Extreme scale computing promises transformative knowledge discovery (e.g., predictive models, change detection, sensitivity analysis) via faster and more realistic simulations. Such scientific progress, however, is heavily dependent on the availability of exploratory data analysis (EDA) techniques that support the formulation and investigation of new hypotheses. Presently, extreme scale EDA tools are missing, which forces data reductions and limits analysis to the hypotheses that prompted collection. This scenario inhibits surprise discoveries, which are often the tipping points for scientific breakthroughs. To realize extreme scale EDA, I believe a visual analytics approach that effectively blends the strengths of humans with those of machines is the most viable strategy.

In this presentation, I will discuss the Exploratory Data analysis ENvironment (EDEN), which corroborates the promise of extreme scale EDA. EDEN is a visual analytics system that harnesses interactive information visualization techniques to allow human- centered analysis of extreme scale data. After a brief introduction, I will describe EDEN’s linked visual representations, interaction methods, and utilization of “information scent�. I will also present practical case studies that illustrate the effectiveness of EDEN in large- scale earth system simulation analysis. Along the way, I will highlight how I have benefited from multi-disciplinary partnerships with domain experts. I will conclude the presentation with new directions for this research, including in situ visual analytics for climate simulations and extensions to support the analysis of network flow information.

Dr. Chad A. Steed is a visual analytics researcher in the Computational Data Analytics Group at Oak Ridge National Laboratory (ORNL). He also holds a Joint Faculty Appointment with the University of Tennessee’s Computer Science Department. Before joining ORNL, he worked as a computer scientist at the Naval Research Laboratory for 9 years. He has a Ph.D. (2008) degree in computer science from Mississippi State University, where he studied visualization and computer graphics. Dr. Steed’s research spans a range of topics including visual analytics, human-computer interaction, scientific and information visualization, data mining, social networks, databases, and web development. His current focus is on the formulation and practical application of extreme scale visual analytics techniques that combine automated analytics with interactive visualizations to enhance cognition and decision support for multi-disciplinary challenges. For more information about Dr. Steed and his research, visit http://csteed.github.io/.