Interpretable Mixture Density Estimation by use of Differentiable Tree-module
This work addresses the need for interpretable uncertainty estimation in real-world ML services, though it appears incremental as it builds on existing mixture density estimation methods.
The paper tackles the problem of interpreting complex mixture density estimates for reliable machine learning services by proposing a method that uses an interpretable tree structure, achieving both high speed and interpretability through a fast inference procedure.
In order to develop reliable services using machine learning, it is important to understand the uncertainty of the model outputs. Often the probability distribution that the prediction target follows has a complex shape, and a mixture distribution is assumed as a distribution that uncertainty follows. Since the output of mixture density estimation is complicated, its interpretability becomes important when considering its use in real services. In this paper, we propose a method for mixture density estimation that utilizes an interpretable tree structure. Further, a fast inference procedure based on time-invariant information cache achieves both high speed and interpretability.