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 .
This office action is in response to submission of application on 3/28/2023.
Claims 1-20 are presented for examination.
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 an abstract idea without significantly more.
Step 1: Is the claim to a process, machine, manufacture or composition of matter?
Claims 1-8 are directed to a method, claims 9-16 are directed to a system, and claims 17-20 are directed to a computer program product; therefore, all claims are directed to one of the four statutory categories.
Step 2A Prong One: Does the claim recite an abstract idea, law of nature, or natural
phenomenon?
Claim 1 recites limitations of:
assigning a sampling weight to each source distribution; - mental process (observation, evaluation, judgement) as a human mind is able to designate weights to a source distribution.
computing a sample based maximum mean discrepancy (MMD) measure between the target dataset and the mixed dataset; - mathematical concept (relationships, formulas or equations, calculations) of calculating a maximum mean discrepancy
determining that the target dataset is in a convex hull of the plurality of source datasets when the MMD measure is less than or equal to a threshold; - mental process (observation, evaluation, judgement) as a human mind is able to observe where the MMD is less than or equal to a threshold or not
determining that the target dataset is not in the convex hull of the plurality of source datasets when the MMD measure is greater than the threshold. - mental process (observation, evaluation, judgement) as a human mind is able to observe where the MMD is greater than a threshold or not
Step 2A Prong Two: Does the claim recite addition elements that integrate the judicial
exception into a practical application? Claim 1 recites limitations of:
obtaining a target dataset drawn from an unknown target distribution and a plurality of source datasets, wherein each source dataset is drawn from an unknown source distribution; - obtaining a target dataset merely amount to data gathering which is insignificant extra-solution activity. See MPEP § 2106.05(g), item (3), which identifies necessary data gathering and outputting as an example of extra-solution activity.
constructing a mixed dataset comprising a plurality of samples drawn from source distributions according to the sampling weights of the source distributions; - constructing a mixed dataset merely amounts to data gathering which is insignificant extra-solution activity. See MPEP § 2106.05(g), item (3), which identifies necessary data gathering and outputting as an example of extra-solution activity.
Step 2B: Does the claim recite additional elements that amounts to significantly more than the
judicial exception?
The additional elements are:
obtaining a target dataset drawn from an unknown target distribution and a plurality of source datasets, wherein each source dataset is drawn from an unknown source distribution; - obtaining a target dataset merely amount to data gathering which is insignificant extra-solution activity. See MPEP § 2106.05(g). Data gathering is well-understood, routine, and conventional. See MPEP 2106.05(d)(II)(iv).
constructing a mixed dataset comprising a plurality of samples drawn from source distributions according to the sampling weights of the source distributions; - constructing a mixed dataset merely amount to data gathering which is insignificant extra-solution activity. See MPEP § 2106.05(g). Data gathering is well-understood, routine, and conventional. See MPEP 2106.05(d)(II)(iv).
The additional elements do not amount to significantly more than the abstract idea. Therefore,
the claim is not patent eligible.
Independent claim 9 and 17 recites the same relevant limitations and a similar analysis applies.
Claim 9 recites the additional elements of, “A system comprising memory for storing instructions, and a processor configured to execute the instructions to:” - components recited at a high level are
construed as generic computer components used to implement the abstract idea. See MPEP
2106.05(f)(2). Claim 17 recites the additional elements of, “A computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor of a system to cause the system to:” They do not integrate the abstract idea into a practical application. Nor do they amount to significantly more. Therefore, the independent claims are not patent eligible.
The above analysis similarly applies to the dependent claims.
Dependent claim 2, 10, and 18 recites, where the source distribution is chosen with a probability proportional to an optimal sampling weight of the source distribution. - mathematical concept (relationships, formulas or equations, calculations) for using probability to choose a source distribution.
Dependent claim 3, 11, and 19 recites, wherein the sampling weight is initially uniform on each source dataset. - mental process (observation, evaluation, judgement) as a human mind is able to observe that the source data is uniform.
Dependent claim 4 and 12 recites, executing a mirror descent optimization algorithm to improve the sampling weight for minimizing a loss function. - mathematical concept (relationships, formulas or equations, calculations) for performing a mirror descent optimization algorithm.
Dependent claim 5 and 13 recites, wherein the mirror descent optimization algorithm uses a Kullback-Leibler (KL) divergence. - mathematical concept (relationships, formulas or equations, calculations) for using Kullback-Leibler (KL) divergence.
Dependent claim 6, 14, and 20 recites, wherein the sampling weight is computed by minimizing a squared norm of a difference between a mean embeddings of a candidate mixed distribution and the unknown target distribution. - mathematical concept (relationships, formulas or equations, calculations) for computing the sampling weight
Dependent claim 7 and 15 recites, further comprising determining that a data shift is a target data shift when the target dataset is in the convex hull of the plurality of source datasets. - mental process (observation, evaluation, judgement) as a human mind is able to observe that the data shift is a target data shift
Dependent claim 8 and 16 recites, further comprising determining that a data shift is a covariate data shift when the target dataset is not in the convex hull of the plurality of source datasets. - mental process (observation, evaluation, judgement) as a human mind is able to observe that the data shift is a covariate data shift.
The dependent claims do not integrate the abstract idea into a practical application, nor do they amount to significantly more than the abstract idea.
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.
Claim(s) 1-3, 6-11, 14-20 are rejected under 35 U.S.C. 103 as being unpatentable over Stojanov et al. (Data-Driven Approach to Multiple-Source Domain Adaptation, herein Stojanov) in view of Gretton et al. (A Kernel Two-Sample Test, herein Gretton).
Regarding claim 1,
Stojanov teaches,
obtaining a target dataset drawn from an unknown target distribution and a plurality of source datasets, wherein each source dataset is drawn from an unknown source distribution; (Stojanov, page 2, section 1, "one labeled training dataset and one unlabeled test dataset (termed source and target domain), arising from two joint distributions PSX,Y and PTX,Y , respectively." note: obtaining a target dataset maps to one unlabeled test dataset (target domain), drawn from an unknown target distribution maps to arising from... PTX,Y, a plurality of source datasets maps to one labeled training dataset... source domain, and wherein each source dataset is drawn from an unknown source distribution maps to arising from two joint distributions PSX,Y)
assigning a sampling weight to each source distribution; (Stojanov, page 7, section 4,
“
PNG
media_image1.png
111
416
media_image1.png
Greyscale
” and page 4, section 4, “performing kernel mean embedding of the conditional distributions of the features given a class label c”, note: Stojanov defines the source distribution as PSX,Y. In this formula, [
PNG
media_image2.png
38
260
media_image2.png
Greyscale
], the source distribution is represented as [
PNG
media_image3.png
39
76
media_image3.png
Greyscale
] as a kernel mean embedding, and [
PNG
media_image4.png
36
20
media_image4.png
Greyscale
] is the weight. Therefore, each source distribution maps to [
PNG
media_image3.png
39
76
media_image3.png
Greyscale
], and weight maps to [
PNG
media_image4.png
36
20
media_image4.png
Greyscale
].)
constructing a mixed dataset comprising a plurality of samples drawn from source distributions according to the sampling weights of the source distributions; (Stojanov, page 5, section 3, "
PNG
media_image5.png
77
407
media_image5.png
Greyscale
”, note: The formula expresses that for each source class c, take the source's data and multiply it by the assigned weight and then add all of those weighted contributions together. The result is a new dataset that is a blend of the sources, where each source is proportional to its weight. mixed dataset maps to this.)
computing a sample based maximum mean discrepancy (MMD) measure between the target dataset and the mixed dataset; (Stojanov, page 5, section 3, "
PNG
media_image6.png
129
404
media_image6.png
Greyscale
”, note: “
PNG
media_image7.png
27
78
media_image7.png
Greyscale
” target dataset maps to the first part of this equation, mixed dataset maps to the second part of this equation, maximum mean discrepancy maps to mean discrepancy.
Stojanov does not explicitly teach,
determining that the target dataset is in a convex hull of the plurality of source datasets when the MMD measure is less than or equal to a threshold;
and determining that the target dataset is not in the convex hull of the plurality of source datasets when the MMD measure is greater than the threshold.
Gretton teaches,
determining that the target dataset is in a convex hull of the plurality of source datasets when the MMD measure is less than or equal to a threshold; (Gretton, page 731, section 3, "This is achieved by comparing the test statistic5 MMD[F,X,Y] with a particular threshold: if the threshold is exceeded, then the test rejects the null hypothesis… The acceptance region of the test is thus defined as the set of real numbers below the threshold.", note: MMD is less than or equal to the threshold maps to acceptance region of the test is thus defined as the set of real numbers below the threshold, target is in the convex hull maps to the interpretation that if the mixed distribution closely matches the target (MMD is small), then the target can be expressed as a mixture of sources which would be inside the convex hull.)
and determining that the target dataset is not in the convex hull of the plurality of source datasets when the MMD measure is greater than the threshold. (Gretton, page 731, section 3, "This is achieved by comparing the test statistic5 MMD[F,X,Y] with a particular threshold: if the threshold is exceeded, then the test rejects the null hypothesis… The acceptance region of the test is thus defined as the set of real numbers below the threshold.", note: MMD is greater than a threshold maps to if the threshold is exceeded, then the test rejects the null hypothesis.)
It would be obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teaching of Gretton into Stojanov because Stojanov cites Gretton directly in his own paper as the source of MMD. Both papers use MMD to measure whether two distributions are the same or different. Stojanov sets up the problem and computes the MMD between a weighted mixture of sources and a target distribution, but never makes a formal decision from that value. Gretton provides the threshold test needed to turn that MMD value into a yes or no decision. This would assist in correctly identifying the type of distribution shift so that the appropriate correction can be applies, which would preserve the model accuracy.
Regarding claim 2,
The combination of Gretton and Stojanov teaches, The method of claim 1, wherein the mixed dataset comprises samples drawn independently from a source distribution, where the source distribution is chosen with a probability proportional to an optimal sampling weight of the source distribution. (Stojanov, page 4-5, section 3, "
PNG
media_image8.png
108
397
media_image8.png
Greyscale
”, note: This equation demonstrates the probabilistic mixture where Pnew(Y=c) would be the probability of selecting source c which is the weight, Pnew(X|y+c) is the distribution that is being drawn from once the source is selected. The summation over all of c means that each source is picked with equal probability equal to its weight and draw a sample from it independently.)
Regarding claim 3,
The combination of Gretton and Stojanov teaches, The method of claim 1, wherein the sampling weight is initially uniform on each source dataset. (Stojanov, page 8, section 4, "We then subsampled 250 points for each patient (domain), such that P(Y=1) varies uniformly between 0.2 and 0.8 across all patients (both sources and target) for each experiment." also see FIG. 4 and Table 2, as well as page 6, section 4, "
PNG
media_image9.png
83
410
media_image9.png
Greyscale
”)
Regarding claim 6,
The combination of Gretton and Stojanov teaches, The method of claim 3, wherein the sampling weight is computed by minimizing a squared norm of a difference between a mean embeddings of a candidate mixed distribution and the unknown target distribution. (Gretton, page 727, lemma 4, "
PNG
media_image10.png
32
194
media_image10.png
Greyscale
” and Stojanov, page 5, section 3, "
PNG
media_image11.png
60
403
media_image11.png
Greyscale
”, note: Gretton and Stojanov both show that MMD^2 is almost mathematically identical to the squared distance between the mean embeddings.)
Regarding claim 7,
The combination of Stojanov and Gretton teaches, The method of claim 1, further comprising determining that a data shift is a target data shift when the target dataset is in the convex hull of the plurality of source datasets. (Stojanov, page 2, section 1, "An alternative assumption to address this is the setting in which PY changes and PX|Y remains the same, a setting termed target shift”)
Regarding claim 8,
The combination of Stojanov and Gretton teaches, The method of claim 1, further comprising determining that a data shift is a covariate data shift when the target dataset is not in the convex hull of the plurality of source datasets. (Stojanov, page 2, section 1, "a large body of work has focused on the setting in which it is assumed that PX changes and PY|X remains the same… This setting is called covariate shift or sample selection bias")
Claims 9-11 and 14-16 corresponds to method claims 1-3, and 6-8. They are not patentably distinguishable as they only differ between a system and a method. Therefore, claims 9-11 and 14-16 are rejected for the same reasons as 1-3 and 6-8, respectively.
Claims 17-20 corresponds to method claims 1-3 and 6. They are not patentably distinguishable as they only differ between a machine and a method. Therefore, claims 17-20 are rejected for the same reasons as 1-3 and 6, respectively.
Claim(s) 4, 5, 12, and 13 are rejected under 35 U.S.C. 103 as being unpatentable over Stojanov et al. (Data-Driven Approach to Multiple-Source Domain Adaptation, herein Stojanov) in view of Gretton et al. (A Kernel Two-Sample Test, herein Gretton) and in further view of Amir Beck and Marc Teboulle (Mirror Descent and Nonlinear Projected Subgradient Methods for Convex Optimization, herein Beck).
Regarding claim 4,
The combination of Stojanov, Gretton, and Beck teaches, The method of claim 3, further comprising executing a mirror descent optimization algorithm to improve the sampling weight for minimizing a loss function. (Beck, page 169, section 2, "
PNG
media_image12.png
102
302
media_image12.png
Greyscale
”, note: this algorithm takes a step in the direction that reduces the loss, the projects the result back onto the simplex so the weights stay valid. Mirror descent optimization algorithm maps to MDA.)
It would be obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teaching of Stojanov, Gretton, and Beck because Beck demonstrates that mirror descent is a solution for keeping weights valid on a simplex. Therefore, it would enhance accuracy because if the weights go invalid during optimization the mixed dataset is wrong, which means the MMD value and the convex hull decision would be wrong. Therefore, this would ensure that it would produce correct and reliable results.
Regarding claim 5,
The combination of Stojanov, Gretton, and Beck teaches, The method of claim 4, wherein the mirror descent optimization algorithm uses a Kullback-Leibler (KL) divergence. (Beck, page 173, section 5, "
PNG
media_image13.png
85
300
media_image13.png
Greyscale
”, note: mirror descent is not one fixed algorithm and it has a family of algorithms which depends on which distance measure is plugged in. When the entropy function is used as the distance measure to be plugged in, the resulting Bregman distance formula would become the KL divergence.)
It would be obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teaching of Stojanov, Gretton, and Beck because once mirror descent is chosen for optimization, there still needs to be a distance measure to use within it. Beck demonstrates that KL divergence because it keeps weights positive throughout optimization which means that no source distribution ever gets a weight of zero and get eliminated from the mixture completely. If a weight hits zero, the mixed dataset loses a source, which hinders the MMD value and produces the wrong convex hull decision. Therefore, the KL divergence prevents this from happening, ensuring that the optimization produces the most accurate possible weights.
Claims 12 and 13 corresponds to method claims 4 and 5. They are not patentably distinguishable as they only differ between a system and a method. Therefore, claims 12 and 13 are rejected for the same reasons as 4 and 5, respectively.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to UYEN-NHU PHAM TRAN whose telephone number is (571)272-1559. The examiner can normally be reached Monday - Friday 7:30-5.
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, Miranda Huang can be reached at (571) 270-7092. 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.
/U.P.T./
Examiner, Art Unit 2124
/MIRANDA M HUANG/Supervisory Patent Examiner, Art Unit 2124