DETAILED ACTION
This Action is responsive to Claims filed 02/11/2026.
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 .
Status of the Claims
Claims 1-3, 5-6, 8-10, 12-13, 15-17, and 19 have been amended. Claim 20 is cancelled. Claim 21 is new. Claims 1-19 and 21 are currently pending.
Response to Arguments
Applicant’s arguments, see Page 8, filed 02/11/2026, with respect to claims 2, 3, 9, 10, 16, and 17 have been fully considered and are persuasive. The 35 U.S.C. 112(b) Rejection of claims 2, 3, 9, 10, 16, and 17 has been withdrawn.
Applicant’s arguments, see Page 8, filed 02/11/2026, with respect to claims 8-14 have been fully considered and are persuasive. The 35 U.S.C. 101 Rejection over non-statutory subject matter of claims 8-14 has been withdrawn.
Applicant’s arguments, see Pages 9-10, filed 02/11/2026, regarding the 35 U.S.C. 103 prior art Rejection(s) have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
In response to applicant's arguments against the references individually, one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986).
Galle teaches “An initial "clustering" of the data points is performed at S110. The algorithm starts by considering each point as a potential cluster. The number of clusters thus corresponds to the number of points, and each point is assigned to its own cluster. In other embodiments, fewer than all data points are assigned to a unique cluster, such that some clusters initially have more than one data point.” ([0050]), which the examiner submits continues to read on the “existing clusters” limitation. Given that Cali was not relied-upon for teaching clusters/clustering, the Examiner submits that if one of ordinary skill in the art were to combine Cali and Galle as recited by the office action (Pages 13-14), then a preliminary clustering of documents operated on by Cali in the fashion of Galle would necessarily create at least one authentic cluster and/or one or more counterfeit clusters.
Applicant's arguments, see Pages 10-11, filed 02/11/2026, regarding the 35 U.S.C. 101 Rejection of Claims 1-20 have been fully considered but they are not persuasive.
As presently drafted, the limitations of the independent claims are recited highly generally. A computer-implemented method is highly generic computing components. The “loading…” step, while interpreted in the context of the claim as a data input step, for example, lacks tangible structural detail or implementation, and could, at broadest, also recite a mental process step (a human mind is capable of generically “loading” a generic “embedding” into a generic “embedding space” with clusters). Because of this, even interpreting the limitation as data input, the limitation itself is highly general. The “determining…”, “comparing…”, and “responsive…” steps remain highly general, and recite abstract idea mental process steps with no structural detail or implementation differentiating the limitations merely from operations performed by a generic computer. Similar to the “loading…” step, the “reporting…” step is highly generic, and is only being interpreted as data output in the context of the claim. Given this, the Examiner submits the bulk of the alleged improvement is a direct result of the abstract idea mental process steps listed above. No improvement is expressly indicated in the “loading…” or “reporting…” steps. The paragraphs of the instant Specification cited in the arguments (34 and 35), recite generic axioms and, in fact, make reference to technician feedback in the analysis, itself a mental process not directly referenced in the claims. Per MPEP 2106.05(a), the specific improvement cannot come from the abstract idea(s). See the updated 35 U.S.C. 101 Rejection below.
Claim Objections
Claim 21 objected to because of the following informalities:
“generating a second global threshold for plurality…” should be “…for the plurality…”
Appropriate correction is required.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1-19 and 21 rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Based on the newly amended limitation “the set of outlier embeddings…”, it is unclear from the claim language (although the Examiner understands the intent based on the Examiner Interview), what the distinction is between a pre-existing set of counterfeit clusters, and an at least one preexisting outlier.
By the claim language there must be:
At least one authentic cluster at initialization
One or more counterfeit clusters at initialization
At least one outlier embedding determined to be an outlier, but not in a counterfeit cluster at initialization (in order to assign a subsequent outlier to a “set…including at least one other outlier…”)
It is unclear from the independent claims or instant Specification why the system would be initialized with an un-clustered outlier at initialization, rather than as an identified outlier cluster in and of itself. This also creates ambiguity regarding the size and number of counterfeit clusters within the system. If there are not sufficient outliers to breach the threshold as recited in subsequent claims, would only a single counterfeit cluster form? This goes against “a plurality” of counterfeit clusters. If the threshold is not reached at initialization, would all samples exist as unclustered outliers? This goes against preexisting counterfeit clusters. Further amendments and detail would improve clarity.
Claim Rejections - 35 USC § 101
The text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
Claims 1-19 and 21 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; and because the claims as a whole, considering all claim elements both individually and in combination, do not amount to significantly more than the abstract idea, see Alice Corporation Pty. Ltd. v. CLS Bank International, et al, 573 U.S. (2014). In determining whether the claims are subject matter eligible, the Examiner applies the 2019 USPTO Patent Eligibility Guidelines. (2019 Revised Patent Subject Matter Eligibility Guidance, 84 Fed. Reg. 50, Jan. 7, 2019.)
Step 1:
Claims 1-7 and 21 recite a computer-implemented method, which falls under the statutory category of a process. Claims 8-14 recite a computer program product comprising a computer-readable storage medium having a set of instructions stored therein which, which would fall under the statutory category of a manufacture, were claims 8-14 amended to exclude software or signals per se. Claims 15-19 recite a computer system comprising: a processor(s) set; and a computer readable storage medium having program instructions stored therein, which falls under the statutory category of a machine.
Step 2A – Prong 1:
Claim 1 recites an abstract idea, law of nature, or natural phenomenon. The limitations of “determining medoid distances between the first embedding and the respective medoids of the plurality of existing clusters;”, “comparing a first medoid distance to a corresponding cluster threshold of a first cluster;”, and “responsive to no medoid distance to any cluster being less than respectively corresponding cluster thresholds, assigning the first embedding to a set of outlier embeddings not matching any cluster, the set of outlier embeddings including at least one other outlier embedding;” under the broadest reasonable interpretation, cover a mental process including an observation, evaluation, judgment or opinion that could be performed in the human mind or with the aid of pencil and paper.
Determining medoid distances is practically performed within the human mind or with the aid of pen and paper. Comparing a value to a threshold is practically performed within the human mind or with the aid of pen and paper. Assigning an embedding to a cluster based on the comparison is practically performed within the human mind or with the aid of pen and paper.
Step 2A – Prong 2:
The additional elements of claim 1 do not integrate the abstract idea into a judicial exception. The claim recites the additional elements “A computer-implemented method” and “a first image”, which are recognized as generic computer components recited at a high level of generality. Although they have and execute instructions to perform the abstract idea itself, this also does not serve to integrate the abstract idea into a practical application as it merely amounts to instructions to "apply it." (See MPEP 2106.04(d)(2) indicating mere instructions to apply an abstract idea does not amount to integrating the abstract idea into a practical application).
The additional elements of “a first embedding”, “an embedding space”, “a plurality of existing clusters”, “a counterfeit class cluster”, and “a first medoid distance” are recognized as non-generic computer components, but are recited at a high level of generality and are found to generally link the abstract idea to a particular technological environment or field of use (See MPEP 2106.05(h)).
The additional elements recited in the limitations “loading a first embedding of a first image into an embedding space where there is a plurality of existing clusters of other image embeddings, the plurality of existing clusters having defined respective medoids and corresponding cluster thresholds, the plurality of existing clusters including at least one authentic class cluster and a plurality of counterfeit class clusters;” and “reporting the first image as representing a counterfeit resource.” Are found to be mere pre- or post-extra-solution or data transmittal steps (See MPEP 2106.05(g)).
Step 2B:
The additional elements of claim 1 do not amount to more than the judicial exception. The only limitation on the performance of the described method is a limitation reciting “A computer-implemented method” and “a first image” . These elements are insufficient to transform a judicial exception to a patentable invention because the recited elements are considered insignificant extra-solution activity (generic computer system, processing resources, links the judicial exception to a particular, respective, technological environment). The claim thus recites computing components only at a high-level of generality such that it amounts to no more than mere instructions to apply the exception using generic computer components; mere instructions to apply an exception using a generic computer component cannot provide an inventive concept (see MPEP 2106.05(f)).
The additional elements of claim 1 do not integrate the abstract idea into a judicial exception. The claim recites the additional elements “a first embedding”, “an embedding space”, “a plurality of existing clusters”, “a counterfeit class cluster”, and “a first medoid distance” are recognized as non-generic computer components, but are recited at a high level of generality and are found to generally link the abstract idea to a particular technological environment or field of use (See MPEP 2106.05(h)).
The additional elements recited in the limitations “loading a first embedding of a first image into an embedding space where there is a plurality of existing clusters of other image embeddings, the plurality of existing clusters having defined respective medoids and corresponding cluster thresholds, the plurality of existing clusters including at least one authentic class cluster and a plurality of counterfeit class clusters;” and “reporting the first image as representing a counterfeit resource.” are recognized as well-understood, routine, or conventional activity (See MPEP 2106.05(d)(II)(i) first list and (d)(II)(iv) third list, respectively).
Taken alone or in ordered combination, these additional elements do not amount to significantly more than the above-identified abstract idea. There is no indication that the combination of elements improves the functioning of a computer or improves any other technology. Their collective functions merely provide conventional computer implementation.
For the reasons above, claim 1 is rejected as being directed to non-patentable subject matter under §101. This rejection applies equally to independent claims 8 and 15.
Claim 8 recites similar limitations to Claim 1, with the exception of “A computer program product comprising a computer-readable storage medium having a set of instructions stored therein which, when executed by a processor, causes the processor to perform a method comprising:” (generic computer components); therefore, both Claims are similarly rejected.
Claim 15 recites similar limitations to Claim 1, with the exception of “A computer system comprising: a processor(s) set; and a computer readable storage medium having program instructions stored therein; wherein: the processor set executes the program instructions that cause the processor set to perform a method comprising:” (generic computer components); therefore, both Claims are similarly rejected.
Dependent Claims:
Claim 2 (claims 9 and 16) recites insignificant extra-solution activity steps “inputting the first image into a trained siamese network to generate the first embedding.”
Claim 3 (claims 10 and 17) recites an instructions to apply step “training a siamese network of shared parameters with pairs of images representing authentic and counterfeit resources to create a trained siamese network, the trained siamese network generating image embeddings of each image;” (See MPEP 2106.05(f)) and mere extra-solution activity “generating, by the trained siamese network and under expert supervision, K-medoids models for creating a cluster of authentic image embeddings and a cluster of counterfeit image embeddings; wherein: the plurality of existing clusters were created by a selected K-medoids model.”
Claim 4 (claims 11 and 18) recites abstract idea mental process steps “calculating integrity of cluster (IOC) for clusters created by the K-medoids models, the IOC being based on a count of ground truth labels in each cluster, a total number of image embeddings in each cluster, and a total number of clusters in the embedding space;” and “selecting the selected K-medoids model from a set of K-medoids models based on a comparison of the calculated IOC of clusters created by each K-medoids model, the selected K-medoids model having a preferred IOC.”
Claim 5 (claims 12 and 19) recites abstract idea mental process steps “determining a count of outlier embeddings in the set of outlier embeddings including the first embedding and the at least one other outlier embedding;” and “responsive to the count meeting at least a threshold number of embeddings, generating a new cluster in the embedding space.”
Claim 6 (claims 13) recites abstract idea mental process steps “generating a global threshold for the plurality of existing clusters in the embedding space;”, “selecting a best center among the set of outlier embeddings in the embedding space, wherein selected close outlier embeddings make up a set of non-outlier embeddings having the best center as a new medoid of the new cluster;”, and “generating a cluster threshold for the new cluster based on a minimum distance from the new medoid to enclose the set of non-outlier embeddings.”
Claim 7 (claim 14) recites refinements to the calculation abstract idea mental process step(s).
Claim 21 recites abstract idea mental process steps “generating…”, “selecting…”, and “generating…”.
Claim Rejections - 35 USC § 103
The text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
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.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claim(s) 1-3, 7-10, and 14-17 is/are rejected under 35 U.S.C. 103 as being unpatentable over Cali et al. (US 2019/0213408 A1), hereinafter Cali; Galle et al. (US 2013/0262465 A1), hereinafter Galle; and Duan et al. (Cluster-based outlier detection, 2008), hereinafter Duan.
In regards to claim 1: The present invention claims: “A computer-implemented method comprising: loading a first embedding of a first image into an embedding space where there is a plurality of…other image embeddings,” Cali teaches “A computer-implemented method for assessing if characters in a sample image are formed from a predefined font. The method comprises forming a first embedded space representation for the predefined font, extracting sample characters from the sample image, forming a second embedded space presentation of the sample characters, and comparing the first and second embedded space representation to assess if the sample characters are of the predefined font.” (Abstract)
“…including at least one authentic class…and a plurality of counterfeit class…;” Cali teaches their method is for the assessment of counterfeit documents against real documents or data (See abstract and at least [0017] for comparing/classification, [0013] for reference to counterfeit detection).
“and reporting the first image as representing a counterfeit resource.” Cali teaches “Preferably, the final similarity score for each text field is the averaged score for all the characters of that text field. The final output is a list of confidence values, one value per text field that indicates if the text field has the genuine font or not. These values can be combined or analysed separately to come to a final result on whether the sample document is authentic.” ([0099]).
Cali fails to explicitly teach:
“…a plurality of existing clusters…” However, Galle, in a similar field of endeavor of document classification, teaches “An initial "clustering" of the data points is performed at S110. The algorithm starts by considering each point as a potential cluster. The number of clusters thus corresponds to the number of points, and each point is assigned to its own cluster. In other embodiments, fewer than all data points are assigned to a unique cluster, such that some clusters initially have more than one data point.” ([0050]), which the examiner submits reads on the “existing clusters” limitation.
“the plurality of existing clusters having defined respective medoids and corresponding cluster thresholds, the plurality of existing clusters…” Galle teaches “The exemplary threshold-based clustering algorithm may employ some or all of the following… 2. It relaxes the "leader" constraint. In conventional algorithms, a cluster is represented by a leader point (also known as a medoid) that is a real data point. This point is used as reference to compute similarities or distances. This freedom is particularly advantageous for news article clustering where several sources are involved and there is no clear central article.” ([0024] and [0026]) and “7. MedoidShift, a variant of MeanShift. (Yaser Ajmal Sheikh, Erum Arif Khan, and Takeo Kanade. Modeseeking by Medoidshifts. In ICCV, pages 1-8. IEEE, 2007). Instead of computing the mean of the points inside the hypersphere of radius -c, MedoidShift computes the median, thus reducing the possible set of representative of the clusters to the set of original points. This implies a leader-based clustering, which in the case of news-event performs worse in general.” ([0113]). Galle also teaches “This includes assigning the data points to the clusters based on a comparison measure of each data point with a representative point of each cluster, and a threshold of the comparison measure.” ([0008]).
“determining medoid distances between the first embedding and the respective medoids of the plurality of existing clusters;” Galle teaches “For generality, the terms "comparison measure" or "comparison" or other similar phraseology is used herein to encompass both similarity measures and distance or divergence measures.” ([0022], see comparison measure from [0008] above). See [0026] for the medoid being used “as reference to compute similarities or distances.”
“comparing a first medoid distance to a corresponding cluster threshold of a first cluster;” Galle teaches “At S102, a comparison measure threshold -i: is established. As explained below, this threshold determines whether a data point contributes positively or negatively to a score for a clustering, based on a computed comparison measure ( e.g., distance), with respect to a representative point of a cluster to which the data point is assigned in the clustering. The comparison measure threshold -i: may be user-defined and/or defined automatically, based on a training set of similar documents.”([0046]).
Galle teaches “Clustering algorithms are useful tools for analyzing data. Many algorithms exist for this task, although their application to a particular problem is very much data-dependent. For example, in the case of news article clustering, clustering may be based on the detection of events inside a given collection of news articles coming from multiple sources. However, since the events themselves are often unpredictable in advance and the articles often arrive in small batches, the identification of clusters is challenging.” ([0002]) and “Threshold-based clustering algorithms tend to be better suited to clustering in such a setting, where the given input is a threshold on the similarity ( denoted by [tau]) that relates to how close documents in the same cluster should be to each other.” ([0004]). It would have been obvious to one of ordinary skill in the art at the time of the Applicant’s filing to use known methods and benefits from Galle in a counterfeit system such as Cali in order to accurately cluster and classify documents or text within documents to determine their authenticity.
While Galle teaches “The fully-incremental algorithm supposes that all points arrive one by one. For the first point p1 it creates a cluster c1 such that y(c 1)={p1}. Forpointp1 arriving at time t, it computes a similarity measure between p1 and each of the clusters already existing at time t. If none of these similarities is greater than a threshold -c, a new cluster is created whose only point is p 1 . Else, p 1 is assigned to the most similar cluster.” ([0121]), the combination of Cali and Galle fails to explicitly teach the limitations of:
“responsive to no medoid distance to any cluster being less than respectively corresponding cluster thresholds, assigning the first embedding to a set of outlier embeddings not matching any cluster, the set of outlier embeddings including at least one other outlier embedding;” Duan, in a similar field of endeavor of classification clustering, teaches a theoretical maximum and minimum number of samples that can make up a cluster before one is formed (Sections 5.1 and 5.2). (Examiner’s Note: Given the ambiguity surrounding this limitation (See the 112(b) Rejection above), the Examiner interprets this limitation broadly and considers a combination of Cali, Galle, and Duan sufficient to read on withholding clustering a given datapoint until a critical number of unclustered outliers is reached).
Duan teaches “Outlier detection has important applications in the field of data mining, such as fraud detection…”, “Outliers are traditionally considered as single points; however, there is a key observation that many abnormal events have both temporal and spatial locality, which might form small clusters that also need to be deemed as outliers. In other words, not only a single point but also a small cluster can probably be an outlier.” (Abstract), and “Comparing with single point outliers, cluster-based outliers are more interesting. Many single point outliers are related to occasional trivial events, while cluster-based outliers concern some important lasting abnormal events. Generally speaking, it is reckless to form a cluster with only 2 or 3 objects,” (Section 5.2). It would have been obvious to one of ordinary skill in the art at the time of the Applicant’s filing to take known elements of Duan’s cluster analysis when combining the methods of Cali and Galle to more accurately represent abnormal clusters when determining a counterfeit document.
In regards to claim 2: The present invention claims: “inputting the first image into a trained siamese network to generate the first embedding.” Cali teaches “The embedded space model may have been generated, or formed, using a Siamese network. The use of the Siamese network in conjunction with the convolutional neural network helps form, or train, an embedded space model that efficiently forms high quality embedded space representations.” ([0021]).
In regards to claim 3: The present invention claims: “training a siamese network of shared parameters with pairs of images representing authentic and counterfeit resources to create a trained siamese network, the trained siamese network generating image embeddings of each image;” Cali teaches “Extracting sample characters from the sample image may comprise obtaining a list of embedded representations of the character images using the Siamese network.” ([0022]) and “The training, of forming, of the embedded space forming CNN, or embedded space model, uses a Siamese network. This method has two key aspects, firstly the training is done using batches of pairs of samples, as opposed to a batch of individual samples, and secondly it uses a contrastive loss function. A contrastive loss function penalizes pairs which are labelled as different but have a low Euclidean distance in the embedding space and also penalizes pairs which are labelled as having the same font but have a large Euclidean distance in the embedded space.” ([0092]).
“and generating, by the trained siamese network and under expert supervision, K-medoids models for creating a cluster of authentic image embeddings and a cluster of counterfeit image embeddings; wherein: the plurality of existing clusters were created by a selected K-medoids model.” See above how Cali teaches the Siamese network, while Galle teaches the clustering of documents based on classification. While Galle does not explicitly teach a K-medoid model, Galle does teach “In this setting, so-called k-based algorithms tend to perform poorly. These algorithms take as input the number of expected clusters and try to fit the given points into k clusters guided by a selected score function. Typically, a user specifies several possible values fork ( or an interval of values) and the score function is extended in order to be able to choose the best value for k, by including a complexity-penalizing term. However, it is generally not evident what could be the expected number of events at a given moment, and this number may change over time. Also, there are possibly outlier articles to deal with. These are documents which do not talk about any particular event. However, k-based algorithms are very sensitive to the presence of outliers.” ([0003]) and “The comparison measure threshold -i: may be user-defined and/or defined automatically, based on a training set of similar documents.” ([0046]). The Examiner submits, in a system clustering real and counterfeit document or text images as a combination of Cali and Galle suggests, a person of ordinary skill in the art may find a k-medoid model to be sufficient in the creation of two primary clusters, versus the high number of complex clusters taught in Galle, especially if the user may define the measurement tau, thereby potentially limiting the number of clusters by threshold.
In regards to claim 5: The present invention claims: “determining a count of outlier embeddings in the set of outlier embeddings including the first embedding and the at least one other embedding; and responsive to the count meeting at least a threshold number of embeddings, generating a new cluster in the embedding space.” See above how Duan teaches a theoretical maximum and minimum number of samples that can make up a cluster before one is formed (Sections 5.1 and 5.2). A combination of Cali, Galle, and Duan would read on a forming a new cluster with more than one outlier.
In regards to claim 6: The present invention claims: “generating a global threshold for the plurality of existing clusters in the embedding space;” See above where Galle teaches measurement thresholds for cluster, including where a new cluster may be made “The fully-incremental algorithm supposes that all points arrive one by one. For the first point p1 it creates a cluster c1 such that y(c 1)={p1}. For point p1 arriving at time t, it computes a similarity measure between p1 and each of the clusters already existing at time t. If none of these similarities is greater than a threshold -c, a new cluster is created whose only point is p 1 . Else, p 1 is assigned to the most similar cluster.” ([0121]).
“selecting a best center among the set of outlier embeddings in the embedding space, wherein selected close outlier embeddings make up a set of non-outlier embeddings having the best center as a new medoid of the new cluster; and generating a cluster threshold for the new cluster based on a minimum distance from the new medoid to enclose the set of non-outlier embeddings.” See above where Galle teaches creating a new cluster with an outlier datapoint (making it a new cluster). Galle also teaches merging clusters ([0089]) and “Merge of clusters will occur when, during the first phase, two clusters turnout to have identical elements or when, during the second phase, two clusters overlap more than a selected amount, such as 80%.” ([0090]). The Examiner submits a person of ordinary skill in the art would reasonably understand the merging of single data-point clusters would reasonably read on finding a new medoid for a cluster and forming a new cluster of data points around it based on the thresholding used by Galle.
In regards to claim 7: The present invention claims: “wherein the corresponding cluster thresholds are calculated for each cluster as a radius.” Galle teaches “7. MedoidShift, a variant of MeanShift. (Yaser Ajmal Sheikh, Erum Arif Khan, and Takeo Kanade. Modeseeking by Medoidshifts. In ICCV, pages 1-8. IEEE, 2007). Instead of computing the mean of the points inside the hypersphere of radius [tau], MedoidShift computes the median, thus reducing the possible set of representative of the clusters to the set of original points. This implies a leader-based clustering, which in the case of news-event performs worse in general.” ([0113]).
In regards to claims 8-10 and 12-14: Claims 8-10 and 12-14 recite similar limitations to claims 1-3 and 5-7, with the exception of “A computer program product comprising a computer-readable storage medium having a set of instructions stored therein which, when executed by a processor, causes the processor to perform a method comprising:” of claim 8; therefore, claims both sets of claims are similarly rejected.
In regards to claims 15-17 and 19: Claims 15-17 recite similar limitations to claims 1-3 and 5, with the exception of “A computer system comprising: a processor(s) set; and a computer readable storage medium having program instructions stored therein; wherein: the processor set executes the program instructions that cause the processor set to perform a method comprising:” of claim 15; therefore, claims both sets of claims are similarly rejected.
In regards to claim 21: New Claim 21 recites similar steps to the above-rejected claims, with the exception of recitation of a time period initiating an iterative process of the aforementioned steps. Galle [0051] and [0054] teaches “An iterative optimization of the clustering of the data points is then initiated which includes alternating steps S112 and S114 (and optionally S116) for a number of iterations until a stopping point is reached. In general, each of steps S112 and S114 is performed at least twice. For example, optimization may continue while the clustering score does not converge from one iteration to the next (i.e., while the clustering score continues to improve by at least a threshold amount). In other embodiments, the number of iterations can be fixed, such as at least 3, or at least 5, or at least 10 iterations. The clustering in the last ( or a later one) of these iterations is the input to S120.” and “At S116, clusters which completely overlap each other may be merged to form a single cluster. Then, if at S118, a stopping point has not been reached, the method returns to S112 for one or more iterations ofS112, S114, and optionally S116. In other embodiments, the cluster merging (S116) may be performed later, once the iterations are complete. At each new iteration of S112, the representative points used are those computed in the prior iteration at S114, so the assignments of the data points to the clusters is computed based on the distances (i.e., similarity) to the new representative points. Thus for example, if the representative point shifts away from a data point previously assigned to that cluster, the data point will no longer be assigned to this cluster if it is further than the predetermined threshold -i: and there is another cluster to which it is closer than the threshold -i:.”, respectively. The Examiner submits Galle’s teaching of a predefined number of iterations broadly reads on the new claim’s “pre-defined time period” before repeating steps/recentering clusters/merging outliers into other clusters.
Claim(s) 4, 11, and 18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Cali, Galle, and Duan as applied to claims 1, 8, and 15 above, and further in view of Menendez et al. (Medoid-based clustering using ant colony optimization, 2016), hereinafter Menendez.
In regards to claim 4: The present invention claims: “calculating integrity of cluster (IOC) for clusters created by the K-medoids models, the IOC being based on a count of ground truth labels in each cluster, a total number of image embeddings in each cluster, and a total number of clusters in the embedding space; and selecting the selected K-medoids model from a set of K-medoids models based on a comparison of the calculated IOC of clusters created by each K-medoids model, the selected K-medoids model having a preferred IOC.” While the combination of Cali, Galle, and Duan teaches clustering data for counterfeit detection, they fail to teach the above limitations; however, Menendez, in a similar field of endeavor of classification clustering teaches “One of the main challenges around the clustering problem is how to choose a good number of clusters (Tibshirani et al. 2001). The majority of clustering algorithms require the specification of the number of clusters a priori as a parameter of the algorithm. An alternative to having the number of clusters fixed is based on the use of a metric to evaluate the clusters’ quality, allowing an algorithm to test a variable number of clusters.” (Page 126). Section 3.2 also goes into detail regarding the use of the number and clusters and the size of clusters (Page 130) and “The silhouette compares tightness and separation of clusters. It is calculated by data instance and gives information about those data instances that are well assigned to a cluster and those that should be moved. The silhouette of all data instances provides an appreciation of the clusters’ quality (in a similar way of a Riemann integral). The area of the shape defined by silhouette is useful to determine the quality of the number of clusters selection (see Fig. 3).” (Page 130). Cali also goes into detail regarding labeling data before classification ([0019]-[0020]).
Menendez teaches “Medoid-based clustering methods are helpful—compared to classical centroid-based techniques—when centroids cannot be easily defined. This paper proposes two medoid-based ACO clustering algorithms, where the only information needed is the distance between data: one algorithm that uses an ACO procedure to determine an optimal medoid set (METACOC algorithm) and another algorithm that uses an automatic selection of the number of clusters (METACOC-K algorithm).” (Abstract), with tables 3 and 4 demonstrating the accuracy of their algorithm(s) over previous work. It would have been obvious to one of ordinary skill in the art at the time of the Applicant’s filing to combine the known benefits of Menendez’s algorithm(s) in a clustering system such as a combination of Cali, Galle, and Duan to realize the benefits over traditional centroid-based methods.
In regards to claim 11: Claim 11 recite similar limitations to claim 4, with the exception of “A computer program product comprising a computer-readable storage medium having a set of instructions stored therein which, when executed by a processor, causes the processor to perform a method comprising:” of claim 8; therefore, claims both sets of claims are similarly rejected.
In regards to claim 18: Claim 18 recite similar limitations to claim 4, with the exception of “A computer system comprising: a processor(s) set; and a computer readable storage medium having program instructions stored therein; wherein: the processor set executes the program instructions that cause the processor set to perform a method comprising:” of claim 15; therefore, claims both sets of claims are similarly rejected.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to GRIFFIN T BEAN whose telephone number is (703)756-1473. The examiner can normally be reached M - F 7:30 - 4:30.
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, Li Zhen can be reached at (571) 272-3768. 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.
/GRIFFIN TANNER BEAN/Examiner, Art Unit 2121
/Li B. Zhen/Supervisory Patent Examiner, Art Unit 2121
Prosecution Insights