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 .
Response to Arguments
Applicant's arguments filed July have been fully considered but they are not persuasive. Claims Claims 2, 4, 7, and 9 have been canceled. Thus, Claims 1, 3, 5, 6, 8, and 10 are pending and presented for examination.
Applicant’s arguments regarding 35 U.S.C. 101 abstract idea rejection have been fully considered but they are not persuasive.
Applicant recites the “Alice Test” to express tat the exceptions are combined in a meaningful way beyond generally linking the use of the exception with the environment, however examiner disagrees. As the claims are currently recited, elements are still mentioned at such a level of generality that they don’t overcome 101.
“Preforming unsupervised learning to identify first transaction data”. Identifying transaction data is inherently a mental process, picking out data that has certain noticeable elements from the others is a well-known mental process. The addition of “unsupervised learning” isn’t enough to overcome this issue, as it is just for the environment (computer) to process a mental process. It is recommended to add more elements, such as why this mental process can only occur on a computer rather than having the unsupervised learning being taught as the way for the computer to perform the process. A suggestion, perhaps, is to expand on why the unsupervised learning process is meaningful such that it generates results that a human being could reasonably do.
“excluding data from target data”. Similarly, this is just picking out sets of data from the original set of data. This does not provide more then the mental process which is replicated by the process of judgement.
“learning a second hierarchical mixed model using training data that remains after the exclusion”. Creating a model based on data, especially mentioned with any descriptor of how the model is “learned” falls under just an element of the field of use. Training models on a set of data is commonplace in the field of use so there needs to be more provided such that it doesn’t broadly mention learning a model. Learning a model, also can be understood to be a mental process. A “model” is very vague and could be understood as just a representation of the data. Therefore, one could feasibly make something similar to a graph to represent the model of the training data.
“visualizing, on a display device, a ratio by displaying both first and second graphics representing different numbers of target data”. Visualizing data has been found by the courts to be an extra-solution activity in particular, falls under Berkheimer evidence of Presenting Offers and gathering statistics, which is found in MPEP 2106.04(d)(ll). Visualizing graphics broadly on the data is not enough to overcome the rejection. Furthermore, visualizing graphics is something that can be performed with a pen and paper. All that is required is to observe the data then create a graphical representation of a piece of pen and paper.
Regarding the next argument proposed by the applicant which is that the invention improves the functioning of a computer or improves another technology or technical field as introduced in Page 10 – 11, there are a couple of elements to take note of. Using models to detect fraudulent data is elements that are not improvement on the current technology in the space. The models being “hierarchical mixed models” doesn’t add much either as there is nothing in the claims that outline how these models are improvements of the space. They are just “learned” however the elements that make these models special and therefore lead to an improvement are not recited. The features in the claims therefore do not address how they assist with “inaccurate predictions due to imbalanced data”, all that is claimed is the hierarchical mixed models but not how the improvement is performed.
Furthermore, the improvements of a user interface are not recited in the claims either. There needs to be a better understanding of how “visualization allow for the prediction accuracy to be classified together with interpretability” is being performed.
Therefore, regarding the applicant’s analysis of Step 2A Prong 1 and Prong 2, the examiner respectfully disagrees.
Similarly, the same issues are found in Claim 6 and 8 and similar issues are in 3, 5 and 10. Please find the amended 35 U.S.C. 101 rejection below.
Therefore, the 35 U.S.C 101 rejection is maintained.
Applicant’s arguments regarding 35 U.S.C. 112 have been fully considered and are persuasive.
Applicant’s arguments regarding 35 U.S.C. 103 rejection have been fully considered but they are not persuasive.
Applicant highlights this quotation from Claim 1:
and visualize, on a display device, the ratio of the target data aggregated for each score to the entire target data, by displaying on the display device both a first graphic representing all target data with a size according to a total number of target data, and a second graphic representing target data according to a number of data of the target data for a corresponding score,
However, Carcillo teaches the both graphs with the ratio (frequency) of the fraud scores (from the calculation using the classifier which in combination would be the classifier from Zoldi which calculates the scores) compared to the entire target data which is seen in graph a in which the data has a size (size is mentioned broadly here, and therefore is understood to be a ratio of the data to the target data, which can be observed by the scores of a.) as well as a second graph which compare the number of data (proportion gives the relative number) for a corresponding score – which is seen in graph d.)
Applicant highlights this quotation from Claim 1:
To teach this limitation, a quotation from the 103 is provided.
calculate, for each leaf node, a ratio of the classified training data that is predicted to be a negative example, (Zoldi teaches Fig.3 which teaches that for each Node a Pf score (fraud probability) is calculated, which represents the ratio of the classified training data using the data predicted to be a negative example (nonfraud).),
and exclude, from the target data, data that satisfies a condition for classification into the leaf node for which the calculate ratio is equal to or greater than a predetermined threshold. (Zoldi teaches this limitation: “A high fidelity probability region (i.e., in the vicinity of the cluster centroid) can be found from the class probability distribution using a threshold probability, and this radius demarcates two regions of different detection capabilities. [..] If the new data falls into a high-fidelity central region (condition), the threshold probability is assigned to the new data. Otherwise the left-over class probability on the node is assigned to the new data.“ [0033])
Regarding the argument that the Zoldi does not “exclude” data, Zoldi teaches separating the data into different sets of data (with the threshold probability or with the left-over class probability) – thereby excluding the data based on the threshold probability into two separate sets of data. This, in combination with Guo, thereby teaches a known way of creating the negative example by Guo to create class probabilities for the system. Regarding the “ratio for each leaf node” figure 3 directly teaches Pf “fraud probability at a leaf node” which involves a ratio of the number of non-fraud samples and number of fraud samples at each leaf node. Therefore, examiner respectfully disagrees.
The complete rejection can be found below and Claim 1, 6, and 8 are not found to be patentable over the recited art. The full detail of the analysis is in the amended 35 U.S.C. 103 rejection below.
Therefore, the 35 U.S.C. 103 rejection is maintained.
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.
Claim(s) 1, 6, and 8 and dependent claims 3, 5, and 10 rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1
According to the first part of the analysis, in the instant case, claims 1 – 9 are directed to a process, machine, manufacture, or composition of matter. As each of the claims falls within one of the four statutory categories (i.e., process, machine, manufacture, or composition of matter).
Independent Claim 1:
Step 2A Prong 1:
(This step for identifying and picking out data that does not have fraudulent characteristics is practically implementable in the human mind and is understood to be a recitation of a mental process (i.e., judgment).)
extract target data by excluding the first transaction data from the transaction data; (This step for determining for excluding a subset of data is practically implementable in the human mind and is understood to be a recitation of a mental process (i.e., judgment).)
(This step for determining for learning and creating a model using training data is practically implementable in the human mind with the aid of a pen and paper and is understood to be a recitation of a mental process (i.e., judgment).)
exclude, from the target data, the target data which is set as negative example training data and is classified as a negative example by the first hierarchical mixed model; (This step for excluding data and classifying data is practically implementable in the human mind and is understood to be a recitation of a mental process (i.e., judgment).)
(This step for determining for learning and creating a model using training data is practically implementable in the human mind with the aid of a pen and paper and is understood to be a recitation of a mental process (i.e., judgment).)
calculate as a score a ratio of target data a ratio of target data for which the training data is discriminated as the positive example by the second hierarchical mixed model; (This step for calculating a score based on the ratio is a mathematical concept. The step for discriminating based on the calculated ratio is understood to be a mental process (i.e., judgement) )
wherein the processor further executes instructions to discriminate the training data classified in a leaf node using a discriminant placed in each leaf node of the first hierarchical mixed model, calculate, for each leaf node, a ratio of the classified training data that is predicted to be a negative example, and exclude, from the target data, data that satisfies a condition for classification into the leaf node for which the calculate ratio is equal to or greater than a predetermined threshold. (The first step involves discriminating training data within a leaf node which is practically implementable in the human mind with the aid of a pen and paper and therefore is understood to be a mental process (i.e., judgement). The step involves calculating a ratio for the data, which is understood to be a mathematical process. The last step involves identifying data that satisfies a condition which is understood to be a mental process (i.e., judgement).)
If a claim limitation under its broadest reasonable interpretation, covers performance of the limitation that can be performed in the human mind and/or using pen and paper as a physical aid, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
Step 2A Prong 2:
The judicial exception is not integrated into a practical application.
A fraud detection device for detecting a fraudulent transaction in an operation of a financial institution comprising:
a memory storing instructions; and
one or more processors configured to execute the instructions to: (The device itself, the memory, and the processor is understood to be generic computer equipment. See MPEP 2106.05(f).)
obtain transaction data: (The step regarding obtaining data is understood to be insignificant extra-solution activity. See MPEP 2106.05(g).)
perform unsupervised learning (This is considered an element of the field of use. See MPEP 2106.05(h).)
learn a first hierarchical mixed model (This is considered an element of the field of use. See MPEP 2106.05(h).)
learn a second hierarchical mixed model (This is considered an element of the field of use. See MPEP 2106.05(h).)
and visualize, on a display device, the ratio of the target data aggregated for each score to the entire target data, by displaying on the display device both a first graphic representing all target data with a size according to a total number of target data, and a second graphic representing target data according to a number of data of the target data for a corresponding score, (The step regarding visualization of the ratio to each score is understood to be insignificant extra-solution activity. See MPEP 2106.05(g).)
Step 2B:
The judicial exception is not integrated into a practical application.
A fraud detection device for detecting a fraudulent transaction in an operation of a financial institution comprising:
a memory storing instructions; and
one or more processors configured to execute the instructions to: (The device itself, the memory, and the processor is understood to be generic computer equipment. See MPEP 2106.05(f).)
obtain transaction data; (The step regarding obtaining transaction data is understood to be an element covered by Berkheimer evidence, in particular “Receiving or transmitting data”. See MPEP 2106.04(d)(ll).)
perform unsupervised learning (This is considered an element of the field of use. See MPEP 2106.05(h).)
learn a first hierarchical mixed model (This is considered an element of the field of use. See MPEP 2106.05(h).)
learn a second hierarchical mixed model (This is considered an element of the field of use. See MPEP 2106.05(h).)
and visualize, on a display device, the ratio of the target data aggregated for each score to the entire target data, by displaying on the display device both a first graphic representing all target data with a size according to a total number of target data, and a second graphic representing target data according to a number of data of the target data for a corresponding score, (The step regarding visualization of the ratio is understood to be an element covered by Berkheimer evidence, in particular “Presenting Offers and gathering statistics”. See MPEP 2106.04(d)(ll).)
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because when considered separately and in combination, they do not add significantly more (also known as an “inventive concept”) to the exception.
Independent Claim 6:
Step 2A Prong 1:
A fraud detection method for detecting a fraudulent transaction in an operation of a financial institution comprising:
(This step for identifying and picking out data that does not have fraudulent characteristics is practically implementable in the human mind and is understood to be a recitation of a mental process (i.e., judgment).)
extracting target data by excluding the first transaction data from the transaction data; (This step for determining for excluding a subset of data is practically implementable in the human mind and is understood to be a recitation of a mental process (i.e., judgment).)
(This step for determining for learning and creating a model using training data is practically implementable in the human mind with the aid of a pen and paper and is understood to be a recitation of a mental process (i.e., judgment).)
excluding, from the target data, the target data which is set as negative example training data and is classified as a negative example by the first hierarchical mixed model; (This step for excluding data and classifying data is practically implementable in the human mind and is understood to be a recitation of a mental process (i.e., judgment).)
(This step for determining for learning and creating a model using training data is practically implementable in the human mind with the aid of a pen and paper and is understood to be a recitation of a mental process (i.e., judgment).)
calculating as a score a ratio of target data a ratio of target data for which the training data is discriminated as the positive example by the second hierarchical mixed model; (This step for calculating a score based on the ratio is a mathematical concept. The step for discriminating based on the calculated ratio is understood to be a mental process (i.e., judgement) )
discriminating the training data classified in a leaf node using a discriminant placed in each leaf node of the first hierarchical mixed model; (The step for separating data using a discriminant is understood to be a mental process (i.e., judgement)
and calculating, for each leaf node, a ratio of the classified training data that is predicted to be a negative example; (The step involves calculating a ratio for the data, which is understood to be a mathematical process.)
excluding, from the target data, data that satisfies a condition for classification into the leaf node for which the calculate ratio is equal to or greater than a predetermined threshold. (The step involves identifying data that satisfies a condition which is understood to be a mental process (i.e., judgement).)
If a claim limitation under its broadest reasonable interpretation, covers performance of the limitation that can be performed in the human mind and/or using pen and paper as a physical aid, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
Step 2A Prong 2:
The judicial exception is not integrated into a practical application.
obtaining transaction data: (The step regarding obtaining data is understood to be insignificant extra-solution activity. See MPEP 2106.05(g).)
preforming unsupervised learning (This is considered an element of the field of use. See MPEP 2106.05(h).)
learning a first hierarchical mixed model (This is considered an element of the field of use. See MPEP 2106.05(h).)
learning a second hierarchical mixed model (This is considered an element of the field of use. See MPEP 2106.05(h).)
visualizing, on a display device, the ratio of the target data aggregated for each score to the entire target data, by displaying on the display device both a first graphic representing all target data with a size according to a total number of target data, and a second graphic representing target data according to a number of data of the target data for a corresponding score, (The step regarding visualization of the ratio to each score is understood to be insignificant extra-solution activity. See MPEP 2106.05(g).)
Step 2B:
obtaining transaction data; (The step regarding obtaining transaction data is understood to be an element covered by Berkheimer evidence, in particular “Receiving or transmitting data”. See MPEP 2106.04(d)(ll).)
performing unsupervised learning (This is considered an element of the field of use. See MPEP 2106.05(h).)
learning a first hierarchical mixed model (This is considered an element of the field of use. See MPEP 2106.05(h).)
learning a second hierarchical mixed model (This is considered an element of the field of use. See MPEP 2106.05(h).)
visualizing, on a display device, the ratio of the target data aggregated for each score to the entire target data, by displaying on the display device both a first graphic representing all target data with a size according to a total number of target data, and a second graphic representing target data according to a number of data of the target data for a corresponding score, (The step regarding visualization of the ratio is understood to be an element covered by Berkheimer evidence, in particular “Presenting Offers and gathering statistics”. See MPEP 2106.04(d)(ll).)
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because when considered separately and in combination, they do not add significantly more (also known as an “inventive concept”) to the exception.
Independent Claim 8:
Step 2A Prong 1:
(This step for identifying and picking out data that does not have fraudulent characteristics is practically implementable in the human mind and is understood to be a recitation of a mental process (i.e., judgment).)
extracting target data by excluding the first transaction data from the transaction data; (This step for determining for excluding a subset of data is practically implementable in the human mind and is understood to be a recitation of a mental process (i.e., judgment).)
(This step for determining for learning and creating a model using training data is practically implementable in the human mind with the aid of a pen and paper and is understood to be a recitation of a mental process (i.e., judgment).)
excluding, from the target data, the target data which is set as negative example training data and is classified as a negative example by the first hierarchical mixed model; (This step for excluding data and classifying data is practically implementable in the human mind and is understood to be a recitation of a mental process (i.e., judgment).)
(This step for determining for learning and creating a model using training data is practically implementable in the human mind with the aid of a pen and paper and is understood to be a recitation of a mental process (i.e., judgment).)
calculating as a score a ratio of target data a ratio of target data for which the training data is discriminated as the positive example by the second hierarchical mixed model; (This step for calculating a score based on the ratio is a mathematical concept. The step for discriminating based on the calculated ratio is understood to be a mental process (i.e., judgement) )
discriminating the training data classified in a leaf node using a discriminant placed in each leaf node of the first hierarchical mixed model; (The step for separating data using a discriminant is understood to be a mental process (i.e., judgement)
and calculating, for each leaf node, a ratio of the classified training data that is predicted to be a negative example; (The step involves calculating a ratio for the data, which is understood to be a mathematical process.)
excluding, from the target data, data that satisfies a condition for classification into the leaf node for which the calculate ratio is equal to or greater than a predetermined threshold. (The step involves identifying data that satisfies a condition which is understood to be a mental process (i.e., judgement).)
If a claim limitation under its broadest reasonable interpretation, covers performance of the limitation that can be performed in the human mind and/or using pen and paper as a physical aid, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
Step 2A Prong 2:
The judicial exception is not integrated into a practical application.
A non-transitory computer readable information recording medium storing a fraud detection program applied to a computer which detects a fraudulent transaction in an operation of a financial institution, when executed by a processor, that performs a method for: (The non-transitory computer readable information recording medium and the processor is understood to be generic computer equipment. See MPEP 2106.05(f).)
obtaining transaction data: (The step regarding obtaining data is understood to be insignificant extra-solution activity. See MPEP 2106.05(g).)
preforming unsupervised learning (This is considered an element of the field of use. See MPEP 2106.05(h).)
learning a first hierarchical mixed model (This is considered an element of the field of use. See MPEP 2106.05(h).)
learning a second hierarchical mixed model (This is considered an element of the field of use. See MPEP 2106.05(h).)
visualizing, on a display device, the ratio of the target data aggregated for each score to the entire target data, by displaying on the display device both a first graphic representing all target data with a size according to a total number of target data, and a second graphic representing target data according to a number of data of the target data for a corresponding score, (The step regarding visualization of the ratio to each score is understood to be insignificant extra-solution activity. See MPEP 2106.05(g).)
Step 2B:
A non-transitory computer readable information recording medium storing a fraud detection program applied to a computer which detects a fraudulent transaction in an operation of a financial institution, when executed by a processor, that performs a method for: (The non-transitory computer readable information recording medium and the processor is understood to be generic computer equipment. See MPEP 2106.05(f).)
obtaining transaction data; (The step regarding obtaining transaction data is understood to be an element covered by Berkheimer evidence, in particular “Receiving or transmitting data”. See MPEP 2106.04(d)(ll).)
performing unsupervised learning (This is considered an element of the field of use. See MPEP 2106.05(h).)
learning a first hierarchical mixed model (This is considered an element of the field of use. See MPEP 2106.05(h).)
learning a second hierarchical mixed model (This is considered an element of the field of use. See MPEP 2106.05(h).)
visualizing, on a display device, the ratio of the target data aggregated for each score to the entire target data, by displaying on the display device both a first graphic representing all target data with a size according to a total number of target data, and a second graphic representing target data according to a number of data of the target data for a corresponding score, (The step regarding visualization of the ratio is understood to be an element covered by Berkheimer evidence, in particular “Presenting Offers and gathering statistics”. See MPEP 2106.04(d)(ll).)
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because when considered separately and in combination, they do not add significantly more (also known as an “inventive concept”) to the exception.
Dependent Claims 2 – 5, 7, and 9 are also ineligible for the same reasons given with respect to claim 1. The dependent claims describe further mental processes and/or do not include additional active functional limitations/steps:
Claim 3:
Step 2A, Prong 1:
The fraud detection device according to claim 1, wherein the (The first step involves discriminating training data within a leaf node which is practically implementable in the human mind with the aid of a pen and paper and therefore is understood to be a mental process (i.e., judgement). The step involves calculating a ratio for the data, which is understood to be a mathematical process. The last step involves identifying a condition for the node to match a threshold which is understood to be a mental process (i.e., judgement).)
Step 2A Prong 2:
The judicial exception is not integrated into a practical application.
processor (The processor is understood to be generic computer equipment. See MPEP 2106.05(f).)
Step 2B:
processor (The processor is understood to be generic computer equipment. See MPEP 2106.05(f).)
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because when considered separately and in combination, they do not add significantly more (also known as an “inventive concept”) to the exception.
Claim 5:
Step 2A, Prong 1:
The fraud detection device according to claim 1, wherein the (This step for identifying conditions for exclusion and excluding data based on conditions is practically implementable in the human mind and is understood to be a recitation of a mental process (i.e., judgment).)
Step 2A Prong 2:
The judicial exception is not integrated into a practical application.
processor (The processor is understood to be generic computer equipment. See MPEP 2106.05(f).)
Step 2B:
processor (The processor is understood to be generic computer equipment. See MPEP 2106.05(f).)
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because when considered separately and in combination, they do not add significantly more (also known as an “inventive concept”) to the exception.
Claim 10:
Step 2A, Prong 1:
The fraud detection device according to claim 1, wherein the
Step 2A Prong 2:
The judicial exception is not integrated into a practical application.
(The step regarding visualization of the ratio is understood to be an element covered by Berkheimer evidence, in particular “Presenting Offers and gathering statistics”. See MPEP 2106.04(d)(ll).)
Step 2B:
(The step regarding visualization of the ratio is understood to be an element covered by Berkheimer evidence, in particular “Presenting Offers and gathering statistics”. See MPEP 2106.04(d)(ll).)
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because when considered separately and in combination, they do not add significantly more (also known as an “inventive concept”) to the exception.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claim(s) 1, 3, 5, 6, 8 and 10 is/are rejected under 35 U.S.C. 103 as being unpatentable over Guo (International Application WO2019100844A1) in view of Malini’s Analysis on credit card fraud identification techniques based on KNN and outlier detection and further in view of Zoldi (U.S. Publication US20170083920A1) and in view of Marcjan (U.S. Publication US20190130406A1) and finally in view of Carcillo’s Streaming active learning strategies for real-life credit card fraud detection: assessment and visualization.
Regarding Claim 1, Guo teaches:
A fraud detection device for detecting a fraudulent transaction in an operation of a financial institution comprising:
a memory storing instructions; and
one or more processors configured to execute the instructions to: (“Corresponding to the above method embodiments, the present specification also provides an embodiment of an electronic device. The electronic device includes a processor and a memory for storing machine executable instructions; wherein the processor and the memory are typically interconnected by an internal bus.” [Guo Page 11])
obtain transaction data: (Guo teaches: “Marking the sample carrying the risk label in the training sample” [Page 1] in which the training sample is the transaction data.)
perform (“In this specification, the modeler can collect historical transaction data of a large number of users in advance, extract transaction features (extract target data) of several dimensions from the collected historical transaction data as modeling features, and construct feature vectors as training samples based on these modeling features. Then, based on these training samples, a feature matrix is constructed as a training sample set. “[Guo Page 5]. Guo teaches that the data is identified then extracted based on the modeling features that represent fraudulent transaction data.)
learn a first hierarchical mixed model using training data, among the target data, which takes positive examples for the data indicating the fraudulent transaction and takes negative examples for the remaining data other than the positive examples; (“Step 102: Mark the sample carrying the risk tag in the training sample set as a positive sample, and the sample not carrying the risk tag as a negative sample;
Step 104: Train the supervised risk (model) based on the positive and negative samples of the tag, and perform risk assessment on the negative sample separately based on the model that is completed by the training to obtain a risk score” [Guo Page 5])
and exclude, from the target data, the target data which is set as negative example training data and is classified as a negative example by the (“Step 106: Filter (exclude) a negative sample in the training sample set that the risk score is greater than a preset threshold” [Guo Page 5])
learn a (Guo teaches: “Step 108, retraining the supervised risk model based on the filtered (remaining data) positive and negative samples in the training sample set.” [Page 5]. Guo teaches the creation of these models.)
calculate as a score a (Guo teaches: “After the supervised machine learning model is retrained based on the filtered training sample set, the modeler can deploy the model in the wind control system and perform risk assessment on the user's daily online transaction based on the model. The online transaction is scored to obtain a risk score (calculates a score based on the data), and then based on the obtained risk score to determine whether the transaction is risky” [Page 9])
Guo does not distinctly disclose:
perform unsupervised learning to identify first transaction data that is not fraudulent transaction data, among the transaction data
Guo does not distinctly disclose applying unsupervised learning, however Guo does teach identifying transaction data that is not fraudulent.
However, Malini teaches unsupervised learning being the way that (Malini teaches on Page 5: “unsupervised outlier detection” in which the normal transaction data is represented as the non-outlier data and the fraudulent data is the outlier data.)
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to combine the feature selection of Guo with the unsupervised learning method taught by Malini so that the system can learn to handle finding new trends in the data. (Malini teaches this on Page 4: “In outlier detection method unsupervised learning is preferred to detect the fraud because it can lead to new explanations and representation of the observation data.”)
Guo as modified by Malini does not disclose:
a hierarchical mixed model.
However, Zoldi teaches a hierarchical mixed model: Fig. 3 and paragraph 0042 of Malini teaches a decision tree structure ([0021] of the applicant’s specification describes the mixed model being represented by a tree structure) for use as a trained model.
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to replace the model taught by Guo as modified by Malini with the tree model and analysis taught by Zoldi so that the model can partition the data into small chunks and data analysis can be performed on smaller separate classified datasets. (Zoldi teaches in 0086: “As described above, the decision tree partitions the entire dataset into small chunks, depending on its intrinsic feature distribution, and the dataset is distributed onto many nodes so that the clustering is performed on the small partitioned datasets.”)
Guo as modified by Malini and further modified by Zoldi does not distinctly disclose:
second hierarchical mixed model.
However, Marcjan teaches the use of a second model in paragraph 0089, in which the newly labeled data is passed into a second trained risk determination model.
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to combine the hierarchical mixed model taught by Guo as modified by Malini and further modified by Zoldi with the method of creating a second model as taught by Marcjan to yield the predictable result of allowing the system to evaluate one set of data in different ways as each model could compare different elements of the data.
Guo as modified by Malini, Zoldi, and Marcjan does not distinctly disclose:
a ratio of target data for which the training data is discriminated
However, Zoldi teach in another embodiment: a ratio of target data for which the training data is discriminated. (Zoldi teaches: “taking the ratio of the count of number of fraud samples to the total number of all the samples.”[0090].)
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to combine the calculation of the score as taught by Guo as modified by Malini, Zoldi, and Marcjan with the ratio calculation of this embodiment of Zoldi obtain the predictable result of allowing the system to analyze and create a class probability of each node in the system. (This is taught by Zoldi in paragraph 0090, which shows taking the ratio of the count of number of fraud samples to the total number of all the samples to obtain the left-over class probability of the node.)
Guo as modified by Malini, Zoldi, and Marcjan does not distinctly disclose:
and visualize, on a display device, the ratio of the target data aggregated for each score to the entire target data, by displaying on the display device both a first graphic representing all target data with a size according to a total number of target data, and a second graphic representing target data according to a number of data of the target data for a corresponding score,
However, Carcillo teaches: and visualize, on a display device, the ratio of the target data aggregated for each score to the entire target data, by displaying on the display device both a first graphic representing all target data with a size according to a total number of target data, and a second graphic representing target data according to a number of data of the target data for a corresponding score, (Carcillo teaches Fig. 4 on page 8 which shows a visualization of the ratio of the fraud scores to the entire target data (a) and the second graphic representing target data according to a number of data of the target data for a corresponding score (d)).
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to combine the analysis method of Guo as modified by Malini, Zoldi, and Marcjan with the visualization step of Carcillo for the known result of being able to compare the results of the machine learning model.
Guo as modified by Malini, Zoldi, and Marcjan does not distinctly disclose:
discriminate the training data classified in a leaf node using a discriminant placed in each leaf node of the first hierarchical mixed model, calculate, for each leaf node, a ratio of the classified training data that is predicted to be a negative example, and exclude, from the target data, data that satisfies a condition for classification into the leaf node for which the calculate ratio is equal to or greater than a predetermined threshold.
However, Zoldi teaches:
discriminate the training data classified in a leaf node using a discriminant placed in each leaf node of the first hierarchical mixed model, (Zoldi teaches: “Specifically, a clustering approach is applied to the training samples in a leaf node (or decision node) and to group the training samples into a plurality of subsets (clusters). New data traverse the tree and will be classified by determining the memberships (discriminant) to each cluster and the characteristics of the cluster to improve the predictability of the decision tree model.” [0010].),
calculate, for each leaf node, a ratio of the classified training data that is predicted to be a negative example, (Zoldi teaches Fig.3 which teaches that for each Node a Pf score (fraud probability) is calculated, which represents the ratio of the classified training data using the data predicted to be a negative example (nonfraud).),
and exclude, from the target data, data that satisfies a condition for classification into the leaf node for which the calculate ratio is equal to or greater than a predetermined threshold. (Zoldi teaches this limitation: “A high fidelity probability region (i.e., in the vicinity of the cluster centroid) can be found from the class probability distribution using a threshold probability, and this radius demarcates two regions of different detection capabilities. [..] If the new data falls into a high-fidelity central region (condition), the threshold probability is assigned to the new data. Otherwise the left-over class probability on the node is assigned to the new data.“ [0033])
The motivation for the combination is the same as the one found earlier in the rejection.
Regarding Claim 3, Guo as modified by Malini, Zoldi, Marcjan, and Carcillo teaches all the limitations of Claim 1, and Zoldi further teaches: The fraud detection device according to claim 1, wherein the processor further executes instructions to discriminate the training data classified in a leaf node using a discriminant placed to each leaf node (Zoldi teaches: “Specifically, a clustering approach is applied to the training samples in a leaf node (or decision node) and to group the training samples into a plurality of subsets (clusters). New data traverse the tree and will be classified by determining the memberships (discriminant) to each cluster and the characteristics of the cluster to improve the predictability of the decision tree model.” [0010].),
calculate, for each leaf node, as the score the ratio of the target data for which the training data is discriminated as the positive example (Zoldi teaches Fig.3 which teaches that for each Node a Pf score (fraud probability) is calculated, which represents the ratio of the classified training data using the data predicted to be a negative example (nonfraud).),
and identify a condition of the node for which the calculated score is equal to or greater than a predetermined threshold as the condition with high accuracy of fraudulent transaction. (Zoldi teaches this limitation: “A high fidelity probability region (i.e., in the vicinity of the cluster centroid) can be found from the class probability distribution using a threshold probability, and this radius demarcates two regions of different detection capabilities. [..] If the new data falls into a high-fidelity central region (condition), the threshold probability is assigned to the new data. Otherwise the left-over class probability on the node is assigned to the new data.“ [0033])
The motivation for this combination is the same as Claim 1.
Regarding Claim 5, Guo as modified by Malini, Zoldi, Marcjan, and Carcillo teaches all the limitations of Claim 1, and Zoldi further teaches: The fraud detection device according to claim 1, wherein the processor further executes instructions to identify conditions for exclusion of the target data each time the first hierarchical mixed model is learned (Zoldi teaches identifying conditions for exclusion of the target data each time the hierarchal mixed model is learned, see paragraph 0065 which starts with step 1 (learning the hierarchical mixed model) and step 3, in which the high-fidelity radius is identified as a condition for classifying and excluding data.),
and exclude the data that satisfies the any of the conditions from the target data. (Zoldi teaches this limitation in which the data is separated: “If the new data falls into a high-fidelity central region, the threshold probability is assigned to the new data. Otherwise (exclude) the left-over class probability on the node is assigned to the new data.“ [0033].)
The motivation for this combination is the same as Claim 1.
Regarding Claim 6, Guo teaches:
A fraud detection method for detecting a fraudulent transaction in an operation of a financial institution comprising:
obtaining transaction data: (Guo teaches: “Marking the sample carrying the risk label in the training sample” [Page 1] in which the training sample is the transaction data.)
performing (“In this specification, the modeler can collect historical transaction data of a large number of users in advance, extract transaction features (extract target data) of several dimensions from the collected historical transaction data as modeling features, and construct feature vectors as training samples based on these modeling features. Then, based on these training samples, a feature matrix is constructed as a training sample set. “[Guo Page 5]. Guo teaches that the data is identified then extracted based on the modeling features that represent fraudulent transaction data.)
learning a first hierarchical mixed model using training data, among the target data, which takes positive examples for the data indicating the fraudulent transaction and takes negative examples for the remaining data other than the positive examples; (“Step 102: Mark the sample carrying the risk tag in the training sample set as a positive sample, and the sample not carrying the risk tag as a negative sample;
Step 104: Train the supervised risk (model) based on the positive and negative samples of the tag, and perform risk assessment on the negative sample separately based on the model that is completed by the training to obtain a risk score” [Guo Page 5])
excluding, from the target data, the target data which is set as negative example training data and is classified as a negative example by the (“Step 106: Filter (exclude) a negative sample in the training sample set that the risk score is greater than a preset threshold” [Guo Page 5])
learning a (Guo teaches: “Step 108, retraining the supervised risk model based on the filtered (remaining data) positive and negative samples in the training sample set.” [Page 5]. Guo teaches the creation of these models.)
calculating as a score a (Guo teaches: “After the supervised machine learning model is retrained based on the filtered training sample set, the modeler can deploy the model in the wind control system and perform risk assessment on the user's daily online transaction based on the model. The online transaction is scored to obtain a risk score (calculates a score based on the data), and then based on the obtained risk score to determine whether the transaction is risky” [Page 9])
Guo does not distinctly disclose:
performing unsupervised learning to identify first transaction data that is not fraudulent transaction data, among the transaction data
Guo does not distinctly disclose applying unsupervised learning, however Guo does teach identifying transaction data that is not fraudulent.
However, Malini teaches unsupervised learning being the way that (Malini teaches on Page 5: “unsupervised outlier detection” in which the normal transaction data is represented as the non-outlier data and the fraudulent data is the outlier data.)
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to combine the feature selection of Guo with the unsupervised learning method taught by Malini so that the system can learn to handle finding new trends in the data. (Malini teaches this on Page 4: “In outlier detection method unsupervised learning is preferred to detect the fraud because it can lead to new explanations and representation of the observation data.”)
Guo as modified by Malini does not disclose:
a hierarchical mixed model.
However, Zoldi teaches a hierarchical mixed model: Fig. 3 and paragraph 0042 of Malini teaches a decision tree structure ([0021] of the applicant’s specification describes the mixed model being represented by a tree structure) for use as a trained model.
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to replace the model taught by Guo as modified by Malini with the tree model and analysis taught by Zoldi so that the model can partition the data into small chunks and data analysis can be performed on smaller separate classified datasets. (Zoldi teaches in 0086: “As described above, the decision tree partitions the entire dataset into small chunks, depending on its intrinsic feature distribution, and the dataset is distributed onto many nodes so that the clustering is performed on the small partitioned datasets.”)
Guo as modified by Malini and further modified by Zoldi does not distinctly disclose:
second hierarchical mixed model.
However, Marcjan teaches the use of a second model in paragraph 0089, in which the newly labeled data is passed into a second trained risk determination model.
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to combine the hierarchical mixed model taught by Guo as modified by Malini and further modified by Zoldi with the method of creating a second model as taught by Marcjan to yield the predictable result of allowing the system to evaluate one set of data in different ways as each model could compare different elements of the data.
Guo as modified by Malini, Zoldi, and Marcjan does not distinctly disclose:
a ratio of target data for which the training data is discriminated
However, Zoldi teach in another embodiment: a ratio of target data for which the training data is discriminated. (Zoldi teaches: “taking the ratio of the count of number of fraud samples to the total number of all the samples.”[0090].)
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to combine the calculation of the score as taught by Guo as modified by Malini, Zoldi, and Marcjan with the ratio calculation of this embodiment of Zoldi obtain the predictable result of allowing the system to analyze and create a class probability of each node in the system. (This is taught by Zoldi in paragraph 0090, which shows taking the ratio of the count of number of fraud samples to the total number of all the samples to obtain the left-over class probability of the node.)
Guo as modified by Malini, Zoldi, and Marcjan does not distinctly disclose:
visualizing, on a display device, the ratio of the target data aggregated for each score to the entire target data, by displaying on the display device both a first graphic representing all target data with a size according to a total number of target data, and a second graphic representing target data according to a number of data of the target data for a corresponding score,
However, Carcillo teaches: visualizing, on a display device, the ratio of the target data aggregated for each score to the entire target data, by displaying on the display device both a first graphic representing all target data with a size according to a total number of target data, and a second graphic representing target data according to a number of data of the target data for a corresponding score, (Carcillo teaches Fig. 4 on page 8 which shows a visualization of the ratio of the fraud scores to the entire target data (a) and the second graphic representing target data according to a number of data of the target data for a corresponding score (d)).
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to combine the analysis method of Guo as modified by Malini, Zoldi, and Marcjan with the visualization step of Carcillo for the known result of being able to compare the results of the machine learning model.
Guo as modified by Malini, Zoldi, and Marcjan does not distinctly disclose:
discriminating the training data classified in a leaf node using a discriminant placed in each leaf node of the first hierarchical mixed model, and calculating, for each leaf node, a ratio of the classified training data that is predicted to be a negative example, and excluding, from the target data, data that satisfies a condition for classification into the leaf node for which the calculate ratio is equal to or greater than a predetermined threshold.
However, Zoldi teaches:
discriminating the training data classified in a leaf node using a discriminant placed in each leaf node of the first hierarchical mixed model, (Zoldi teaches: “Specifically, a clustering approach is applied to the training samples in a leaf node (or decision node) and to group the training samples into a plurality of subsets (clusters). New data traverse the tree and will be classified by determining the memberships (discriminant) to each cluster and the characteristics of the cluster to improve the predictability of the decision tree model.” [0010].),
calculating, for each leaf node, a ratio of the classified training data that is predicted to be a negative example, (Zoldi teaches Fig.3 which teaches that for each Node a Pf score (fraud probability) is calculated, which represents the ratio of the classified training data using the data predicted to be a negative example (nonfraud).),
and excluding, from the target data, data that satisfies a condition for classification into the leaf node for which the calculate ratio is equal to or greater than a predetermined threshold. (Zoldi teaches this limitation: “A high fidelity probability region (i.e., in the vicinity of the cluster centroid) can be found from the class probability distribution using a threshold probability, and this radius demarcates two regions of different detection capabilities. [..] If the new data falls into a high-fidelity central region (condition), the threshold probability is assigned to the new data. Otherwise the left-over class probability on the node is assigned to the new data.“ [0033])
The motivation for the combination is the same as the one found earlier in the rejection.
Regarding Claim 8, Guo teaches:
A non-transitory computer readable information recording medium storing a fraud detection program applied to a computer which detects a fraudulent transaction in an operation of a financial institution, when executed by a processor, that performs a method for: (“Corresponding to the above method embodiments, the present specification also provides an embodiment of an electronic device. The electronic device includes a processor and a memory for storing machine executable instructions; wherein the processor and the memory are typically interconnected by an internal bus.” [Guo Page 11])
obtaining transaction data: (Guo teaches: “Marking the sample carrying the risk label in the training sample” [Page 1] in which the training sample is the transaction data.)
performing (“In this specification, the modeler can collect historical transaction data of a large number of users in advance, extract transaction features (extract target data) of several dimensions from the collected historical transaction data as modeling features, and construct feature vectors as training samples based on these modeling features. Then, based on these training samples, a feature matrix is constructed as a training sample set. “[Guo Page 5]. Guo teaches that the data is identified then extracted based on the modeling features that represent fraudulent transaction data.)
learning a first hierarchical mixed model using training data, among the target data, which takes positive examples for the data indicating the fraudulent transaction and takes negative examples for the remaining data other than the positive examples; (“Step 102: Mark the sample carrying the risk tag in the training sample set as a positive sample, and the sample not carrying the risk tag as a negative sample;
Step 104: Train the supervised risk (model) based on the positive and negative samples of the tag, and perform risk assessment on the negative sample separately based on the model that is completed by the training to obtain a risk score” [Guo Page 5])
excluding, from the target data, the target data which is set as negative example training data and is classified as a negative example by the (“Step 106: Filter (exclude) a negative sample in the training sample set that the risk score is greater than a preset threshold” [Guo Page 5])
learning a (Guo teaches: “Step 108, retraining the supervised risk model based on the filtered (remaining data) positive and negative samples in the training sample set.” [Page 5]. Guo teaches the creation of these models.)
calculating as a score a (Guo teaches: “After the supervised machine learning model is retrained based on the filtered training sample set, the modeler can deploy the model in the wind control system and perform risk assessment on the user's daily online transaction based on the model. The online transaction is scored to obtain a risk score (calculates a score based on the data), and then based on the obtained risk score to determine whether the transaction is risky” [Page 9])
Guo does not distinctly disclose:
performing unsupervised learning to identify first transaction data that is not fraudulent transaction data, among the transaction data
Guo does not distinctly disclose applying unsupervised learning, however Guo does teach identifying transaction data that is not fraudulent.
However, Malini teaches unsupervised learning being the way that (Malini teaches on Page 5: “unsupervised outlier detection” in which the normal transaction data is represented as the non-outlier data and the fraudulent data is the outlier data.)
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to combine the feature selection of Guo with the unsupervised learning method taught by Malini so that the system can learn to handle finding new trends in the data. (Malini teaches this on Page 4: “In outlier detection method unsupervised learning is preferred to detect the fraud because it can lead to new explanations and representation of the observation data.”)
Guo as modified by Malini does not disclose:
a hierarchical mixed model.
However, Zoldi teaches a hierarchical mixed model: Fig. 3 and paragraph 0042 of Malini teaches a decision tree structure ([0021] of the applicant’s specification describes the mixed model being represented by a tree structure) for use as a trained model.
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to replace the model taught by Guo as modified by Malini with the tree model and analysis taught by Zoldi so that the model can partition the data into small chunks and data analysis can be performed on smaller separate classified datasets. (Zoldi teaches in 0086: “As described above, the decision tree partitions the entire dataset into small chunks, depending on its intrinsic feature distribution, and the dataset is distributed onto many nodes so that the clustering is performed on the small partitioned datasets.”)
Guo as modified by Malini and further modified by Zoldi does not distinctly disclose:
second hierarchical mixed model.
However, Marcjan teaches the use of a second model in paragraph 0089, in which the newly labeled data is passed into a second trained risk determination model.
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to combine the hierarchical mixed model taught by Guo as modified by Malini and further modified by Zoldi with the method of creating a second model as taught by Marcjan to yield the predictable result of allowing the system to evaluate one set of data in different ways as each model could compare different elements of the data.
Guo as modified by Malini, Zoldi, and Marcjan does not distinctly disclose:
a ratio of target data for which the training data is discriminated
However, Zoldi teach in another embodiment: a ratio of target data for which the training data is discriminated. (Zoldi teaches: “taking the ratio of the count of number of fraud samples to the total number of all the samples.”[0090].)
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to combine the calculation of the score as taught by Guo as modified by Malini, Zoldi, and Marcjan with the ratio calculation of this embodiment of Zoldi obtain the predictable result of allowing the system to analyze and create a class probability of each node in the system. (This is taught by Zoldi in paragraph 0090, which shows taking the ratio of the count of number of fraud samples to the total number of all the samples to obtain the left-over class probability of the node.)
Guo as modified by Malini, Zoldi, and Marcjan does not distinctly disclose:
visualizing, on a display device, the ratio of the target data aggregated for each score to the entire target data, by displaying on the display device both a first graphic representing all target data with a size according to a total number of target data, and a second graphic representing target data according to a number of data of the target data for a corresponding score,
However, Carcillo teaches: visualizing, on a display device, the ratio of the target data aggregated for each score to the entire target data, by displaying on the display device both a first graphic representing all target data with a size according to a total number of target data, and a second graphic representing target data according to a number of data of the target data for a corresponding score, (Carcillo teaches Fig. 4 on page 8 which shows a visualization of the ratio of the fraud scores to the entire target data (a) and the second graphic representing target data according to a number of data of the target data for a corresponding score (d)).
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to combine the analysis method of Guo as modified by Malini, Zoldi, and Marcjan with the visualization step of Carcillo for the known result of being able to compare the results of the machine learning model.
Guo as modified by Malini, Zoldi, and Marcjan does not distinctly disclose:
discriminating the training data classified in a leaf node using a discriminant placed in each leaf node of the first hierarchical mixed model, and calculating, for each leaf node, a ratio of the classified training data that is predicted to be a negative example, and excluding, from the target data, data that satisfies a condition for classification into the leaf node for which the calculate ratio is equal to or greater than a predetermined threshold.
However, Zoldi teaches:
discriminating the training data classified in a leaf node using a discriminant placed in each leaf node of the first hierarchical mixed model, (Zoldi teaches: “Specifically, a clustering approach is applied to the training samples in a leaf node (or decision node) and to group the training samples into a plurality of subsets (clusters). New data traverse the tree and will be classified by determining the memberships (discriminant) to each cluster and the characteristics of the cluster to improve the predictability of the decision tree model.” [0010].),
calculating, for each leaf node, a ratio of the classified training data that is predicted to be a negative example, (Zoldi teaches Fig.3 which teaches that for each Node a Pf score (fraud probability) is calculated, which represents the ratio of the classified training data using the data predicted to be a negative example (nonfraud).),
and excluding, from the target data, data that satisfies a condition for classification into the leaf node for which the calculate ratio is equal to or greater than a predetermined threshold. (Zoldi teaches this limitation: “A high fidelity probability region (i.e., in the vicinity of the cluster centroid) can be found from the class probability distribution using a threshold probability, and this radius demarcates two regions of different detection capabilities. [..] If the new data falls into a high-fidelity central region (condition), the threshold probability is assigned to the new data. Otherwise the left-over class probability on the node is assigned to the new data.“ [0033])
The motivation for the combination is the same as the one found earlier in the rejection.
Claim(s) 1, 3, 5, 6, 8 and 10 is/are rejected under 35 U.S.C. 103 as being unpatentable over Guo (International Application WO2019100844A1) in view of Malini’s Analysis on credit card fraud identification techniques based on KNN and outlier detection and further in view of Zoldi (U.S. Publication US20170083920A1) and in view of Marcjan (U.S. Publication US20190130406A1) and finally in view of Carcillo’s Streaming active learning strategies for real-life credit card fraud detection: assessment and visualization and finally in view of Xu EnsembleLens: Ensemble-based Visual Exploration of Anomaly Detection Algorithms with Multidimensional Data
Regarding Claim 10, Guo as modified by Malini, Zoldi, Marcjan, and Carcillo teaches all the limitations of Claim 1, but does not teach: The fraud detection device according to claim 1, wherein the processor further executes instructions to visualize the score by changing colors according to the size of the score.
However, Xu teaches figure 6, which visualizes score by changing colors depending on the size of the score.
PNG
media_image1.png
395
669
media_image1.png
Greyscale
Before the effective filing date of the invention it would have been obvious to one of ordinary skill in the art to modify the graph generation of Guo as modified by Malini, Zoldi, Marcjan, and Carcillo with the graph color scheme of Xu for the improvement of viewing outliers and variations in features more effectively. This improvement is taught by Xu: “Outlier scores are encoded differently” [Page 7] which describe the intention of viewing outliers as well as “we use different columns to depict different features of raw data.” [Page 7].
Conclusion
THIS ACTION IS MADE FINAL. 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 Jonathan M Bakhit whose telephone number is (571)272-0454. The examiner can normally be reached Monday - Thursday 8:00AM - 6:00PM EST.
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, Alexey Shmatov can be reached on (571) 270-3428. 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
J.M.B.
Examiner
Art Unit 2123
571-272-1000.
/ALEXEY SHMATOV/ Supervisory Patent Examiner, Art Unit 2123
Prosecution Insights