Bayesian Inference
Bayesian inference is a statistical framework for updating beliefs about unknown parameters based on observed data, aiming to quantify uncertainty and make robust predictions. Current research emphasizes developing efficient algorithms, such as those based on neural networks (e.g., simulation-based inference, variational autoencoders), to handle complex models and high-dimensional data, often incorporating techniques like amortized inference and gradient-based methods (e.g., Stein variational gradient descent). These advancements are significantly impacting various scientific fields, enabling more accurate and reliable inference in applications ranging from cosmology and medical diagnostics to robotics and materials science.
Papers
A Survey of Uncertainty Estimation in LLMs: Theory Meets Practice
Hsiu-Yuan Huang, Yutong Yang, Zhaoxi Zhang, Sanwoo Lee, Yunfang Wu
On Cold Posteriors of Probabilistic Neural Networks: Understanding the Cold Posterior Effect and A New Way to Learn Cold Posteriors with Tight Generalization Guarantees
Yijie Zhang