Paper accepted at AStA

Our paper Mixture of Experts Distributional Regression: Implementation Using Robust Estimation with Adaptive First-order Methods was accepted for publication at AStA Advances in Statistical Analysis in the special issue Bridging the gap between AI and Statistics.

Abstract

In this work, we propose an efficient implementation of mixtures of experts distributional regression models which exploits robust estimation by using stochastic first-order optimization techniques with adaptive learning rate schedulers. We take advantage of the flexibility and scalability of neural network software and implement the proposed framework in mixdistreg, an R software package that allows for the definition of mixtures of many different families, estimation in high-dimensional and large sample size settings and robust optimization based on TensorFlow. Numerical experiments with simulated and real-world data applications show that optimization is as reliable as estimation via classical approaches in many different settings and that results may be obtained for complicated scenarios where classical approaches consistently fail.

For more details, see our ArXiv preprint!