representing handwritten digits also contain a large amount of discriminative information.”; i.e., handwritten digits are personal data).
As per claim 22, Agerberg discloses the method of claim 20. In addition, Agerberg discloses wherein the reference object is an internal reference object that is a subset of the dataset D (pg. 2, 1. Introduction, “Global stable ranks encode some geometrical aspects of R.”; See also pg. 8, Figure 6, Reference Object: noisy circle of radius 3, which overlaps with point p in X).
As per claim 23, Agerberg discloses the method of claim 22. In addition, Agerberg discloses wherein the internal reference object is a subset of D corresponding to a predetermined type of category of points of the data (pg. 4, 3. Global Sable Ranks, Testd and Traind; see also, pg. 8, 4. Relative Stable Ranks in the Plane, Example 2, “noisy circle”).
As per claim 24, Agerberg discloses the method of claim 20. In addition, Agerberg discloses wherein the reference object is an external reference object that is a subset of the ambient distance space U and is not a subset of dataset D (pg. 7, 4. Relative Stable Ranks in the Plane, Example 1, Reference Object: single point).
As per claim 25, Agerberg discloses the method of claim 20. In addition, Agerberg discloses wherein the filter function fx on the reference object T produces a distance between a point in the reference object T and the data point x (pg. 3, 2. Pipeline Step A1: filter function. “In our particular construction of a filter, the input consists of a point p in Rr and a vector field on R represented by a function V: R → Rr. In this article the focus is on two types of a vector field: a constant vector field, and a vector field Vc: R → Rr determined by a point c, called center, in Rr which assigns to x in R the vector Vc(x) := c−x from x to c …. For example, if the consider vector field is given by Vp, the associated filter function assigns to x in R the distance between x and p.”; pg. 9, 5. Relative Stable Ranks on MNIST, “Following the steps defined in the pipeline, for a point under consideration p and for elements in the reference object x in R we choose as filter function fp(x) = ||p − x||2, i.e. the Euclidean distance between the point under consideration and the elements of the reference object.”).
As per claim 26, Agerberg discloses the method of claim 20. In addition, Agerberg discloses wherein the filter function fx is a geometrical projection along a vector field on the product between the dataset D and the reference object T (pg. 3, 2. Pipeline Step A1: filter function. “In our particular construction of a filter, the input consists of a point p in Rr and a vector field on R represented by a function V: R → Rr. In this article the focus is on two types of a vector field: a constant vector field, and a vector field Vc: R → Rr determined by a point c, called center, in Rr which assigns to x in R the vector Vc(x) := c−x from x to c.”); pg. 8, 4. Relative Stable Ranks in the Plane, Example 3).
As per claim 27, Agerberg discloses the method of claim 20. In addition, Agerberg discloses wherein the ambient distance space U is a space of n sequences of real or complex numbers R., wherein the distance is measured according to Euclidean, Chebyshev, Minkowski, or cosine distance (Abstract, “In our framework the dataset serves as a reference object; we then consider different points in the ambient space and endow them with a geometry defined in relation to the reference object”; pg. 3, line 4, “The initial input is a finite subset R ⊂ Rr called a reference object.”; pg. 8, Example 2, “Reference object: a noisy circle (of radius 3) represented by green dots in Figure 6”).
As per claim 28, Agerberg discloses the method of claim 27. In addition, Agerberg discloses wherein the ambient distance space U is a space of n sequences of real numbers Rn, using Euclidean distance (Abstract, “In our framework the dataset serves as a reference object; we then consider different points in the ambient space and endow them with a geometry defined in relation to the reference object”; pg. 3, 2. Pipeline, “The initial input is a finite subset R ⊂ Rr called a reference object.”; pg. 8, Example 2, “Reference object: a noisy circle (of radius 3) represented by green dots in Figure 6”).
As per claim 29, Agerberg discloses the method of claim 20. In addition, Agerberg discloses wherein the at least one probability distribution is selected from a uniform distribution, a uniform distribution restricted to a domain range, a normal distribution, or a piecewise linear function-based distribution (pg. 3, 2. Pipeline Step A: probabilities. “For example we could take the uniform probability which is the constant function with value 1/|R|.”).
As per claim 30, Agerberg discloses the method of claim 20. In addition, Agerberg discloses wherein each data point in the dataset D is encoded by its geometrical properties via the computed stable ranks (pg. 2, 1. Introduction, “Global stable ranks encode some geometrical aspects of R. Relative stable ranks encode some geometrical aspects of points in Rn relative to R.”).
As per claim 31, Agerberg discloses the method of claim 20. In addition, Agerberg discloses wherein the stable ranks generated for each data point are used for comparative analysis across multiple anonymized datasets (pg. 5, 3. Global Stable Ranks, “To further investigate whether the difference corresponds to a dataset shift or is due to random factors we pool the training and the test set together and perform random partitions. This is done 10 times for each digit, average stable ranks are then computed and the distance between the training and test sets resulting from these random partitions is compared to the distances obtained for the original training and test split.”; pg. 14. 5.3 Distinguishing out-of-sample points from two subsets of the reference object. “To quantify the capacity to discrimate [sic] between digits based on their stable ranks, we train a Support-vector machine classifier on the 20 stable ranks, for each homological degree, using the kernel obtained by taking inner products between stable ranks in the L2 function space [ARSC21]. We can then evaluate the model on the remaining samples of digit 1 and 7 from the MNIST test set”).
As per claim 32, Agerberg discloses the method of claim 20. In addition, Agerberg discloses wherein the number of sample points from dataset T selected for each point x of dataset D is dynamically adjusted based on statistical measures of dataset variation (pg. 3, 2. Pipeline Step A2: distribution and probabilities, “prob(x)” [determines how the filter values are turned into sampling probabilities]; pg. 9, 5. Relative Stable Ranks on MNIST, “As distribution we choose a Gaussian, whose parameters µp, σp are chosen in order to concentrate the probability mass on elements of the reference object close to p, yet ensuring the probability mass is distributed on sufficiently many elements for the samples to be diverse enough.”).
As per claim 34, Agerberg discloses the method of claim 20. In addition, Agerberg discloses the method further comprising applying a machine learning model to the anonymized signatures to detect patterns or correlations in the transformed data (pg. 3, 2. Pipeline “These functional representations of R are then used as inputs for various analysis pipelines such as SVMs.”).
As per claim 35, Agerberg discloses the method of claim 34. In addition, Agerberg discloses wherein the machine learning model applies a clustering algorithm to the anonymized signatures (pg. 3, 2. Pipeline “These functional representations of R are then used as inputs for various analysis pipelines such as SVMs.”; examiner’s note: SVMs separates data points into different classes, e.g., using a boundary)
As per claim 36, Agerberg discloses the method of claim 34. In addition, Agerberg discloses wherein the machine learning model applies a classification model trained on previously labeled anonymized signatures (pg. 3, 2. Pipeline “These functional representations of R are then used as inputs for various analysis pipelines such as SVMs.”; examiner’s note: SVMs separates data points into different classes, e.g., using a boundary; pg. 4, 3. Global Stable Ranks “Recall that MNIST is a dataset of handwritten digits widely used in machine learning, composed of 60000 training samples and 10000 test samples. The samples are considered as points in R784, since the images have 28×28 = 784 pixels. For every d in {0,1,...,9}, consider two reference objects Testd ⊂ R784 and Traind ⊂ R784 formed by these handwritten digits in respectively the test and the training sets of MNIST which are labeled by d.”).
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim 33 is rejected under 35 U.S.C. 103 as being unpatentable over Agerberg in view of Thomas et al. US 20210209870 (hereinafter Thomas ‘870).
As per claim 33, Agerberg discloses the method of claim 20. In addition, Agerberg discloses wherein the anonymized signatures are (pg. 3, 2. Pipeline “These functional representations of R are then used as inputs for various analysis pipelines such as SVMs.”). Although Agerberg does not expressly disclose storing the anonymized signatures in a database for future analysis, storing topological information in a database for future analysis would have been conventional to one of ordinary skill in the art. The storage of useful generated topological information in a database would enable such information to be analyzed at a later, indeterminate point in time. For example, Thomas ‘870 discloses generating topological representation of a machine performance event and storing the information as a reference model in a database for future comparisons. Para 0050 and 0060. It would have been obvious to one ordinary skill in the art before the effective filing date of the claimed invention such that the anonymized signatures are stored in a database for future comparison and analysis. One would have been motivated to do so to enable the signatures to be readily available for downstream use, such as an input to an SVM, at a later, indeterminate point in time, without recreating the signatures.
Although claims 37-39 are not rejected under 35 USC 102 nor 103, an indication of allowable subject matter is premature as there are outstanding written description issues as identified above in the 112a rejections.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JUNG W KIM whose telephone number is (571)272-3804. The examiner can normally be reached Monday-Friday, 10 a.m. - 6 p.m..
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, Amy Cohen Johnson can be reached at 571-272-2238. 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.
/JUNG W KIM/Supervisory Patent Examiner, Art Unit 2494
Prosecution Insights