“Geometric Algorithms for Transformation of Data to Knowledge”
Karthik Ramani, Purdue University
Design and manufacturing (D&M) together serve the overarching goal of achieving superior function via form and physical behavior, while satisfying customer needs. Increasingly this process has become heavily driven by information sciences and technologies (IS&T). Key trends in globalization are also moving us toward a knowledge-based economy in which innovation at the interfaces of IS&T, D&M and other fields has come to play a key role in adding value to the entire life cycle of products. The explosion of data in many other fields is also posing serious problems where computational capabilities are lagging behind due to lack of appropriate algorithms for transforming the information to knowledge. These problems are in fact very general and transcend many disciplines. This talk will provide such examples in geometric design, structural biology and analysis.
The development of 3D computer graphics and related modeling and digitizing tools has led to the creation of a large number of 3D models in many fields. In engineering, reusing and sharing the knowledge embedded in 3D shapes is becoming an important way to accelerate the design process, improve product quality and reduce costs. This process requires effective methods for indexing, clustering and retrieving the 3D/2D shapes in large digital repositories. Traditional approaches such as keywords and annotating are insufficient to describe or search for 3D shapes intuitively since the important ideas are usually implied in the 3D content. Therefore it is necessary to provide designers a means to search and retrieve 3D models based on their content. We have proposed new representations where the similarity between 3D shapes is transformed to compute the similarities between the shape descriptors.
These insights led to the discovery of new methods to improve the efficiency and naturalness of computer-aided design and analysis through, for example, sketch-based parameterization and finite-element analysis. A new 3D inner distance shape descriptor for deformation invariant search of flexible proteins is introduced. Finally, the talk summarizes new research that is applicable across domains and types of data for embedding high dimensional data into diffusion spaces for multi-scale analysis.
Karthik Ramani is a professor of mechanical engineering at Purdue. He earned his B.Tech from the Indian Institute of Technology, Madras, in 1985, an MS from The Ohio State University in 1987 and a PhD from Stanford University in 1991, all in mechanical engineering. His interests are in digital and computational geometry, high-dimensional mathematics, shape search and computer support for early design. He receives support from NSF-CISE, the National Institutes of Health, General Electric and Siemens/Parametric Technology Corp./Boeing, and he previously served as chief scientist at Imaginestics, a knowledge-based software company that has launched the world’s first online search engine for the global supply chain. He has also worked as a summer intern in Delco Products, Advanced Composites, and as a summer faculty intern in Dow Plastics, Advanced Materials.
He has received many awards, including the Dupont Young Faculty Award, an NSF Research Initiation Award, an NSF CAREER Award, the Ralph Teetor Educational Award from the Society of Automotive Engineers, the Outstanding Young Manufacturing Engineer Award from the Society of Manufacturing Engineers, and the Ruth and Joel Spira Award for Outstanding contributions to the Mechanical Engineering Curriculum. In 2002 he was recognized by Purdue through a University Faculty Scholars Award and won the NSF partnership for innovation award. In 2005 he won the Discovery in Mechanical Engineering Award for his work in shape search. In 2006 he was a finalist for Indiana’s innovation of the year award. In 2006 and 2007 he won the most cited journal paper award from Computer-Aided Design, the research excellence award throughout the college of engineering at Purdue and the Thomas French Award for outstanding educator from The Ohio State University. In 2009 he won the outstanding commercialization award from Purdue University and the ASME Best Paper Award at the International Design Engineering Technical Conference.
He serves on the editorial board of the Elsevier Journal of Computer-Aided Design and the ASME Journal of Mechanical Design. He also serves on the NSF Advisory Committee for Industrial Innovation and Partnerships.