Smarajit Bose

A Heterogeneous Mixture of Experts Framework for Interpretable Machine Learning

The proposed approach combines interpretability, adaptive inductive bias selection, and probabilistic coherence within a unified mixture-of-experts framework to achieve interpretable expert assignments while achieving predictive performance competitive with homogeneous MoDT and Random Forests.