Causal Foundation Models: Disentangling Physics from Instrument Properties

Jeroen Audenaert, Daniel Muthukrishna, Paul F. X. Gregory, David W Hogg, V Ashley Villar

Keywords: foundation models, time series, contrastive learning, causal representation learning, disentanglement, few-shot learning, astronomy

TL;DR: We propose a causal foundation model for time series data that disentangles true physical variations from instrumental effects, yielding improved downstream performance on simulated astronomical data compared to conventional foundation models.

Abstract: Foundation models for structured time series data must contend with a fundamental challenge: observations often conflate the true underlying physical phenomena with systematic distortions introduced by measurement instruments. This entanglement limits model generalization, especially in heterogeneous or multi-instrument settings. We present a causally-motivated foundation model that explicitly disentangles physical and instrumental factors using a dual-encoder architecture trained with structured contrastive learning. Leveraging naturally occurring observational triplets (i.e., where the same target is measured under varying conditions, and distinct targets are measured under shared conditions) our model learns separate latent representations for the underlying physical signal and instrument effects. Evaluated on simulated astronomical time series designed to resemble the complexity of variable stars observed by missions like NASA's Transiting Exoplanet Survey Satellite (TESS), our method significantly outperforms traditional single-latent space foundation models on downstream prediction tasks, particularly in low-data regimes. These results demonstrate that our model supports key capabilities of foundation models, including few-shot generalization and efficient adaptation, and highlight the importance of encoding causal structure into representation learning for structured data.

Submission Number: 94