|
|
Project Descriptions: 2008
The Computer Science Department's research efforts cover a broad spectrum of areas. Much of our research is cross-disciplinary, and many faculty members are engaged in this effort.
Applicants will be able to select three projects as their top preferences on the application form. Efforts will be made to match students with their preferred topics as closely as possible.
| Title: |
AuthentiFace |
| Description: |
Have you ever wanted to simply sit in front of your laptop and it instantly logs you in? Would you like to be part of a team that is developing such an exciting technology? Our group is currently researching systems capable of identifying rapidly and reliably, especially under adverse conditions, using novel methods for face recognition. In this project, you will develop an interface for login into a laptop using the image capture from the laptop's webcam. You will also have a chance to work with a team of talented, excited, goal-driven developers/researchers in a relax and friendly environment. This is a unique opportunity for an undergraduate to be exposed to state-of-art algorithms that is currently holding the best published result.
|
| Specific Requirements: |
We are looking for a driven and dynamic individual. Knowledge of object-oriented programming, computer vision and computer graphics techniques is desirable |
| Faculty Sponsor: |
Ioannis A. Kakadiaris |
| Web Link: |
http://www.cbl.uh.edu/URxD/ |
| Topic ID: |
IK1 |
| Title: |
Compiler-Driven I/O Management |
| Description: |
Access times to external memory, in particular magnetic disks, have not kept pace with the speed of processors. As a consequence, more and more programs become I/O-bound. This is especially serious for implicit I/O, those I/O operations that occur as consequence of insufficient memory, because implicit I/O tends to be hidden from the user. (For example, paging in virtual memory management is usually implicit I/O.) It is crucial to develop techniques that permit intelligent, automatic restructuring of high-performance programs so that the implicit I/O operations are as few as possible.
Objectives: Design and implement selected techniques.
Students will learn compiler optimization techniques as applied to I/O management. |
| Specific Requirements: |
N/A |
| Faculty Sponsor: |
Ernst L. Leiss |
| Web Link: |
http://www.cbl.uh.edu/~reu/2007/EL2.html |
| Topic ID: |
EL2 |
| Title: |
Computer Security |
| Description: |
We are looking for enthusiastic, passionate and bright students for all our projects. This project involves the study and analysis of how man-in-the-middle attacks can be prevented in cryptographic protocols.
What Student will learn: Computer security in general and cryptographic protocols in particular.
|
| Specific Requirements: |
None |
| Faculty Sponsor: |
Rakesh Verma |
| Web Link: |
http://www.cs.uh.edu/~rmverma |
| Topic ID: |
RV4 |
| Title: |
Computing Physiology 24/7 |
| Description: |
The student will work as part of a large team consisting of Ph.D. students and research engineers in the Computational Physiology Lab (CPL) as well as the Mayo College of Medicine (obesity studies), the University of Texas Medical School (sleep studies), the Methodist Hospital (surgical studies), and the Communication Lab of SUNY Buffalo (psycho-physiological studies). Strong emphasis is placed on addressing the algorithmic, computational, and human-computer interaction aspects of the problem. This is not only an intriguing scientific project but also a unique opportunity for the undergraduate student to gain experience of modern multi-disciplinary and multi-institutional team research.
|
| Specific Requirements: |
|
| Faculty Sponsor: |
Ioannis Pavlidis |
| Web Link: |
http://www.cpl.uh.edu/html/ |
| Topic ID: |
IP1 |
| Title: |
Data compression package for TinyOS |
| Description: |
The project is to develop a data compression and processing package for TinyOS (version 2.0) that implements standard compression techniques such as LZ77, LZW, wavelet transformation etc. A secondary objective is to benchmark the computation speed and energy cost of different operations. Students will have the opportunity to work with embedded sensor devices and operating systems. The broader context of the project is for wireless monitoring of critical infrastructures. For more information, please feel free to contact rzheng@cs.uh.edu
|
| Specific Requirements: |
Strong background in C/C++ programming. Knowledge in nesC and data compression is a plus. |
| Faculty Sponsor: |
Rong Zheng |
| Web Link: |
Wireless research group at UH, http://wireless.cs.uh.edu
TinyOS, http://www.*tinyos*.net |
| Topic ID: |
RZ1 |
| Title: |
Data Mining and Information Extraction |
| Description: |
We are looking for enthusiastic, passionate and bright students for all our projects. This project will investigate how to mine and analyze information from medical documents (research papers, drug studies, etc.) and financial documents (SEC filings).
What Student will learn: Student will learn text mining, information extraction techniques and meta-analysis.
|
| Specific Requirements: |
None - interest in statistics is desirable |
| Faculty Sponsor: |
Rakesh Verma |
| Web Link: |
http://www.cs.uh.edu/~rmverma |
| Topic ID: |
RV1 |
| Title: |
Digital Watermarks |
| Description: |
Watermarks have attracted increased attention as concerns about establishing ownership of digital media have escalated. Robust invisible watermarks allow one to attach an indelible stamp of ownership; clearly the methods employed must be impervious to operations such as rescaling, filtering, or superimposing an additional watermark. Robustness is related to the redundancy of the watermark (e. g., if a certain small pattern is repeated many times in a watermark, the removal of the watermark through cropping an image is foiled). Similarly, the invisibility of a watermark is related to the extent of changes in the information that makes up the media. This imposes limits on the amount of information that can be encoded in the watermark.
Objectives: Study digital watermarks and determine their properties.
Students will learn about aspects of digital watermarks.
|
| Specific Requirements: |
N/A |
| Faculty Sponsor: |
Ernst L. Leiss |
| Web Link: |
http://www.cbl.uh.edu/~reu/2007/EL1.html |
| Topic ID: |
EL1 |
| Title: |
Identification of Patterns in Message Traffic |
| Description: |
Understanding the behavior of a distributed application involves extracting the global communication topology from a matrix of communicating neighbors developed from message traffic analysis. The basic problem to be addressed is whether a communication graph represents, e.g., a grid, tree, a stencil or another known topology. The underlying theoretical problem is graph subisomorphism which is known to be NP-hard. However, several packages have been developed to solve this problem efficiently for many graphs. This project involves pushing the boundaries of graph isomprphism and subisomorphism heuristics ¡V which classes, types, and scales of graphs can be solved effectively with today's technologies? The project is lies at the intersection of parallel systems and algorithms and involves experimentation and analysis. Contact jaspal@uh.edu for more information. |
| Specific Requirements: |
|
| Faculty Sponsor: |
Jaspal Subhlok |
| Web Link: |
http://www2.cs.uh.edu/~jsteach/HPSL/ |
| Topic ID: |
JS1 |
| Title: |
Image Analysis for Cardiovascular Screening |
| Description: |
Every year 1.4 million Americans suffer a heart attack; in 2004, over 800,000 of these attacks were fatal. Developing novel technologies necessary for successful screening of individuals at risk for heart disease is becoming progressively more critical, as the number of fatal attacks increases. Toward this end, our group, in collaboration with cardiologists worldwide, is developing advanced computational tools for the analysis of medical data from Computer Tomography and Intravascular Ultrasound. The student will develop a system for specific tasks related to the analysis of Computer Tomography data. The student will interact with an expert group of collaborators, researchers, and other students in the fields of cardiology and computer vision-all while gaining invaluable experience in academic research, software development, and image analysis. This is not only an exciting scientific project but also a project with a unique societal impact. |
| Specific Requirements: |
We are looking for a driven and dynamic individual. Knowledge of object-oriented programming, computer vision and computer graphics techniques is desirable. |
| Faculty Sponsor: |
Ioannis A. Kakadiaris |
| Web Link: |
http://www.cbl.uh.edu/CARDIA |
| Topic ID: |
IK2 |
| Title: |
Information Extraction |
| Description: |
We are looking for enthusiastic, passionate and bright students for all our projects. This project involves the implementation of a specialized pattern language to extract information from documents.
What Student will learn: Student will learn information extraction and text mining techniques.
|
| Specific Requirements: |
Programming experience in at least one of C/C++/Java |
| Faculty Sponsor: |
Rakesh Verma |
| Web Link: |
http://www.cs.uh.edu/~rmverma |
| Topic ID: |
RV3 |
| Title: |
Information Extraction and Text Mining |
| Description: |
We are looking for enthusiastic, passionate and bright students for all our projects. This project will investigate how to automatically summarize different types of documents and evaluate the quality of the summaries.
What Student will learn: Student will learn text mining, information extraction techniques and meta-analysis.
|
| Specific Requirements: |
None - interest in statistics is desirable |
| Faculty Sponsor: |
Rakesh Verma |
| Web Link: |
http://www.cs.uh.edu/~rmverma |
| Topic ID: |
RV5 |
| Title: |
Information Retrieval, Search Engines |
| Description: |
We are looking for enthusiastic, passionate and bright students for all our projects. This project will investigate meta-search engines and ranking. One specific issue will be how to combine different rankings of the same information nuggets.
What Student will learn: Student will learn information retrieval techniques including ranking, search engines and meta-search engines.
|
| Specific Requirements: |
None - interest in statistics is desirable |
| Faculty Sponsor: |
Rakesh Verma |
| Web Link: |
http://www.cs.uh.edu/~rmverma |
| Topic ID: |
RV2 |
| Title: |
Integrated Shape Models |
| Description: |
The student will be involved in research efforts targeted at deriving mathematical models for combining 2D shape representations that are based on appearance and feature data of 3D objects. The student will work closely with other members of the Quantitative Imaging Lab (QIL) who are already working on this project and have developed the individual models. The student will leverage fundamentals of computer vision, image processing, mathematical and statistical modeling, and software engineering in his/her effort. The ability to derive mathematical representations of 3D objects is critical in designing next generation intelligent visual systems. |
| Specific Requirements: |
Knowledge of Visual C++/C++ and/or Java and basic math (calculus and linear algebra). Knowledge of computer vision and computer graphics is desirable. |
| Faculty Sponsor: |
Shishir Shah |
| Web Link: |
http://www.cs.uh.edu/~shah |
| Topic ID: |
SS1 |
| Title: |
Intrusion Detection Monitoring System |
| Description: |
In order to avoid being detected, computer hackers typically go through a long chain of computers to break into a target machine.? The computers along the chain are called stepping-stones for the attack. Real-time algorithms that can detect such intruders have been designed.? The project involves the integration of known algorithms into a system that can provide system administrator warnings. |
| Specific Requirements: |
We are seeking students with some programming experience in C/C++ or Java, knowledge of OS or computer graphics a plus. |
| Faculty Sponsor: |
Stephen Huang |
| Web Link: |
http://www2.cs.uh.edu/shuang/IDRG/index.html
http://www.cs.uh.edu/shuang/ |
| Topic ID: |
SH1 |
| Title: |
Mining and Visualization of Cell Population Data |
| Description: |
Recently, flow cytometry has become an important technique in hematology, and it can be used to detect and identify the minor cell population from bone marrow or blood. Most of the current studies are still based on manual gating by medical scientist and researchers. This process is not only labor intensive, but also may mislays some potential cells in other dimensions. Our current would like to develop an automatic framework to visualize and analyze the flow cytometry readouts. The result can provide doctors and physicians useful information to diagnose blood or lymphatic diseases, such as Leukemia, Myeloma, and Lymphoma. We are seeking students with data mining and/or Matlab skills to help us analyze the data. |
| Specific Requirements: |
We are seeking students with data mining and/or Matlab skills to help us analyze the data. |
| Faculty Sponsor: |
Stephen Huang |
| Web Link: |
http://probes.invitrogen.com/resources/education/tutorials/4Intro_Flow/player.html
http://www.cs.uh.edu/shuang/ |
| Topic ID: |
SH2 |
| Title: |
Transitioning from Algorithms to Software |
| Description: |
Algorithm analysis is a well-studied discipline, as is software development. However, at the interface between these two disciplines much can and does go wrong. In fact, many programmers have experienced situations where a good algorithm (that is, correct and efficient) resulted in either wrong or unacceptably slow software. The causes of the differences between the behaviors of algorithms and software can be categorized into several areas, namely the implications of the non-uniform memory in real architectures (both caches and virtual memory management are implicated), system issues (memory mappings, passing of parameters, garbage collection, and optimization techniques are important here), implicit assumptions (including exception handling), and the finiteness of the number representation (which does not only have relevance for numerical applications, but is also important if one tests for equality or assumes mathematical identities hold).
Objectives: Explore assumptions of different computing paradigms and determine how their differences affect software.
Students will learn how to obtain good software from algorithms.
|
| Specific Requirements: |
N/A |
| Faculty Sponsor: |
Ernst L. Leiss |
| Web Link: |
|
| Topic ID: |
EL3 |
|
| Title: |
Visual Metrology |
| Description: |
The student will be developing a system for user-guided and automated measurement of objects in visual scenes. Ongoing research in this area is motivated by the necessity to understand the mathematical theory underlying invariant transformations of visual metrics under projective geometries. A variety of applications ranging from forensics, surveillance and security, graphical reconstruction and modeling, robotics and navigation, etc. will be impacted by this work. The student will interact with other members of the Quantitative Imaging Lab (QIL) who are already working on this project and have developed prototype algorithms and applications.
|
| Specific Requirements: |
Knowledge of Visual C++/C++ and/or Java and basic math (calculus and linear algebra). Knowledge of computer vision and computer graphics is desirable. |
| Faculty Sponsor: |
Shishir Shah |
| Web Link: |
http://www.cs.uh.edu/~shah |
| Topic ID: |
SS2 |
|
