Siamese Network Authentication via User Interaction Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current authentication methods in server systems fail to detect and prevent unauthorized access when an unauthorized user gains physical access to an authorized user's device, leading to security and functionality issues, particularly in scenarios like 'friendly fraud' where a child or friend uses the authorized user's device without their knowledge.
Innovation Solution
Implementing machine learning techniques to authenticate users based on their interaction patterns with client devices, such as smartphones or laptops, by generating encoding values from interaction data and comparing them to reference values from prior sessions to determine similarity scores, thereby authenticating authorized users and detecting unauthorized access.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If traditional authentication methods (username/password) are used, then users can access services conveniently, but security is compromised when credentials are discovered by malicious third-parties
Solution Approach 1:
The authentication process is segmented into multiple independent factors: something you know (credentials) and something you do (interaction patterns). This segmentation prevents a single point of failure, as malicious third-parties would need to compromise both credential storage and replicate genuine user interaction patterns to gain unauthorized access.
Solution Approach 2:
Machine learning models serve as an intermediary layer between credential verification and service access. The model analyzes interaction patterns as a mediator factor, comparing current session behavior against learned user patterns to determine whether to grant access, even when credentials are valid.
2Reliability
If machine learning analysis of interaction patterns is implemented, then security is improved by detecting unauthorized access, but device complexity increases
Solution Approach 1:
The system performs self-service by automatically collecting interaction data, training machine learning models on user patterns, and executing authentication decisions without requiring manual configuration or intervention. The model continuously learns and adapts to user behavior patterns autonomously, reducing operational complexity despite the sophisticated security mechanisms.
Solution Approach 2:
The interaction analysis system serves multiple functions: it collects data for security authentication, builds user profiles for personalization, and detects anomalous behavior patterns. This multi-functionality justifies the added complexity by providing comprehensive security and enhanced user experience from a single integrated system.
3Reliability
If interaction data is collected and analyzed, then unauthorized access is detected, but loss of information occurs through potential privacy concerns
Solution Approach 1:
The system applies local quality by analyzing interaction patterns at the granularity of specific device interactions (swipe gestures, typing rhythms, navigation patterns) rather than collecting broad personal information. The machine learning models process localized behavioral data points that are sufficient for authentication purposes while minimizing intrusion into private user information.
Data Source
AI summary
Techniques are disclosed relating to machine learning techniques for performing user authentication based on the manner in which a user interacts with a client device, including the use of Siamese networks to detect unauthorized use of a device and/or account. In some embodiments, a server system may receive a request to authorize a transaction associated with a user account. The request may include transaction details and, separate from those transaction details, interaction data indicative of a manner in which a requesting user interacts with a client device during a user session. The server system may apply a machine learning model to the interaction data to create an encoding value that is based on the manner in when the requesting user interacts with the client device during the user session. The server system may then compare the encoding value to a reference encoding value and, based on the comparison, determine whether to authorize the transaction.