Exploring Star Formation and ISM through Artificial Intelligence

Duo Xu (University of Toronto, CITA) — September 8, 2025

Abstract

Astronomy is entering a new era driven by artificial intelligence, enabling efficient analysis of large and complex datasets beyond traditional methods. In studies of star formation and the interstellar medium, AI provides tools to extract physical insights and link simulations with observations. First, I will show how discriminative AI can identify and quantify stellar feedback and protostellar outflows, using 3D models trained on synthetic data to recover both structural and statistical properties. Second, I will discuss how generative AI can infer the intrinsic physical conditions of molecular clouds, such as gas density, magnetic fields, and radiation fields, from sparse observational inputs. In particular, I will highlight diffusion-based methods, including Denoising Diffusion Probabilistic Models (DDPM) and the Diffusion Schrödinger Bridge (DSB), which provide physics-aware and interpretable inference while remaining robust to out-of-distribution data. Together, these studies demonstrate the promise of integrating discriminative and generative AI approaches to advance our understanding of star formation and ISM dynamics.