# US 10,503,971 B1 · Platform for document classification

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 10,503,971 B1
- Document ID: US10503971B1
- Status: Granted
- Kind: B1
- Application number: 16/540,287
- Filed: 2019-08-14
- Granted: 2019-12-10
- Earliest claimed: 2019-03-19
- Assignees: Capital One Services, LLC
- Applicants: Capital One Services, LLC
- Inventors: Steven Dang; Jason Gould; Jennifer Jiang; Christopher Akatsuka; Douglas Slattery; Vijaya Pasam
- Web page: https://patentgrep.com/patent/US10503971B1

## CPC

- G06V 30/413: Classification of content, e.g. text, photographs or tables
- G06F 16/353: into predefined classes
- G06F 16/93: Document management systems
- G06F 17/18: for evaluating statistical data {, e.g. average values, frequency distributions, probability functions, regression analysis (forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04)}
- G06F 18/2431: Multiple classes
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/09: Supervised learning
- G06N 20/00: Machine learning
- G06N 20/20: Ensemble learning
- G06Q 10/10: Office automation; Time management
- G06Q 40/03: Credit; Loans; Processing thereof
- G06V 10/764: using classification, e.g. of video objects
- G06N 3/045: Combinations of networks
- G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
- G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks
- G06N 20/10: using kernel methods, e.g. support vector machines [SVM]
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR

## Abstract

A device obtains image data associated with a document. Using a first machine learning model, the device determines, for the document, a first classification of one of a plurality of document types and a first confidence score associated with the first classification, and a second classification of one of the plurality of document types and a second confidence score associated with the second classification based on the image data. The device determines a difference between the first confidence score and the second confidence score, compares the difference and a threshold value, and accept the first classification of the document when the difference satisfies the threshold value.

## Claims

### Claim 1

A method, comprising:
determining, by a processor, for a document, and using a first machine learning model, a first classification of a document type, of a plurality of document types, and a first confidence score associated with the first classification,
the first confidence score indicating a confidence level that the document type corresponds to the first classification;
comparing, by the processor, the first confidence score and a first threshold value;
selectively:
accepting, by the processor, the first classification of the document type when the first confidence score satisfies the first threshold value; or
obtaining, by the processor, information associated with the document when the first confidence score fails to satisfy the first threshold value;
determining, by the processor, for the document, and using a second machine learning model, a second classification of the document type and a second confidence score associated with the second classification,
the second confidence score indicating a confidence level that the document type corresponds to the second classification;
comparing, by the processor, the second confidence score and a second threshold value; and
accepting, by the processor, the second classification of the document type when the second confidence score satisfies the second threshold value.

### Claim 2 (depends on 1)

The method of claim 1, further comprising:
receiving a multiple-page document;
using a splitting technique to form multiple single-page documents from the multiple-page document; and
determining the document from the multiple single-page documents.

### Claim 3 (depends on 1)

The method of claim 1, wherein the document contains at least two pages and the at least two pages contain different document types of the plurality of document types.

### Claim 4 (depends on 1)

The method of claim 1, further comprising:
training the first machine learning model to determine the first confidence score based on historical data associated with a plurality of confidence scores of accurate document classifications, and
wherein the first confidence score is determined based on training the first machine learning model.

### Claim 5 (depends on 1)

The method of claim 1, further comprising:
preprocessing the document by one or more of:
an image contrast adjustment,
an image noise filter,
an image histogram modification,
cropping an image of the document, or
a skew correction operation; and
wherein the second classification is determined based on the preprocessing.

### Claim 6 (depends on 1)

The method of claim 1, further comprising:
training the second machine learning model using information that identifier a plurality of textual features,
the textual features includes one or more of:
words,
phrases, or
clauses; and
wherein the second classification is determined based on training the second machine learning model.

### Claim 7 (depends on 1)

The method of claim 1, further comprising:
assigning a label after accepting the second classification of the document,
the label being one or more of:
a file name,
a tag,
a flag,
a location in directory system, or
an entry in a log or database.

### Claim 8

A device, comprising:
a memory; and
one or more processors to:
determine, for a document, a first classification of a document type, of a plurality of document types, and a first confidence score associated with the first classification,
the first confidence score indicating a confidence level that the document type corresponds to the first classification;
compare the first confidence score and a first threshold value;
determine for the document, a second classification of the document type and a second confidence score associated with the second classification when the first confidence score fails to satisfy the first threshold value,
the second confidence score indicating a confidence level that the document type corresponds to the second classification;
compare the second confidence score and a second threshold value; and
accept the second classification of the document type when the second confidence score satisfies the second threshold value.

### Claim 9 (depends on 8)

The device of claim 8, wherein the one or more processors are further to:
provide the document and information regarding the second classification to another device.

### Claim 10 (depends on 8)

The device of claim 8, wherein the first classification is determined using one or more:
deep learning or
convolutional neural networks.

### Claim 11 (depends on 8)

The device of claim 8, wherein the first confidence score is greater than the second confidence score.

### Claim 12 (depends on 8)

The device of claim 8, wherein the first classification is one of:
a paystub,
a personal financial document, or
an income reporting form.

### Claim 13 (depends on 8)

The device of claim 8, wherein the one or more processors are further to:
train a machine learning model to determine the first confidence score based on features of another document of a particular document type, and
wherein the first confidence score is determined based on the trained machine learning model.

### Claim 14 (depends on 8)

The device of claim 8, wherein the one or more processors are further to:
train a machine learning model to use a computer vision technique to:
identify a plurality of features of the document, and
perform a dimensionality reduction to reduce the plurality of features to a particular feature set; and
wherein the first classification is determined based on the trained machine learning model.

### Claim 15

A non-transitory computer-readable medium storing instructions, the instructions comprising:
one or more instructions that, when executed by one or more processors, cause the one or more processors to:
determine, for a document, a first classification of a document type, of a plurality of document types, and a first confidence score,
the first confidence score indicating a confidence level that the document corresponds to the first classification;
compare the first confidence score and a first threshold value;
selectively:
accept the first classification of the document type when the first confidence score satisfies the first threshold value; or
obtain information associated with the document when the first confidence score fails to satisfy the first threshold value;
determine, for the document, a second classification of the document type and a second confidence score,
the second confidence score indicating a confidence level that the document corresponds to the second classification;
compare the second confidence score and a second threshold value; and
accept the second classification of the document type when the second confidence score satisfies the second threshold value.

### Claim 16 (depends on 15)

The non-transitory computer-readable medium of claim 15, wherein the first classification is determined using one or more:
deep learning or
convolutional neural networks.

### Claim 17 (depends on 15)

The non-transitory computer-readable medium of claim 15, wherein the first confidence score is greater than the second confidence score.

### Claim 18 (depends on 15)

The non-transitory computer-readable medium of claim 15, wherein the one or more instructions when executed by the one or more processors, further cause the one or more processors to:
train a machine learning model to determine the first confidence score based on features of another document of a particular document type; and
wherein the first confidence score is determined based on the trained machine learning model.

### Claim 19 (depends on 15)

The non-transitory computer-readable medium of claim 15, wherein the one or more instructions when executed by the one or more processors, further cause the one or more processors to:
train a machine learning model to use a computer vision technique to:
identify a plurality of features of the document, and
perform a dimensionality reduction to reduce the plurality of features to a particular feature set; and
wherein the first classification is determined based on the trained machine learning model.

### Claim 20 (depends on 15)

The non-transitory computer-readable medium of claim 15, wherein the one or more instructions when executed by the one or more processors, further cause the one or more processors to:
receive a multiple-page document;
use a splitting technique to form multiple single-page documents from the multiple-page document; and
determine the document from the multiple single-page documents.