Title: Multiscale Methods in Information Visualization and Shape Analysis

Speaker:  Alexandru C. Telea

Abstract:

In the last years, the size (number of data points) and dimensionality of scientific datasets has enormously increased, leading to so-called 'big data' challenges. This poses problems also to visualization methods.
A well-known solution to the data size problem is to represent the data on a fine-to-coarse multiscale (MS). For the dimensionality problem, so-called dimensionality-reduction (DR) methods offer complementary solutions. However well known and studied in scientific visualization and imaging, MS solutions are far less well-established in information visualization. Separately, however well known in information visualization, DR has many interpretation challenges.
In this talk, we present new developments in both MS and DR methods for the visualization of large and high-dimensional infovis and scivis datasets. Key to these methods is an image-based approach, which casts data simplification and presentation problems in the well-known framework of image processing. We illustrate our methods with examples from machine learning, visual analytics, graph exploration, 3D shape analysis, and 3D shape processing.

Biography:
Prof. dr. Alexandru C. Telea obtained his PhD in Computer Science from the Eindhoven University of Technology in 2000. Until 2007, he worked there as assistant professor in visualization. Since 2007, he is full professor of multiscale visual analytics at the Institute Johann Bernoulli, University of Groningen, the Netherlands. His research interests include multiscale visualization methods, 3D shape processing with medial axes, and software visualization, which are covered by over 200 of his publications. He is best known for his book "Data Visualization - Principles and Practice" (2nd edition, 2014), which has become one of the standard (under)graduate textbooks in visualization teaching.