Translation Equivariant Transformer Neural Processes

TL;DR

This paper introduces translation equivariant Transformer Neural Processes (TE-TNPs), enhancing neural process models by incorporating translation symmetry, leading to improved performance on spatio-temporal data.

Contribution

The paper proposes a novel family of translation equivariant TNPs that explicitly incorporate translation symmetry into the model architecture.

Findings

TE-TNPs outperform non-equivariant models on synthetic data

TE-TNPs show improved accuracy on real-world spatio-temporal datasets

Translation equivariance enhances model robustness and generalization

Abstract

The effectiveness of neural processes (NPs) in modelling posterior prediction maps -- the mapping from data to posterior predictive distributions -- has significantly improved since their inception. This improvement can be attributed to two principal factors: (1) advancements in the architecture of permutation invariant set functions, which are intrinsic to all NPs; and (2) leveraging symmetries present in the true posterior predictive map, which are problem dependent. Transformers are a notable development in permutation invariant set functions, and their utility within NPs has been demonstrated through the family of models we refer to as TNPs. Despite significant interest in TNPs, little attention has been given to incorporating symmetries. Notably, the posterior prediction maps for data that are stationary -- a common assumption in spatio-temporal modelling -- exhibit translation equivariance. In this paper, we introduce of a new family of translation equivariant TNPs that incorporate translation equivariance. Through an extensive range of experiments on synthetic and real-world spatio-temporal data, we demonstrate the effectiveness of TE-TNPs relative to their non-translation-equivariant counterparts and other NP baselines.