Learning for Decision and Control in Stochastic Networks E-Book

Longbo Huang

• Herausgeber: Springer International Publishing
• Kategorie: Wissenschaft und neue Technologien
• Serie: Synthesis Lectures on Learning, Networks, and Algorithms
• Sprache: Englisch

Veröffentlichungsjahr: 2023

Beschreibung

This book introduces the Learning-Augmented Network Optimization (LANO) paradigm, which interconnects network optimization with the emerging AI theory and algorithms and has been receiving a growing attention in network research. The authors present the topic based on a general stochastic network optimization model, and review several important theoretical tools that are widely adopted in network research, including convex optimization, the drift method, and mean-field analysis. The book then covers several popular learning-based methods, i.e., learning-augmented drift, multi-armed bandit and reinforcement learning, along with applications in networks where the techniques have been successfully applied. The authors also provide a discussion on potential future directions and challenges.