Tools for BigData Approaches: Numerical Moment Matching and Multi-Layered Parallel Grouping Techniques
Date/Time: November 5th, 3:40 pm
Location: B29 Atanasoff Hall
With the rise of advanced computational power and large-scale database in a broad spectrum of scientific and engineering fields, strong demand emerges especially for novel computational and numerical tools that can facilitate big data-oriented researches. The topic of this talk will be twofold. First, I will present a numerical moment matching technique that is believed to be powerful for representing large-scale irregular population and for incorporating uncertainty behind the real-world big population. This technique will help researchers to foresee how BigData will behave (structurally, economically, biologically, etc.) in conjunction with advanced simulation/prediction models. Next, I will expand upon the multi-layered parallel grouping technique and its successful application to large multi-scale analysis in engineering fields. Notably, a seemingly “super-linear” speedup has been revealed and investigated. This technique is believed to be able to offer unprecedented access to ultimate capacity of parallel computer and to pave a new way to the optimal design of parallel calculations with BigData.
- Seoul National University, S. Korea. B.S. and M.S. in Civil, Urban and Geotechnical Engineering.
- California Institute of Technology, Civil Eng., USA. Ph.D., 2012.
- California Institute of Technology, Computational Sci. & Eng., Ph.D. minor, 2012.
- Postdoctoral Researcher and Lecturer, Department of Civil, Environmental, and Architectural Engineering, University of Colorado at Boulder, 2014.