Research Groups

The database group at UW–Madison was established in 1976 and is one of the top database groups in the world. Database research at UW–Madison focuses on all aspects of data management for modern analytics and data science. Our contributions span across diverse areas, including data platforms for large-scale analytics that run at the speed of the underlying hardware, machine learning systems for end-to-end data integration and enrichment, and foundational work in data management theory. Our goal in the Database Group at UW–Madison is to help users go from data to actionable insights fast. To this end, we put emphasis on 1) building high-impact systems, such as QuickStep (a high-performance in-memory database), Magellan (an end-to-end data integration platform), and HoloClean (a machine learning-based system for data enrichment); and 2) fundamental advances in emerging areas such as data pricing, uncertain data management, and data enrichment. More about the database group here. 

The Visual Computing group takes a broad perspective on how computers are used to create and interact with things we see and do. Our research spans the areas of Computer Animation, Digital Content Creation, Multimedia, Visualization, Virtual Reality, and Human-Robot Interaction. Some current projects include: Virtual Surgery—to simulate operations on the human body for training; High Performance Simulation—to create complex physical effects efficiently; Physical Human-Robot Interaction—to help robots better work together with people; Visual Data Science—to create interactive visualization tools to help people make sense of complex data sets; and Visualization for Machine Learning—to create tools that facilitate the creation and use of data-centric systems.

The group’s research mission is to build human-centered methods and principles for the design of robotic technologies and for their integration into human environments. This interdisciplinary area bridges computer science with human psychology to develop robots with an understanding of the needs, expectations, and concerns of their future users. Following this approach, the group has developed educational robots that can understand whether students are distracted and regain their attention, robotic assistants that know when their users are busy and offer help exactly at the right time, drones that can communicate their intent to bystanders, and companion robots for children that turn reading into a sustained social activity.

We study the theory, algorithms, and applications of machine learning, computer vision, bioinformatics, and natural language processing. Themes in computer vision include computational imaging, design of high-performance 3D and motion cameras, active approaches for medical image analysis, face recognition, image-based modeling and rendering, adversarial attacks and defenses. This research is motivated by applications in robotics, autonomous transportation, consumer imaging, and human-computer interfaces. Machine learning research focuses on analysis of deep learning, adversarial machine learning, multi-armed bandits, active learning, machine teaching, robust learning, non convex optimization including matrix and tensor factorization, and fast randomized large scale learning methods. Research includes applications to bioinformatics, information extraction and data enrichment, adaptive agents, collaborative filtering, and interactive learning systems. Natural language processing research focuses on embedding methods for deep learning, and transfer learning for text data.

The networks group at Wisconsin works on a wide variety of areas including network verification, data center networking, big data systems, Internet robustness, measurements, mobile, wireless networking.  The Wisconsin Internet & Systems Research (WISR) group focuses on making networks and systems “future proof”, i.e., perform well and be robust, even as new, disruptive technologies arise, including both hardware and software. The Wisconsin Wireless and NetworkinG Systems (WiNGS) laboratory builds new, deployable wireless technologies such as high-bandwidth connectivity to moving vehicles, non-WIFI interference detection, etc. The Wisconsin Advanced Internet Laboratory (WAIL) conducts measurement based research to answer fundamental questions about the security, performance, topology and robustness of the Internet including the effects of climate change, detection of Internet level outages. The group also works on lightweight Network Time Protocol (NTP) for mobile and IoT devices and deployment models for edge computing.

Numerical analysis is the study of algorithms that use numerical approximation (as opposed to general symbolic manipulation) for mathematical analysis.  Researchers in our department specialize in approximation theory, splines, wavelets, Gabor systems, splines, and polynomial interpolation. Applications include data and signal analysis, including compression, denoising, deconvolution and more.

The Optimization group’s profile has continued to grow as our research becomes more and more crucial to many areas of science, engineering, and economics. Researchers in the group develop new methods, explore their mathematical properties, and apply them to a wide range of important real-world problems. Examples of ongoing projects include optimization algorithms for machine learning and data science, strategic optimization for improved water quality and enhanced natural resources, and design and operation of more reliable electrical power grids. Our tools and techniques benefit our domain science collaborators by enabling them to formulate and model their problems, solve these formulations using efficient algorithms, and extract relevant information from their data. Our group has been involved in projects of this nature in ecology, medicine, statistics, agriculture, chemistry, control engineering, and genetics. Two important online resources for researchers and users of Optimization—NEOS and Optimization Online—are  both hosted at Wisconsin.

The security group has a wide-range of interests, including network security, language-based security (e.g.,  machinecode analysis and software “debloating”), differential privacy, and adversarial machine learning.  Some of our recent work has focused on methods for detecting online attacks and fraud.  The Software Assurance Marketplace (SWAMP) is creating a state-ofthe-art facility that will serve as an open resource for software developers, assurance tool developers and researchers who wish to perform continuous assurance (CSwA) testing in a safe, secure environment. Funded by the Department of Homeland Security, and directed by academic experts in identity management and high throughput computing, the SWAMP’s mission is to improve the safety and quality of our software ecosystem by creating access to assurance tools and testing, and reporting.