Using Visual Analytics to Interpret Predictive Machine Learning Models
This addresses the challenge of model interpretability for users who need to understand AI decisions while maintaining performance, though it appears incremental as it builds on existing visual analytics approaches.
The paper tackles the problem of interpreting predictive machine learning models without sacrificing predictive power by using visual analytics to inspect input-output relationships as black-boxes, and it demonstrates successful practical applications with two examples.
It is commonly believed that increasing the interpretability of a machine learning model may decrease its predictive power. However, inspecting input-output relationships of those models using visual analytics, while treating them as black-box, can help to understand the reasoning behind outcomes without sacrificing predictive quality. We identify a space of possible solutions and provide two examples of where such techniques have been successfully used in practice.