Explore a seminar on the applications of submodular theory in machine learning and natural language processing. Delve into how submodularity, a concept from economics and discrete mathematics, is becoming increasingly relevant in AI. Discover diverse real-world applications, including dynamic graphical models, clustering, summarization, computer vision, and parallel computing. Learn how submodular frameworks lead to efficient, scalable algorithms with high-quality solutions. Understand the importance of developing simple, mathematically rich machine learning constructs suited to real-world challenges. Gain insights into how practical applications have advanced the mathematical study of submodularity. This 57-minute talk, presented by J. Bilmes at the University of Washington's Computer Science and Engineering AI Seminar in February 2016, offers a comprehensive overview of this emerging field in AI research.