Semi-Supervised Learning for Fraud Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current predictive models face difficulties in discerning legitimate from fraudulent monetary-related transactions due to unknown targets, as they require fully known training targets to function effectively.
Innovation Solution
The implementation of semi-supervised learning systems and methods that utilize a neural network model with known targets to generate values for unknown targets, combined with outlier detection analysis, to identify and investigate potentially fraudulent activities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If supervised learning models are used with fully known training targets, then model accuracy is improved, but the system cannot handle cases with unknown targets such as fraudulent transactions
Solution Approach 1:
The patent segments the target data into three distinct groups: known good targets, known bad targets, and unknown targets. This segmentation allows the system to apply different processing strategies to each group, enabling supervised learning on known targets while identifying and investigating unknown targets separately, thus resolving the contradiction between requiring known targets for accuracy and needing to handle unknown targets for versatility.
Solution Approach 2:
The patent introduces an intermediary investigation process that bridges supervised learning and unknown target detection. The system generates a list of unknown targets for investigation by comparing data characteristics against known patterns, serving as a mediator that allows the supervised model to indirectly handle unknown targets without requiring their labels, thus maintaining both accuracy and adaptability.
2Reliability
If all unknown targets are investigated manually, then detection completeness is improved, but investigation time and resources increase significantly
Solution Approach 1:
The patent applies local quality by prioritizing investigation resources on specific subsets of unknown targets rather than treating all unknown targets equally. The system identifies and flags only those unknown targets that exhibit characteristics most suspicious or deviate most from known patterns, allowing investigators to focus on high-priority cases and achieving effective detection completeness with reduced time investment.
Solution Approach 2:
The patent implements partial action by investigating a selective subset of unknown targets rather than all of them. The system generates a prioritized list of unknown targets for investigation based on their deviation from known patterns, performing investigation on the most suspicious cases first, thus achieving acceptable detection completeness with significantly reduced investigation time and resources.
Data Source
AI summary
Computer-implemented systems and methods for determining a subset of unknown targets to investigate. For example, a method can be configured to receive a target data set, wherein the target data set includes known targets and unknown targets. A supervised model such as a neural network model is generated using the known targets. The unknown targets are used with the neural network model to generate values for the unknown targets. Analysis with an unsupervised model is performed using the target data set in order to determine which of the unknown targets are outliers. A comparison of list of outlier unknown targets is performed with the values for the unknown targets that were generated by the neural network model. The subset of unknown targets to investigate is determined based upon the comparison.