Robust and Sample-Efficient Simulation-Based Inference

Explore the challenges and advancements in simulation-based inference (SBI) through this 36-minute talk by Ayush Bharti, a postdoctoral researcher at Aalto University. Delve into the complexities of statistical inference for simulator-based models used across various scientific and engineering domains. Understand the limitations of traditional methods when faced with unavailable likelihood functions and learn how SBI leverages large-scale simulations to overcome these obstacles. Discover recent developments in robust and sample-efficient SBI techniques designed to address model misspecification and computational constraints. Gain insights into Bharti's research on improving SBI performance in real-world scenarios where models may not perfectly capture the phenomenon under study or when simulation resources are limited. Explore the intersection of probabilistic machine learning, Bayesian optimization, and experimental design as applied to enhancing simulation-based inference methodologies.

Ayush Bharti: Robust and sample-efficient simulation-based inference