Multimodal Fraud Detection via Intermediary Settlement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional fraud detection processes in electronic payment systems are inadequate in addressing the scale of compromised digital accounts and sophisticated socially engineered fraud schemes, often failing to prevent fraudulent transactions until after they have been completed.
Innovation Solution
A multimodal fraud detection system that creates comprehensive customer profiles with predictive identity behavior models, intercepts transactions, and uses machine learning techniques to quantify behavior conformance, generating a predicted fraud score to intervene and prevent fraudulent transactions before settlement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional fraud detection processes are used, then transaction security is improved, but transaction settlement time is delayed and efficiency is reduced
Solution Approach 1:
The system performs preliminary fraud detection actions by creating customer profiles with expected behavior models before transactions occur. The machine learning models continuously learn from historical data to establish baseline behavior patterns, enabling the system to detect anomalies proactively rather than reactively after fraud occurs.
Solution Approach 2:
The patent introduces an intermediary fraud detection layer between the customer and the core banking services. This intermediary system intercepts transactions, evaluates them against the customer's behavior model, and only allows legitimate transactions to proceed to core services, thereby preventing fraud without blocking legitimate transactions.
2Reliability
If robust fraud prevention measures are implemented, then transaction security is improved, but transaction settlement time is increased
Solution Approach 1:
The system enables self-service fraud detection by automatically comparing each transaction against the customer's learned behavior model without requiring manual review. The machine learning models continuously update themselves with new transaction data, allowing the system to adapt to changing customer behaviors autonomously and make real-time fraud determination decisions.
3Reliability
If traditional in-person fraud detection models are used, then basic transaction security is maintained, but sophisticated digital fraud schemes cannot be detected
Solution Approach 1:
The system changes the detection parameters from traditional in-person verification metrics to digital behavior pattern analysis. The machine learning models monitor numerous digital parameters including transaction timing, amount patterns, device information, and behavioral sequences to detect sophisticated digital fraud schemes that traditional models cannot identify.
Data Source
AI summary
In one example, the disclosed multimodal fraud prevention techniques are employed by an intermediary settlement platform. The intermediary settlement platform generally monitors transaction data associated with various payment rails and provides new store and forward functions, which include receiving new customer transactions before they reach core services associated with a Financial Institution (FI); executing new transactions as predicted transactions; extracting behavior metrics from the predicted transactions; quantifying predicted fraud prior to settlement (e.g., before transferring funds) using comprehensive customer identity behavior models; and performing fraud interventions based on the same. In this example, the intermediary settlement platform creates the customer identity behavior models and transforms the customer's transaction data into comprehensive behavior metrics according to the dimensions of the respective model; quantifies the customer behaviors for new transactions; and determines the degree of predicted behavior conformance between the new transaction and non-fraudulent customer behavior metrics.