Alexandru Telea   &   Michael Behrisch

Date: 16:00 – 17:00, Friday, 22.11.2019
Location: RUPPERT-D

Title: Visualizing the Invisible: Understanding Abstract Big Data and Explaining Machine Learning
Abstract: Big data, pervasive today in science and technology, poses many challenges to scientists and practitioners. Information visualization is the technique of choice of exploring such large, high-dimensional, abstract, and discrete data collections to help forming, validating, and refining hypotheses about the underlying phenomena and ultimately solve practical problems. In this context, we are developing a particular class of techniques, called image-based information visualization, that leverage the human’s powerful cognition and perception system and (semi-)automatically support the cognitively demanding task of finding interpretable visualizations. These techniques inherit desirable continuous properties from classical imaging and scientific visualization, thereby being able to handle many of the challenges of big data in a natural way. We present several such recent techniques with applications in software understanding, exploration of large graphs and trail-sets, understanding multidimensional data, and opening the ‘black box’ of machine learning and deep learning.