Graphical Authentication System Using Deep Learning for Native Language Support

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

There is a need for a system that securely validates users in an electronic network, particularly for non-anglophone users who require authentication in their native language to access resources.

Innovation Solution

A system utilizing graphical authentication credentials, which involves a deep learning model trained on user-provided credentials, using optical character recognition to identify characters and verify matches against stored credentials, allowing or denying access based on authentication success.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If traditional authentication systems are used, then security validation can be performed, but non-anglophone users cannot authenticate in their native language

Engineering Contradiction:

Improvelanguage support for authenticationVSAvoidauthentication security

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The patent introduces an optical character recognition (OCR) system as an intermediary between the user's graphical credential input and the authentication verification system. The OCR engine translates graphical credentials in various languages into machine-readable text, enabling the authentication system to process native language inputs while maintaining security through standardized verification protocols

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent replaces traditional text-based mechanical input systems with graphical credential verification using OCR technology. Instead of requiring users to type or select from predefined language options, the system captures graphical credentials (such as handwritten text or images containing authentication information) and uses optical recognition to extract and verify the content, thereby supporting multiple languages without compromising security

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Adaptability or versatility

If graphical authentication credentials are implemented, then native language support is enabled, but system complexity increases

Engineering Contradiction:

Improvenative language authentication capabilityVSAvoidauthentication system structure

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent implements a universal authentication framework where a single graphical credential capture interface handles multiple languages and input types. The OCR-based verification system serves multiple functions: capturing graphical credentials, recognizing text in various languages, extracting authentication information, and verifying credentials against stored data. This multi-functional approach enables native language support while avoiding the need for separate authentication systems for each language

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system employs self-service mechanisms where the OCR engine automatically captures and processes graphical credentials without requiring manual text entry or language selection by users. The authentication system automatically verifies credentials against stored information, eliminating the need for complex manual verification procedures and reducing system operational complexity while supporting multiple languages

Inventive Principle:

Principle #25Self-service

3Measurement precision

If deep learning models are trained on user credentials, then authentication accuracy improves, but training time and computational resources increase

Engineering Contradiction:

Improvecredential verification accuracyVSAvoidmodel training duration

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary actions by pre-training the deep learning OCR model on extensive datasets of graphical credentials in multiple languages before deployment. This preliminary training establishes a robust baseline accuracy for recognizing various script types and styles. The model is then fine-tuned on user-specific credentials during initial authentication interactions, progressively improving accuracy over time without requiring extensive retraining for each user

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system implements feedback mechanisms where authentication results and verification outcomes are used to continuously refine the deep learning model. Successful and unsuccessful authentication attempts provide feedback signals that help the model learn from real-world usage patterns, improving credential verification accuracy over time. The system adjusts model parameters based on feedback without requiring complete retraining, thereby improving precision while minimizing additional training time

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS20240037210A1System and method for capturing and encrypting graphical authentication credentials for validating users in an electronic network

Publication Date: 2024.02.01 BANK OF AMERICA CORP

AI summary

Embodiments of the present invention provide a system for validating users in an electronic network based on graphical authentication credentials. The system is configured for receiving a file comprising graphical authentication credential from a user device of a user, decrypting the file comprising the graphical authentication credential, loading a deep learning model associated with the user, building a deep learning network using the deep learning model, running the file comprising the graphical authentication credential through the deep learning network, and verifying that the graphical authentication credential matches one or more stored credentials associated with the user based in running the file through the deep learning network.