Faculty Research Areas
Computer Networking (Haiyang Wang)
Computer networking underlies today's mobile devices, home networks and the Internet. Although Networking is becoming a critical infrastructure of our information-based society, it still failed to achieve a reliability level of the traditional telephone systems. Research on computer networks at UMD concentrates on designing highly scalable and efficient networking infrastructures, by combining optimization, economics, and computer science. Besides network optimization, projects at UMD also address application specific issues in cloud computing, peer-to-peer (P2P) and multimedia systems. For example, our study on cloud computing aims to mitigate the performance as well as the energy issues in the existing visualization environments. Our students will be able to test and explore different real-world coud systems (e.g., could gaming/synchronization apps) in our networking lab.
Information Retrieval (Carolyn Crouch, Donald Crouch and Steve Holtz)
For the last several years, research in information retrieval has focused on very short queries and how to improve them. These are typically queries of from 1 to 3 words in length. These queries are of particular interest because they mimic the type of query often input by users of the web. Untutored users (e.g.,, those without knowledge of how to structure search queries) often simply input a word or two in their area of interest and hope to retrieve relevant information. But as those who frequently search the web know, the results returned from such a search are often of questionable value.
Graphics and User Interfaces (Doug Dunham)
Research in computer graphics is centered on creating and visualizing repeating patterns in Euclidean, hyperbolic, and spherical geometry, and on other 2-dimensional surfaces. This research also includes investigations of spherical and hyperbolic spline curves. The user interface research mostly involves issues in the computer creation of motifs for the repeating patterns mentioned above.
Data Mining and Machine Learning (Rich Maclin)
Dr. Maclin pursues research in a number of machine learning and data mining related areas. Recent research has focused on a variety of topics such as the creation of effective classifiers for real world data sets (such as a program to recognize Venusian volcanos from Synethic Aperture Radar (SAR) data for NASA), the development and evaluation of ensemble learning methods (especially bagging and boosting), and effective techniques for priming reinforcement learners with prior knowledge. Dr. Maclin also pursues machine learning related research in the field of Bioinformatics.
Computational Linguistics (Ted Pedersen)
Computational Linguistics represents the intersection of Computer Science and Linguistics, and seeks to identify fundamental properties of human language by automatically processing large quantities of online text. Research at UMD focuses on lexical semantics, in particular discovering syntactic and lexical features that are necessary to automatically identify the meaning of words in text.
Natural Language Processing (Ted Pedersen)
Natural Language Processing seeks to make computers able to understand and use human language in written and spoken form. Research at UMD focuses on inventing techniques that allow a computer to organize and understand large quantites of online text though the use of statistical and machine learning techniques. We are currently developing methods that automatically identify the meaning of words in text, and that organize phrases or documents based on their conceptual similarity.
Formal and Automated Reasoning (Hudson Turner)
Logic-based artificial intelligence. Representing knowledge about the effects of actions. Automated reasoning about actions. Planning. Nonmonotonic logics, including declarative logic programming.
Perception and Computer Graphics (Pete Willemsen)
Dr. Willemsen conducts research focused on how humans perceive and act in immersive virtual environments. The goal of this research is to convey an accurate sense of space to users of virtual environments. This work is centered on understanding how people act in immersive virtual environments and comparing that behavior to performance in similarly constructed real conditions. We are acquiring information about how a person's behavior in either a real or virtual space transfers to their behavior in the opposite space, providing important insights about the mental processes underlying human computer interaction within head-mounted display virtual environments.
Simulation and Environment Representation for Virtual Environments (Pete Willemsen)
Creating active virtual environments that resemble real-life situations is still a formidable task. In addition to difficulties with matching visual quality, populating virtual environments with autonomous activity, such as crowds of pedestrians or roadways filled with vehicles, requires substantial effort. It is necessary to provide a sufficient environment model that supports geometric, topological, socio-cultural, as well as relational environment information. These representations must place a strong emphasis on real-time, interactive solutions capable of supporting hundreds, if not thousands, of virtual entities. Designing virtual experiences in which this autonomous, ambient activity is meshed with replicable behavior that can be used by researchers for experiments is still an open issue. Infusing ambient activity with directable behavior requires tools and frameworks capable of merging requests for changes to otherwise autonomous behaviors, while still allowing actions to be reactive to the human subject.