Justine Zeghal – How Can AI Help Us Understand the Universe?

Pedagogical Talk | Cosmology from Home 2026 https://www.cosmologyfromhome.com/ Abstract: Constraints on cosmological parameters are traditionally obtained by compressing high-dimensional data into summary statistics such as the power spectrum, typically assuming these statistics to be Gaussian distributed. While this approach provides a tractable likelihood for Bayesian inference and reduces the dimensionality of the problem, it inevitably discards a significant fraction of the cosmological information encoded in the data. Recently, there has been a promising shift from analytical likelihood-based approaches to simulations-based inference. This new paradigm enables full-field inference, where all cosmological information embedded in the data is extracted. In my talk I’ll describe how machine learning enables such kind of inference.