midr: Learning from Black-Box Models by Maximum Interpretation Decomposition
This provides a method for improving explainability in fields requiring model transparency, but it appears incremental as it builds on existing functional decomposition approaches.
The authors tackled the problem of interpreting black-box predictive models by introducing the R package midr, which implements Maximum Interpretation Decomposition (MID) to derive a low-order additive representation, resulting in a tool for constructing global surrogate models with advanced analytical capabilities.
The use of appropriate methods of Interpretable Machine Learning (IML) and eXplainable Artificial Intelligence (XAI) is essential for adopting black-box predictive models in fields where model and prediction explainability is required. As a novel tool for interpreting black-box models, we introduce the R package midr, which implements Maximum Interpretation Decomposition (MID). MID is a functional decomposition approach that derives a low-order additive representation of a black-box model by minimizing the squared error between the model's prediction function and this additive representation. midr enables learning from black-box models by constructing a global surrogate model with advanced analytical capabilities. After reviewing related work and the theoretical foundation of MID, we demonstrate the package's usage and discuss some of its key features.