Payment Risk Score Interpretation Through Parameter Perturbation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Black box machine learning models in electronic payment transactions lack transparency, making it difficult for users to understand which aspects of the input data drive the decision, hindering monitoring, understanding, and justifying the model's outcomes.
Innovation Solution
A method and system that perturb transaction parameters to generate perturbed risk scores, allowing for the determination of impact parameters by comparing original and perturbed risk scores, thereby enhancing transparency and enabling interpretable justification of payment transaction decisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If black box machine learning models are used for payment transaction risk assessment, then model accuracy and decision-making capability are improved, but model transparency and interpretability deteriorate
Solution Approach 1:
The patent segments the risk assessment process into multiple components by analyzing individual transaction parameters separately. It divides the black box model's decision-making into discrete parameter analyses, allowing users to understand which specific parameters (e.g., transaction amount, location, time) most influenced the risk score without exposing the complex internal model structure.
Solution Approach 2:
The patent introduces an intermediary explanation layer between the black box model and the user. This intermediary component translates the model's internal decisions into human-understandable reasons by analyzing parameter impacts, serving as a mediator that preserves model accuracy while providing transparency about decision factors.
2Loss of information
If transaction parameters are perturbed to analyze model decisions, then model interpretability is improved, but computational resources and processing time increase
Solution Approach 1:
The patent applies partial action by perturbing only the most relevant transaction parameters rather than all parameters simultaneously. It identifies key parameters to analyze and applies perturbation selectively to those, reducing computational overhead while still providing meaningful interpretability insights into the model's decision-making process.
Solution Approach 2:
The patent performs preliminary analysis to identify which transaction parameters have the most significant impact on risk scores before conducting full perturbation analysis. This preliminary screening step filters out less important parameters, reducing the computational burden of subsequent perturbation operations while maintaining interpretability focus on meaningful factors.
Data Source
AI summary
A computer-implemented method includes: receiving an inquiry request message identifying a first payment transaction having a plurality of transaction parameters and a risk score, where the risk score is generated by a machine-learning model based on the plurality of transaction parameters; for each transaction parameter of the plurality of transaction parameters, perturbing a value of the transaction parameter and re-analyzing the first payment transaction with the machine-learning model to generate a perturbed risk score based on the perturbed transaction parameter; determining at least one impact parameter from the plurality of transaction parameters by comparing the perturbed risk scores generated for each of the plurality of transaction parameters; and generating an inquiry response message based on the at least one impact parameter.