Latent Variable Multi-output Gaussian Processes for Hierarchical Datasets

TL;DR

This paper introduces a hierarchical extension of multi-output Gaussian processes that models complex relationships in structured datasets, improving scalability and capturing dependencies more effectively, with applications demonstrated in genomics and motion capture.

Contribution

It proposes a novel hierarchical MOGP model with tailored kernels and latent variables to better represent complex output relationships in structured data.

Findings

Improved modeling of hierarchical data structures.

Enhanced scalability with increasing tasks.

Effective application to genomics and motion capture datasets.

Abstract

Multi-output Gaussian processes (MOGPs) have been introduced to deal with multiple tasks by exploiting the correlations between different outputs. Generally, MOGPs models assume a flat correlation structure between the outputs. However, such a formulation does not account for more elaborate relationships, for instance, if several replicates were observed for each output (which is a typical setting in biological experiments). This paper proposes an extension of MOGPs for hierarchical datasets (i.e. datasets for which the relationships between observations can be represented within a tree structure). Our model defines a tailored kernel function accounting for hierarchical structures in the data to capture different levels of correlations while leveraging the introduction of latent variables to express the underlying dependencies between outputs through a dedicated kernel. This latter feature is expected to significantly improve scalability as the number of tasks increases. An extensive experimental study involving both synthetic and real-world data from genomics and motion capture is proposed to support our claims.