Recent work on kernel methods has produced some exciting and powerful new machine learning techniques (e.g. Support Vector Machines, kernel Fisher discriminants, kernel PCA), along with some promising initial application successes. In this talk we will explore the key principles behind these methods, discuss important recent advances, and summarize applications (including some recent NASA applications) which both demonstrate current successes as well as motivate specific future research directions.
Dennis DeCoste is Technical Group Leader of the Machine Learning Systems Group at the Jet Propulsion Laboratory. He received his Ph.D. in Computer Science / Artificial Intelligence from the University of Illinois at Urbana-Champaign in 1994.
At JPL he leads several research projects on data mining of large time-series and image data sets, for both engineering and science analysis, over a variety of NASA domains, including Terra/MISR, SIM, Space Shuttle, Mars Rovers, and bioinformatics. His adaptive novelty detection software flew onboard the recent successful Deep Space 1 mission. His current research focuses on scaling support vector machine and other kernel methods to handle massive data sets.
Back to Mathematical Challenges in Scientific Data Mining