Multi-Stage Fraud Risk Detection With Reinforcement Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods and systems for preventing fraudulent electronic payments lack holistic integration across pre-authorization and post-authorization stages, leading to inefficiencies and increased operational latencies, as they rely on separate modeling systems without communication and fail to utilize post-authorization data for enhanced detection.
Innovation Solution
Implementing a policy-based reinforcement learning risk decision agent that integrates pre-authorization and post-authorization data using Markov chain modeling to determine the value of impropriety, enabling real-time blocking or facilitation of transactions based on a unified assessment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If separate modeling systems are used for pre-authorization and post-authorization stages, then system modularity is maintained, but detection precision and operational efficiency deteriorate due to lack of holistic integration
Solution Approach 1:
The patent merges separate pre-authorization and post-authorization modeling systems into a unified reinforcement learning agent that processes transaction data across all stages. This integration allows the system to maintain modularity benefits while achieving holistic fraud detection by combining data from multiple authorization stages into a single decision-making framework.
2Device complexity
If separate modeling systems are used for pre-authorization and post-authorization stages, then system simplicity is maintained, but productivity deteriorates due to increased operational latencies
Solution Approach 1:
The reinforcement learning agent performs preliminary risk assessment by analyzing pre-authorization data early in the transaction flow. This allows the system to identify high-risk transactions before they proceed to post-authorization stages, reducing overall processing time and improving productivity by preventing fraudulent transactions early rather than detecting them later.
3Device complexity
If post-authorization data is not utilized for detection, then data processing complexity is reduced, but detection precision deteriorates due to failure to leverage enhanced detection opportunities
Solution Approach 1:
The system implements feedback loops where post-authorization transaction data is fed back into the reinforcement learning agent. This feedback mechanism allows the agent to continuously learn from actual transaction outcomes and refine its risk assessment model, improving detection precision by leveraging information from the entire transaction lifecycle rather than just pre-authorization data.
Data Source
AI summary
The technology described herein relates to systems, methods, and computer storage media, among other things, for determining whether an electronic transmission (e.g., associated with an electronic payment transaction) should be blocked (e.g., based on being a fraudulent transaction). In embodiments, a policy-based reinforcement learning risk decision agent is used to make these determinations for a plurality of stages associated with the electronic payment transaction (e.g., a pre-authorization stage, a post-authorization stage, and a delay-captured stage). The policy-based reinforcement learning risk decision agent can be trained using previous electronic payment transaction data for previous electronic payment transactions. For example, this particular agent can be trained using pre-authorization electronic payment transaction data, post-authorization electronic payment transaction data, and delay-captured electronic payment transaction data for each of the previous electronic payment transactions.