Enabling Mixed Effects Neural Networks for Diverse, Clustered Data Using Monte Carlo Methods

TL;DR

This paper introduces MC-GMENN, a Monte Carlo-based method for training mixed effects neural networks that better model clustered data, improving generalization, interpretability, and applicability to complex datasets.

Contribution

The paper proposes MC-GMENN, a novel Monte Carlo approach enabling generalized mixed effects neural networks to handle diverse, multi-clustered data with improved performance and interpretability.

Findings

MC-GMENN outperforms existing models in generalization and variance quantification.

It is effective for multi-class classification with high-cardinality categorical features.

The method offers a principled way to interpret clustering effects.

Abstract

Neural networks often assume independence among input data samples, disregarding correlations arising from inherent clustering patterns in real-world datasets (e.g., due to different sites or repeated measurements). Recently, mixed effects neural networks (MENNs) which separate cluster-specific 'random effects' from cluster-invariant 'fixed effects' have been proposed to improve generalization and interpretability for clustered data. However, existing methods only allow for approximate quantification of cluster effects and are limited to regression and binary targets with only one clustering feature. We present MC-GMENN, a novel approach employing Monte Carlo methods to train Generalized Mixed Effects Neural Networks. We empirically demonstrate that MC-GMENN outperforms existing mixed effects deep learning models in terms of generalization performance, time complexity, and quantification of inter-cluster variance. Additionally, MC-GMENN is applicable to a wide range of datasets, including multi-class classification tasks with multiple high-cardinality categorical features. For these datasets, we show that MC-GMENN outperforms conventional encoding and embedding methods, simultaneously offering a principled methodology for interpreting the effects of clustering patterns.