Neural Network Payment Authorization With SHAP Feature Ranking

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing payment transaction systems face challenges in achieving high approval rates while minimizing fraudulent transactions, leading to significant financial losses and eroding customer trust due to frequent declines.

Innovation Solution

A neural network-based approach using Shapley Additive Explanations (SHAP) values to identify significant transaction features and authorizing components, enabling the creation of simulated models to recommend features and components that enhance transaction approval rates and reduce fraud.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional fraud scoring models and authentication models are used to detect fraudulent transactions, then fraudulent transactions can be identified, but legitimate payment transactions are wrongfully declined

Engineering Contradiction:

Improvefraud detection accuracyVSAvoidtransaction approval rate

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The patent segments the feature selection process into two distinct ranking systems: one based on fraud model SHAP values and another based on approval model SHAP values. By comparing the rank differences between these two segmented rankings, the system identifies features that are important for fraud detection but currently underutilized in approval decisions, thereby resolving the contradiction between fraud detection accuracy and transaction approval rate

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Instead of directly optimizing for approval rate, the patent inverts the approach by analyzing what features are being overlooked. It ranks features by their importance to fraud detection, then identifies the gap between this ranking and the current feature utilization in approval models. This inverted analysis reveals underutilized features that can improve approval rates without compromising fraud detection

Inventive Principle:

Principle #13The other way round (Inversion)

2Productivity

If payment transactions are strictly approved to increase approval rates, then customer satisfaction improves, but fraudulent transactions may be approved

Engineering Contradiction:

Improvetransaction approval rateVSAvoidfraudulent transaction approval

Core Design Contradiction:

ProductivityVSObject-affected harmful factors

Solution Approach 1:

The patent implements a feedback mechanism by using SHAP values from both fraud and approval models to continuously evaluate feature importance. The system provides feedback on which features are currently underutilized and should be weighted more heavily, allowing the authorization model to dynamically adjust its decision-making process to maintain both high approval rates and strong fraud detection capabilities

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent changes the parameters of the authorization decision by introducing a new feature ranking methodology based on rank difference analysis. This parameter change shifts the focus from static feature weighting to dynamic feature prioritization, where features are selected based on their relative importance gaps between fraud detection and approval models, enabling flexible adjustment of approval thresholds

Inventive Principle:

Principle #35Parameter changes

3Productivity

If multiple authentication models are deployed to reduce false declines, then transaction approval rates improve, but system complexity increases

Engineering Contradiction:

Improvetransaction approval rateVSAvoidauthorization system complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent performs preliminary action by pre-ranking all transaction features based on their SHAP values from both fraud and approval models before actual authorization decisions are made. This pre-computed feature ranking and the identification of underutilized features creates a streamlined selection process that simplifies the authorization system architecture while maintaining high approval rates, avoiding the need for multiple complex authentication models

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS12423702B2Neural network based methods and systems for increasing approval rates of payment transactions

Publication Date: 2025.09.23 MASTERCARD INT INC

AI summary

Embodiments of present disclosure provide methods and systems for increasing transaction approval rate. Method performed includes accessing transaction features and determining via fraud model and approval model, first and second set of rank-ordered transaction features. Method includes computing difference in ranks of transaction features and determining set of utilized and unutilized transaction features and generating simulated authorizing model and computing simulated transaction approval rate and simulated fraud transaction rate for simulated authorizing model. Method includes generating plurality of proxy authorization models. Method includes computing transaction approval rates and fraud transaction rates for each of plurality of proxy authorization models and computing an increase in transaction approval rate and change in fraud transaction rate for each of plurality of proxy transaction approval models. Method includes determining one or more recommended transaction features from set of unutilized transaction features and transmitting one or more recommended transaction features to authorizing entity.