Department of Information and Computing Sciences, Utrecht University.
Find out more about Alexandru Telea.
Machine learning (ML) has witnessed tremendous successes in the last decade in classification, regression, and prediction tasks. However, many ML models are used, and sometimes even designed, as black boxes. When such models do not operate properly, their creators do not often know what is the best way to improve them. Moreover, even when operating successfully, users often require to understand how and why they take certain decisions to gain trust therein. We present how information visualization and visual analytics help towards explaining (and improving) ML models. These cover tasks such as understanding high-dimensional datasets; understanding unit specialization during the training of deep learning models; exploring how training samples determine the shape of classification decision boundaries; and helping users annotating samples in semi-supervised active learning scenarios.
Alexandru Telea is a Professor of Visual Data Analytics at the Department of Information and Computing Sciences, Utrecht University. He holds a PhD from Eindhoven University and has been active in the visualization field for over 22 years. He has been the program co-chair, general chair, or steering committee member of several conferences and workshops in visualization, including EuroVis, VISSOFT, SoftVis, and EGPGV. His main research interests cover unifying information visualization and scientific visualization, high-dimensional visualization, and visual analytics for machine learning. He is the author of the textbook "Data Visualization: Principles and Practice" (CRC Press, 2014).
Department of Computer Science, University of British Columbia.
Find out more about Tamara Munzner.
Design studies are a popular approach to problem-driven research in the field of visualization. I define a design study as a situation where visualization researchers analyze a specific real-world problem faced by domain experts, design a visualization system that supports solving this problem, validate the design, and reflect about lessons learned in order to refine visualization design guidelines. I will discuss the methodology of conducting these studies and processes to help researchers and practitioners avoid potential pitfalls. For example, one consideration for success in such projects is ensuring that the collaboration incentives are aligned for all parties. I will illustrate the potential and challenges of design studies through case studies in three application domains: conducting facilities management and planning with building occupancy data, analyzing consumer behaviour with e-commerce clickstream data, and investigating biological hypotheses through the relationships between many evolutionary trees for both species and genes.
Tamara Munzner is a Professor of Computer Science at the University of British Columbia. She holds a PhD from Stanford and has been active in the visualization field for over 30 years. Her longstanding engagement with the IEEE VGTC community includes service as InfoVis and EuroVis Papers Co-Chair, and chair of the VIS Restructuring Committee, the VIS Executive Committees, and the InfoVis Steering Committee. She published the book "Visualization Analysis and Design" in 2014 as the first in the AK Peters Visualization Series (CRC Press), and continues as series editor. She received the IEEE VGTC Visualization Technical Achievement Award in 2015.