My research aims to make visualization an enabling tool for Big Data analytics and discoveries. As almost all areas of business, medicine and healthcare, engineering designs, and scientific studies become increasingly data-driven, visualization is clearly the best interface for human users to access, analyze and interpret Big Data.
My group creates exploratory and explanatory visualization solutions that are built on our research innovations. Our innovations are not restricted to visualization design, and often extend to computer graphics and human-computer interaction. We also incorporate machine learning and high-performance computing techniques into our work. My research has been largely driven by real-world applications. My students and I have developed visualization solutions to assist decision-making in business based on customer data, fighting crime and fraud, detecting and tracking intrusions for network security, understanding large-scale scientific research data, analysis of biomedical image data for better medical diagnosis and surgical planning, supercomputing resource management and performance optimization, and many others. We have also contributed to a collaborative effort to convert scientific research data into interactive visualization exhibits. These exhibits are presently on display at a science museum and help achieve a broader impact of scientific research.
To support scientific studies with supercomputing at extreme scale, my group has introduced novel in situ visualization solutions, and successfully demonstrated the feasibility and value of visually steering large-scale scientific simulations. The new visualization technologies my group has introduced help drastically raise scientists' productivity by allowing them to verify their data, uncover previously unknown information, and more effectively communicate with others their findings, i.e., telling stories from their data using visualization. To further enhance the value of visualization in the overall scientific workflow, we have been conducting usability studies, which allow us to identify areas for improvement. We have also addressed the need to support collaborative work by designing visualization interfaces for facilitating multichannel communication, as well as reusing and sharing data analysis strategies and results among participants.
I continue to launch new threads of research in data visualization and for developing new visualization applications. I welcome prospective students and collaborators to join our research endeavor.
Following the resurgence of artificial intelligence technology
in recent years, in the field of visualization, there is the growing interest
and opportunity in applying AI and machine learning to perform data transformation
and to assist the generation of visualization, aiming to strike a balance between cost and quality.
The other aim is to make optimal use of machine intelligence and human intelligence.
Combining machine learning and visual analytics
can empower analysts' ability to interpret and discover from large, complex data,
and to predict and make critical decisions.
The other active line of research to pursue is the use of
visualization to enhance AI.
Comparative Visual Analytics for Assessing Medical Records with Sequence Embedding, J. of Visual Informatics (in press). Rongchen Guo et al. [Preprint]
A Visual Analytics System for Multi-model Comparison on Clinical Data Predictions, J. of Visual Informatics (in press). Yiran Li et al. [Preprint]
Supporting Analysis of Dimensionality Reduction Results with Contrastive Learnin, IEEE TVCG 26(1):45-55 (2020). Takanori Fujiwara et al. [DOI]
A Deep Generative Model for Graph Layout, IEEE TVCG 26(1):665-675 (2020). Oh-Hyun Kwon and Kwan-Liu Ma [DOI]
What Would a Graph Look Like in This Layout? A Machine Learning Approach to Large Graph Visualization, IEEE TVCG 24(1):478-488 (2018). Oh-Hyun Kwon, Tarik Crnovrsanin, Kwan-Liu Ma [DOI]
Visual Reasoning of Feature Attribution with Deep Recurrent Neural Networks, IEEE International Conference on Big Data 2018. Chuan Wang et al. [DOI]
Intelligent Classification and Visualization of Network Scans, VizSEC 2007, pp. 237-253. Chris Muelder et al. [DOI]
An Intelligent System Approach to Higher-Dimensional Classification of Volume Data, IEEE TVCG 11(3):273-284 (2005). Fan-Yin Tzeng, Eric Lum, Kwan-Liu Ma. [DOI]
Opening the Black Box - Data Driven Visualization of Neural Network, IEEE Visualization 2005 Conference, pp. 383-390. Fan-Yin Tzeng and Kwan-Liu Ma [DOI]
The increasing quality and decreasing price of wearable VR devices
inspires us to exploit the immersive space for data visualization
and analysis. We are developing new immersive visualizaion applications,
determining how to represent and lay out data in immersive space, and studying
associated interaction problems. We also investigate the use of AR for
collaborative visual analysis tasks.
Collaborative Visual Analysis with Multi-Level Information Sharing Using
a Wall-size Display and See-Through HMDs, IEEE PacificVis 2019 (to appear).
Tianchen Sun, Yucong Ye, Issei Fujishiro, Kwan-Liu Ma
Improving Spatial Orientation in Immersive Environments, ACM SUI 2018. Joseph Kotlarek, I-Chen Lin, and Kwan-Liu Ma [pdf]
A Study of Rendering, and Interaction Methods for Immersive Graph Visualization, IEEE Transactions on Visualization and Computer Graphics, 22(7):1802—1815, 2016. Oh‑Hyun Kwon, Chris Muelder, Kyungwon Lee, and Kwan‑Liu Ma [pdf]
Enabling Interactive Scientific Data Visualization and Analysis with See-through HMDs and a Large Tiled Display, in Proceedings of Workshop on Immersive Analytics, IEEE VR 2016. Ken Nagao, Yucong Ye, Chuan Wang, Issei Fujishiro, and Kwan-Liu Ma [pdf]
We are entering a data-rich era. Advanced computing, imaging, and sensing technologies enable scientists to study natural and physical phenomena at unprecedented precision, resulting in an explosive growth of data. The size of the collected information about the Web and mobile device users is expected to be even greater. To make sense and maximize utilization of such vast amounts of data for knowledge discovery and decision making, we need a new set of tools beyond conventional data mining and statistical analysis. Visualization has been shown very effective in understanding large, complex data, and thus become an indispensable tool for many areas of research and practice. We have been developing new concepts to further advance the visualization technology as a powerful discovery and communication tool.
The ability to do data triage and visualization during a simulation run becomes more and more essential as scientific supercomputing moves from terascale and petascale to exascle. We have been studying the feasibility and requirements of in situ processing over the past few years. In situ analysis and visualization is clearly on our national research roadmap, and many others are joining us in this very important direction.
Our research in information visualization focuses on very large graph visualization, visual data mining, social network analysis, software evolution, and computer security visualization. We are also interested in studying the information visualization aspect of scientific visualization problems.
We aim to add to existing visualization systems the support for storytelling using illustrative visualizations, animations, annotations, and sound.
Visualizing large, complex volume data demands parallel or Grid-based solutions. We intend to realize a fully parallelized visualization pipeline. We have developed scalable parallel rendering algorithms for a variety of volume data, designed highly efficient software and hardware image compositing solutions, and built clusters targeting large-scale visualization applications. Most of the performance studies have been done on the massively parallel computers operated at LANL, LBL, LLNL, and PSC.
We develop visual interfaces that can help scientists keep track of their visualization experience and findings, use it to generate new visualizations, and share it with others. We also investigate how intelligent systems can assist sophisticated, time-consuming visualization tasks, and consequently simplify the user interfaces for performing the tasks.
We aim at improving the expressiveness of visualizations through the use of artistic inspired methods, non-photorealistic rendering techniques, and highly interactive user interfaces. Visualizations should be made by using the appropriate level of abstraction according to the purpose of visualization, and the visualizations should be perceptually effective to deliver the most relevant information in the data.
Volume data arises in a large subset of scientific, engineering, and biomedical applications. We aim to develop new methodologies for more efficient and effective volume segmentation and visualization.
Time-varying data visualization presents some unique challenges. Our goal is to drastically improve the interactivity and explorability of large-scale, time-varying data visualization through the study and development of innovative data reduction methods, rendering and interaction techniques, and system integration strategies tailored to the characteristics of several representative leading-edge applications. Check out our new NSF ITR project.
Current research projects are sponsored by Chengying/Alibaba, Bosch Research, HP Labs, Nokia Research, AT&T Labs, NSF BigData, NSF FODAVA, NSF PetaApps, NSF HECURA, NSF CyberTrust, DOE ASCR. DOE SciDAC, DOE BER, and Air Force STTR.