Astronomy Tea Talk
Cahill, Hameetman Auditorium
Multidimensional parameter space, the final frontier
Ciro Donalek,
Caltech,
The growing complexity of multidimensional data sets poses serious challenges for an effective data analysis and visualization. For example, in a typical sky survey, each detected object is represented as a feature vector with several tens to several hundred dimensions (parameters). Given the high number of parameters available for each object, automatic feature selection is quickly becoming a crucial task in analyzing astronomical data sets. In this talk I will illustrate how a machine learning framework can be employed to enhance our understanding of the data. Moreover, visualization is a key component in a discovery process and in an intuitive understanding of the data. I'll present also two immersive, collaborative, visualization tools where multiple users can navigate through the data simultaneously, either with their own, independent vantage points, or with a shared view. Recently we have started adding support for Oculus Rift, a next generation headset for VR environments, and Leap Motion; a demo will be available.
For more information, please contact Luca Ricci and Dan Perley by phone at 626-395-2460 and 626-395-3734 or by email at lricci@astro.caltech.edu and dperley@astro.caltech.edu.
Event Series
Astronomy Tea Talks