Interpretable Mixture Density Estimation by use of Differentiable Tree-module

TL;DR

This paper introduces an interpretable mixture density estimation method using a differentiable tree structure, enabling reliable uncertainty quantification with fast inference for real-world machine learning applications.

Contribution

It proposes a novel mixture density estimation approach that combines interpretability with high-speed inference through a differentiable tree module.

Findings

Achieves high interpretability in mixture density estimation.

Provides fast inference with a time-invariant information cache.

Demonstrates effectiveness in modeling complex uncertainty distributions.

Abstract

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.