Dynamic Control Group Generation for Fraud-Resistant Transactions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional control groups in transaction processing systems become less representative over time and may include fraudulent transactions, leading to increased loss and inefficiency due to biased selection models.
Innovation Solution
A machine learning technique using a Dragonnet combined with a VAE to automatically generate and update control groups by learning the feature distribution of transactions, weighting low-dollar transactions, and adjusting the control group to ensure representativeness and reduce bias.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional control groups are used in transaction processing systems, then the system can process transactions with a baseline comparison group, but the control group becomes less representative over time and may include fraudulent transactions, leading to increased loss and reduced accuracy in detecting suspicious behavior
Solution Approach 1:
The patent implements dynamic control group generation that automatically adapts to changing transaction patterns over time. The system continuously updates the control group composition based on learned feature distributions, ensuring the control group remains representative of the current transaction population rather than becoming static and outdated.
Solution Approach 2:
The system employs feedback mechanisms where the performance and representativeness of the control group are continuously evaluated. Based on this feedback, the system automatically adjusts and regenerates the control group to maintain its effectiveness in detecting fraudulent transactions and suspicious behavior patterns.
2Measurement precision
If control groups include fraudulent transactions for representative purposes, then the control group accurately reflects the overall transaction population, but the system incurs financial loss by processing these fraudulent transactions
Solution Approach 1:
The patent applies local quality by differentiating between control group transactions used for training/testing and actual transaction processing. The system creates specialized control groups with known characteristics that are used exclusively for model evaluation, while maintaining separate processing paths that prevent fraudulent transactions from causing financial loss.
Solution Approach 2:
The system segments transactions into distinct categories: control group transactions for model evaluation and regular transactions for processing. This segmentation allows fraudulent transactions to be included in the control group for accuracy measurement without exposing the financial system to their harmful effects.
3Device complexity
If the control group selection model introduces bias, then the model may oversimplify transaction patterns, but this reduces the system's ability to detect complex fraudulent behavior
Solution Approach 1:
The patent implements dynamic model complexity adjustment where the selection model adapts its complexity based on the specific transaction context and fraud detection needs. The system can transition between simpler models for normal operations and more complex models when detecting sophisticated fraudulent patterns, optimizing both efficiency and accuracy.
Data Source
AI summary
Techniques are disclosed for automatically generating and updating a control group. In disclosed techniques, a server computer system trains, using a plurality of transactions, a machine learning model. During training the machine learning model learns a feature distribution of both a current set of control group (CG) transactions and a current set of non-control group (non-CG) transactions included in the plurality of transactions. The system inputs the current set of CG transactions into the trained machine learning model. Based on the output of the trained machine learning model for the current set of CG transactions, the system modifies the current set of CG transactions to generate an updated set of CG transactions. Based on the updated set of CG transactions, the server performs one or more preventative measures for a transaction processing system. The disclosed techniques may advantageously improve the accuracy e.g., of a transaction processing system.