Machine Learning Decisioning Model Explainability for Bias Mitigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine learning-based decisioning models are often opaque, leading to a lack of transparency and fairness in decision-making processes, undermining user trust and making it difficult to discern how specific inputs yield particular outputs.
Innovation Solution
A system and method that uses a machine learning algorithm to detect favorable decisioning records closest to unfavorable records, compute bias intensity metrics, generate an explainability artifact, and display it in a user interface to explain and mitigate bias in decisioning models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If machine learning-based decisioning models are used to improve decision-making efficiency, then productivity is improved, but the black box nature of these models causes loss of information regarding transparency and fairness
Solution Approach 1:
The patent introduces an explainability artifact as an intermediary between the machine learning decisioning model and users. This artifact includes bias intensity metrics and factor explanations that mediate the information flow, making the black box model's decision-making process transparent without affecting the model's operational efficiency. The artifact serves as a bridge that preserves both productivity and transparency information.
2Manufacturing precision
If complex machine learning models are deployed to improve decision accuracy, then manufacturing precision is improved, but device complexity increases making the system harder to understand and trust
Solution Approach 1:
The patent segments the complex machine learning model into understandable components by generating explainability artifacts that break down decision-making into discrete factors. Each artifact contains specific bias intensity metrics and factor explanations that divide the complex model's operation into manageable, interpretable elements. This segmentation maintains decision accuracy while reducing the perceived complexity for users.
3Reliability
If bias mitigation measures are implemented to improve fairness, then reliability is improved, but the computational requirements and processing time increase
Solution Approach 1:
The patent performs bias analysis and generates explainability artifacts in advance of actual decision-making operations. By pre-computing bias intensity metrics and identifying biased factors beforehand, the system establishes fairness criteria that can be quickly applied during runtime decisions. This preliminary action ensures reliability and fairness while minimizing the time penalty during actual decision-making processes.
Data Source
AI summary
A system, method, and computer-program product includes obtaining a decisioning dataset comprising a plurality of favorable decisioning records and at least one unfavorable decisioning record; detecting, via a machine learning algorithm, a favorable decisioning record of the plurality of favorable decisioning records that has a vector value closest to a vector value of the unfavorable decisioning record; executing a counterfactual assessment between the favorable decisioning record and the unfavorable decisioning record; generating an explainability artifact based on one or more bias intensity metrics to explain a bias in a machine learning-based decisioning model; and in response to generating the explainability artifact, displaying the explainability artifact in a user interface.