The Gaussian Neural Process

TL;DR

The paper introduces the Gaussian Neural Process, a new model that enhances Neural Processes by modeling correlations, ensuring translation equivariance, and providing universal approximation guarantees, with promising empirical results.

Contribution

It proposes the Gaussian Neural Process, a novel model that improves Neural Processes with correlation modeling, translation equivariance, and theoretical guarantees.

Findings

Models predictive correlations effectively.

Ensures translation equivariance.

Shows encouraging empirical performance.

Abstract

Neural Processes (NPs; Garnelo et al., 2018a,b) are a rich class of models for meta-learning that map data sets directly to predictive stochastic processes. We provide a rigorous analysis of the standard maximum-likelihood objective used to train conditional NPs. Moreover, we propose a new member to the Neural Process family called the Gaussian Neural Process (GNP), which models predictive correlations, incorporates translation equivariance, provides universal approximation guarantees, and demonstrates encouraging performance.