Please see below for the next talk in the fall seminar series organized by the Institute for AI & Computational Research on AI/ML techniques and applications across various scientific domains. You can find a table of upcoming talks here: https://web.uri.edu/iacr/seminars/.Speaker: Michael Puerrer (URI)Date/Time/Location: Oct 2, 3pm, Memorial Union room 308.Title: Simulation-Based Inference: Enabling Scientific Discoveries with Machine LearningAbstract: Modern science often relies on computer simulations to model complex systems — from the evolution of ice sheets and the spread of diseases to the merger of compact binaries. A central challenge is inference: learning about the hidden parameters of these systems from limited and noisy observations. Classical statistical methods rely on evaluating the likelihood function, but for realistic simulations the likelihood is often intractable or unavailable.Simulation-Based Inference (SBI) provides a powerful alternative. By leveraging simulations directly, and combining them with machine learning and Bayesian reasoning, SBI makes it possible to approximate posterior distributions without requiring a closed-form likelihood. I will introduce the core idea of SBI, outline the main methods — from approximate Bayesian computation to neural density estimation and deep generative models — and highlight how neural networks serve as surrogate models for inference.I will then survey applications across scientific domains, showing how SBI is enabling new discoveries in climate science, population genetics, and epidemiology, before turning to a detailed example in gravitational-wave astronomy, where SBI methods provide rapid, accurate parameter estimation for merging black holes and neutron stars.