Sequential Estimation

Sequential estimation focuses on dynamically updating estimates of parameters or states from streaming data, aiming for accurate and timely inferences. Current research emphasizes developing robust algorithms, such as particle filters (often enhanced with neural networks) and confidence sequences, to handle complex, high-dimensional data and non-linear systems, including those with dynamic interactions between estimation and future data generation. These advancements are improving inference in diverse applications like robot localization, change detection, and Bayesian optimization, offering more efficient and accurate solutions for real-world problems involving sequential data.