he development and analysis of randomized iterative methods are pivotal for addressing challenges in mathematical data science. Data manifests in diverse forms in large-scale data settings, often accompanied by noise, corruption, or missing values. In addition to these challenges, data is often difficult to work with due to its scale, representation, and storage. Randomized algorithms can help tackle these challenges, allowing practitioners to solve large-scale problems with low memory cost and avoid or otherwise handle corrupted or missing entries. This presentation introduces simple yet powerful iterative methods for solving large-scale linear systems from the linear algebraic and optimization perspectives and connections between such methods and their applications.
Copyright. All Rights Reserved.