The wide variety of new types of applications and the ever present explosion of data have forced machine learning to rapidly evolve and significantly expand its basic tools and ap-proaches. In addition to classic computational and statistical complexity questions, many new fundamental questions arise in this context: how to best interact with the expert and most ef-fectively utilize the available data in interactive learning settings, how to design communica-tion efficient learning protocols in distributed settings, and how to preserve privacy of highly sensitive data? In this talk I will discuss recent advances in these directions.
I will also discuss how learning theory can bring novel insights into core problems of other fields. As one example, I will focus on submodular functions, commonly used to model laws of diminishing returns in many contexts ranging from Economics to Social Networks. By ana-lyzing them through a learning theoretic lens, we uncover several novel structural results re-vealing ways in which submodular functions can be both surprisingly structured and surprisingly unstructured, with numerous implications in other fields.
Maria Florina Balcan is an assistant professor in the School of Computer Science at Georgia Institute of Technology. Her main research interests are machine learning, computational aspects in economics and game theory, and algorithms. Her honors include the CMU SCS Distinguished Dissertation Award, an NSF CAREER Award, a Microsoft Faculty Research Fellowship, a Sloan Research Fellowship, and several paper awards at COLT. She is currently a board member of the International Machine Learning Society and Program Committee chair for COLT 2014.
sharonw [atsymbol] cs.cmu.edu