Fraud Risk Modeling With Hysteresis and Transformers in Real Time

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing fraud detection systems in financial transactions are unable to provide a reliable estimate of fraud risk in real-time, leading to a choice between delaying transactions and bearing an unacceptably high risk of fraud.

Innovation Solution

A fraud detection system utilizing hysteresis models, decision trees, and transformer networks to analyze transaction data in real-time, integrating immediate transaction details with historical data for pattern recognition and risk assessment, and applying weighted analysis to reach a decision on transaction risk levels.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If existing automated fraud detection systems are used, then fraud risk can be evaluated, but the evaluation cannot be completed within an acceptable time interval between payment method presentation and transaction completion

Engineering Contradiction:

Improvefraud risk evaluation reliabilityVSAvoidtransaction processing time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The fraud detection system is segmented into multiple specialized models (hysteresis model for temporal patterns, decision tree for rule-based evaluation, transformer network for sequence analysis) that process different aspects of transaction data in parallel, enabling comprehensive fraud assessment within acceptable time intervals

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system performs preliminary fraud risk assessment by analyzing historical transaction data and building user profiles before the actual transaction occurs. This pre-computed information is then rapidly applied to the current transaction request, enabling fast real-time decision-making without comprehensive re-analysis

Inventive Principle:

Principle #10Preliminary action

2Measurement precision

If transaction verification is delayed to improve fraud detection accuracy, then fraud risk assessment improves, but customer experience deteriorates due to transaction delays

Engineering Contradiction:

Improvefraud risk assessment accuracyVSAvoidtransaction completion ease

Core Design Contradiction:

Measurement precisionVSEase of operation

Solution Approach 1:

The system applies partial verification by focusing on the most discriminative features and patterns for each transaction type, rather than performing exhaustive analysis of all possible data points. This selective approach maintains high accuracy while reducing processing time to preserve customer experience

Inventive Principle:

Principle #16Partial or excessive action

3Reliability

If comprehensive historical data analysis is performed, then fraud pattern recognition improves, but system complexity increases

Engineering Contradiction:

Improvefraud detection reliabilityVSAvoidsystem architecture complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The complex fraud detection task is segmented into three specialized models: a hysteresis model for temporal pattern recognition, a decision tree for rule-based risk assessment, and a transformer network for sequence analysis. Each model handles specific aspects of data analysis, making the overall system more manageable despite comprehensive analysis

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system transforms historical transaction data into structured representations across multiple dimensions (temporal sequences, feature vectors, pattern matrices) that can be efficiently processed by different model types, enabling comprehensive analysis without linearly increasing computational complexity

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

Data Source

PatentUS20250371547A1Financial transaction fraud prevention system using hysteresis models, decision trees, and transformer networks

Publication Date: 2025.12.04 GIVECORPORATION INC

AI summary

A fraud detection system preemptively identifies and mitigates fraudulent financial transactions in real-time across a diverse range of digital and physical transaction sources. Transaction data may be acquired from various platforms and categorized into rule sets specific to transaction types. Categorized data is aligned with existing risk profiles to construct a dynamic hysteresis model. This model, evaluated by a decision tree algorithm, identifies potential fraud by integrating immediate transaction details with a comprehensive historical data analysis, thus enabling advanced trend analysis and pattern recognition. Key features identified by the decision tree are used to form a heuristics model, which is then analyzed by a transformer network risk model. A feedback loop enhances the system's effectiveness by incorporating decision outcomes back into the model training server, thus refining the training dataset and continuously improving the accuracy of the risk model.