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 .
DETAILED ACTION
The instant application having Application No. 18428925 has a total of 20 claims pending in the application, all of which are ready for examination by the examiner.
I. ACKNOWLEDGEMENT OF REFERENCES CITED BY APPLICANT
Information Disclosure Statement
As required by M.P.E.P 609(c), the applicant’s submissions of the Information Disclosure Statement dated 2/1/24 is acknowledged by the examiner and the cited references have been considered in the examination of the claims now pending. As required by M.P.E.P 609 C(2), a copy of the PTOL-1449 initialed and dated by the examiner is attached to the instant office action.
II. REJECTIONS NOT BASED ON PRIOR ART
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 2-4, 9-11, and 16-18 are 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.
As per claims 2, 9 and 16, these claims call for “a reconstruction loss function based on the synthetic training data samples and the real training data sample.” However, in the previous claim, the synthetic training data samples are being created by the generator, which receives input from the encoder. How can the encoder be trained on something coming out of the generator, when the data it needs to operate (i.e. be trained from) is being created based on output from that same encoder? As far as the Examiner can tell this is circular logic. In order to operate the generator, you need to receive input from the encoder. But to train the encoder, you need output from the generator. How can the generator generate output if the encoder isn’t trained? How can the encoder be trained on output from the generator if it has not been trained to produce the appropriate output? This situation seems mutually exclusive, and causes the claim to be confusing. Therefore the claims are rejected under U.S.C. 112(b) for failing to particularly point out and claim the intended invention.
As per claims 3, 10, and 17, these claims call for the generator to be trained on a classification loss function of the classifier model. However, the Classifier model requires data from the generator in order to operate. As with the encoder, how can the generator be creating data for the Classifier to classify if the generator has not been trained? How can the classifier be providing a loss function result if the generator has not yet been trained? This appears to be mutually exclusive, and therefore rejected under U.S.C. 112(b) for failing to particularly point out and claim the intended invention.
As per claims 4, 11, and 18, these claims are rejected as being dependent on a claim rejected under U.S.C. 112(b).
III. REJECTIONS BASED ON PRIOR ART
Examiners Note: Some rejections will be followed by an ‘EN’ that will denote an examiners note. This will be placed to further explain a rejection.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-2, 6-9, 13-16, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Liu et al (“Alike and Unlike: Resolving Class Imbalance Problem in Financial Credit Risk Assessment”) in view of Chen et al (“Variational Autoencoders and Wasserstein Generative Adversarial Networks for Improving the Anti-Money Laundering Process”).
As per claims 1, 8, and 15, Liu discloses, “a system comprising” (abstract; EN: This denotes the overall system).
“collect a set of real training data samples” (Pg.2126-2127, particularly section 3.1; EN: this denotes the real data). “wherein the real data samples include feature data associated with health class indicators, wherein the health class indicators are indicative of a likelihood of a supplier to continue using a service provided by a service provider” (Pg.2126-2127, particularly section 3.1; EN: this denotes features associated with high risk (i.e. not going to continue service of having a loan) vs low risk (continuing service) with the loan service provider).
“transform the set of real training data samples into sample encodings using an encoder” (pg.2126 section 2.2; EN: this denotes preparing the data with an autoencoder).
”generate a set of synthetic training data samples using a generator and the sample embeddings” (Pg.2126, particularly section 2.3; EN: this denotes the autoencoder material going into the generator, and being used to create synthetic representations of the data).
“generate discrimination feedback data of the real training data samples and the synthetic training data samples using a discriminator” (Pg.2126, particularly section 2.3; EN: this denotes the discriminators being used to determine what is real and what isn’t).
“Train the generator using the generated discrimination feedback data” (Pg.2126, particularly section 2.3; EN: this denotes the discriminators being used to determine what is real and what isn’t and being used to optimize the generator (i.e. train)).
“… suppliers with health class indicators…” (Pg.2126-2127, particularly section 3.1; EN: this denotes features associated with high risk (i.e. not going to continue service of having a loan) vs low risk (continuing service) with the loan service provider).
However, Liu fails to explicitly disclose, “A processor; and a memory comprising computer program code, the memory and the computer program code configured to, with the processor, cause the processor to”, “train a classifier model to classify … using the set of synthetic training data samples”
Chen discloses, “A processor; and a memory comprising computer program code, the memory and the computer program code configured to, with the processor, cause the processor to” (Abstract; EN: this denotes the use of machine learning, which will inherently include some sort of processor and memory to execute the machine learning).
“train a classifier model to classify … using the set of synthetic training data samples” (Pg.83776 particularly section F. Performance Evaluation Methods” EN: This denotes using synthetic data in order to train a classifier to deal with financial data).
Liu and Chen are analogous art because both involve synthetic financial data.
Before the effective filing date it would have been obvious to one skilled in the art of synthetic financial data to combine the work of Liu and Chen in order to make use of the synthetic data for training a classifier.
The motivation for doing so would be because “Experimental results demonstrated that the False Positive Rate (FPR) drops down to as low as 7% in the proposed multi-loss AE model” (abstract) or in the case of Liu, allow the system to use the generated synthetic data to train a classifier to classify that type of data.
Therefore before the effective filing date it would have been obvious to one skilled in the art of synthetic financial data to combine the work of Liu and Chen in order to make use of the synthetic data for training a classifier.
As per claims 2, 9, and 16, Liu discloses, “… to train the encoder using: a divergence loss function based on the …. Encoding distribution… and a reconstruction loss function based on the … real training data samples” (Pg.2126, particularly section 2.2; EN: this denotes training the encoder using a reconstruction loss).
Chen discloses, “wherein the encoder is a Variational Autoencoder (VAE)” (pg.83767, particularly C2; EN: this denotes the use of VAE with financial data).
“wherein the sample encodings include a normalized encoding distribution” (pg.83772, particularly section 6; EN: this denotes normalizing the input before training).
“wherein the memory and the computer program code are configured to, with the processor, further cause the processor to train the encoder using” (Abstract; EN: this denotes the use of machine learning, which will inherently include some sort of processor and memory to execute the machine learning).
“a divergence loss function based on the normalized encoding distribution and a standard normal distribution” (Pg.83775 particularly the Model Architecture for VAE section; EN; this denotes training the divergence loss function based on the input and the standard deviation).
“a reconstruction loss based on the synthetic training data samples and the real training data samples” (Pg.83776 particularly section F. Performance Evaluation Methods” EN: This denotes using synthetic data in order to train a classifier to deal with financial data).
Liu and Chen are analogous art because both involve synthetic financial data.
Before the effective filing date it would have been obvious to one skilled in the art of synthetic financial data to combine the work of Liu and Chen in order to make use of the Variational autoencoders.
The motivation for doing so would be to “encourage the model to generalize better” (Pg.83767, C2, Section B, first paragraph) or in the case of Liu, allow the system to use the VAE in order to have better generalization.
Therefore before the effective filing date it would have been obvious to one skilled in the art of synthetic financial data to combine the work of Liu and Chen in order to make use of the Variational autoencoders.
As per claims 6, 13, and 20 Liu discloses, “receive a set of feature data associated with a target supplier” (pg.2126-2127, particularly section 3.1; EN; this denotes getting input on people to use with the system).
“… target supplier with a class indicator….” (Pg.2126-2127, particularly section 3.1; EN: this denotes features associated with high risk (i.e. not going to continue service of having a loan) vs low risk (continuing service) with the loan service provider).
Chen discloses, “wherein the memory and the computer program code are configured to, with the processor, further cause the processor to” (Abstract; EN: this denotes the use of machine learning, which will inherently include some sort of processor and memory to execute the machine learning).
“classify the … using the trained classifier and the received set of feature data” (Pg.83776 particularly section F. Performance Evaluation Methods” EN: This denotes using the system to make classifications).
As per claims 7 and 14, Liu discloses, “wherein the real training data samples are class-imbalanced in that a difference between a quantity of real training data samples associated with a healthy health class indicator and a quantity of real training data samples associated with an unhealthy health class indicator exceeds a threshold” (Pg.2125, particularly the introduction section; EN: this denotes the data being imbalanced, and solutions to make them balanced, with the threshold being the two sets being out of balance).
“Wherein generating the set of synthetic training data samples includes generating a class-balanced set of synthetic training data samples in that a difference between a quantity of synthetic training data samples associated with a healthy health class indicator and a quantity of synthetic training data samples associated with an unhealthy health class indicator is within he threshold” (Pg.2126-2127, particularly section 3; EN: this denotes the process of balancing the training data using the system).
Claim Rejections - 35 USC § 103
Claims 3-4, 10-11, and 17-18 are rejected under 35 U.S.C. 103 as being unpatentable over Liu et al (“Alike and Unlike: Resolving Class Imbalance Problem in Financial Credit Risk Assessment”) in view of Chen et al (“Variational Autoencoders and Wasserstein Generative Adversarial Networks for Improving the Anti-Money Laundering Process”) and further in view of Bellegarda et al (US 20220391585 A1).
As per claims 3, 10, and 17, Liu discloses, “wherein the generator and discriminator are components of a generative adversarial network” (Pg.2126, particularly Figure 1; EN: this denotes the Generator/Discriminator setup, which makes up a GAN).
“wherein training the generator includes training the generator using:” (pg.2126, particularly section 2.3 and 2.4; EN: This denotes training the generator and discriminator).
“…based on the synthetic training data samples and the real training data samples” (Figure 1 and sections 2.3 and 2.4; EN: this denotes using the generated samples and real samples).
“an adversarial loss function based on the discrimination feedback data of the discriminator…” (Figure 1 and sections 2.3 and 2.4; EN: this denotes using feedback from the discriminator).
“… train the discriminator using the adversarial loss function based on the discrimination feedback data of the discriminator” (Figure 1 and sections 2.3 and 2.4; EN: this denotes using feedback from the discriminator to help optimized the discriminator as well).
Chen discloses, “wherein the memory and the computer program code are configured to, with the processor, cause the processor to…” (Abstract; EN: this denotes the use of machine learning, which will inherently include some sort of processor and memory to execute the machine learning).
However, Chen and Liu fail to explicitly disclose, “Training the generator using … a reconstruction loss function…, an adversarial loss function… and a classification loss function….”
Bellegarda discloses, “Training the generator using … a reconstruction loss function…, an adversarial loss function… and a classification loss function…” (Pg.23, particularly paragraph 0247-0248; EN: this denotes combining reconstruction, classification, and adversarial loss when training the discriminator and generator).
Bellegarda and Liu modified by Chen are analogous art because both involve encoders and GANs
Before the effective filing date it would have been obvious to one skilled in the art of encoders and GANs to combine the work of Bellegarda and Liu modified by Chen in order to make use of combined loss functions.
The motivation for doing so would be to use “a combined loss taking into account aspects of multi-task optimization” (Bellegarda, Pg.23, paragraph 0247) or in the case of Liu modified by Chen, allow the system to take into account the loss function from each aspect when training the GAN of the model.
Therefore before the effective filing date it would have been obvious to one skilled in the art of encoders and GANs to combine the work of Bellegarda and Liu modified by Chen in order to make use of combined loss functions.
As per claims 4, 11, and 18, Liu discloses, “wherein training the generator and training the discriminator further includes training the generator and the discriminator iteratively until an accuracy level of the discrimination feedback data reaches a threshold range” (Pg.2126, particularly section 2.4; EN: this denotes minimizing/maximizing various aspects of the loss function in response to training, with the minimization/maximization being the threshold).
Claim Rejections - 35 USC § 103
Claims 5, 12, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Liu et al (“Alike and Unlike: Resolving Class Imbalance Problem in Financial Credit Risk Assessment”) in view of Chen et al (“Variational Autoencoders and Wasserstein Generative Adversarial Networks for Improving the Anti-Money Laundering Process”) and further in view of Zhu et al (US 20220366231 A1).
As per claims 5, 12, and 19, Liu fails to explicitly disclose, “wherein the classifier model is a graph convolutional network (GCN)-based model” and “Wherein training the classifier model includes training the classifier model using a classification loss function that is a cross-entropy loss function based on classification output data of the classifier model.”
Zhu discloses, “wherein the classifier model is a graph convolutional network (GCN)-based model” and “Wherein training the classifier model includes training the classifier model using a classification loss function that is a cross-entropy loss function based on classification output data of the classifier model” (Pg.6, particularly paragraph 0070; EN: this denotes using GCN for risk analysis and using cross-entropy to train the model).
Zhu and Liu modified by Chen are analogous art because both involve financial machine learning.
Before the effective filing date it would have been obvious to one skilled in the art of financial machine learning to combine the work of Zhu and Liu modified by Chen in order to make use of a GCN trained by cross-entropy.
The motivation for doing so would be because GCN can “demonstrated improved empirical performance of the implemented GCN … in solving climate risk and credit risk problems in finance over conventional uni-relational GCN, providing improvement in prediction accuracy” (Pg.2, particularly paragraph 0024) or in the case of Liu, allow the system to use GCN as a type of neural network to improve prediction accuracy.
Therefore before the effective filing date it would have been obvious to one skilled in the art of financial machine learning to combine the work of Zhu and Liu modified by Chen in order to make use of a GCN trained by cross-entropy.
Conclusion
The examiner requests, in response to this Office action, support be shown for language added to any original claims on amendment and any new claims. That is, indicate support for newly added claim language by specifically pointing to page(s) and line no(s) in the specification and/or drawing figure(s). This will assist the examiner in prosecuting the application.
When responding to this office action, Applicant is advised to clearly point out the patentable novelty which he or she thinks the claims present, in view of the state of the art disclosed by the references cited or the objections made. He or she must also show how the amendments avoid such references or objections See 37 CFR 1.111(c).
Any inquiry concerning this communication or earlier communications from the examiner should be directed to BEN M RIFKIN whose telephone number is (571)272-9768. The examiner can normally be reached Monday-Friday 9 am - 5 pm.
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 at (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 571-272-1000.
/BEN M RIFKIN/ Primary Examiner, Art Unit 2123
Prosecution Insights