Anatoli Juditsky

This book provides a modern perspective on the connection between convex optimization and high-dimensional statistics, commonly used in machine learning. The authors propose a convex optimization-based approach to address some of the critical challenges in modern data science by focusing on establishing performance limits for statistical procedures under the key assumption of computational tractability. This innovative approach lays the groundwork for a scalable, principled framework to efficiently achieve near-optimal performance in high-dimensional statistics, supplying both optimization-based theoretical tools and an algorithmic toolbox for statistical analysis.