Robotics Lecture is at 10am and repeats at noon. Dr. O'Kane will present the research of his lab, South Carolina Autonomous Robotics Research (SCARR), some of which was mentioned and shown in the Fall 2012 issue of the College's TechnoKids Newsletter. Dr. O'Kane's research is in planning algorithms for robotics and autonomous systems. As robot technology becomes more practical, it becomes increasingly important to design robots that are suitable for domains that are unpredictable and inhospitable, while ensuring that the resulting systems are robust and inexpensive. Because sensing and uncertainty are central issues in robotics, it is essential to understand how to solve robotics problems when sensing is limited and uncertainty is great. Professor O'Kane's interests span sensor-based algorithmic robotics and related areas, including planning under uncertainty, artificial intelligence, computational geometry, sensor networks, and motion planning. See the Edison Lecture Program for more details.

Void Star Creations is an indie game company based out of Lexington, South Carolina. They'll be talking about the game development process and doing a tech demo of one of their games. They're also looking for feedback. Oh, also: Pizza! http://voidstarcreations.com/home For more info see the facbook event page

COLLOQUIUM Department of Computer Science and Engineering University of South Carolina Ontology-driven Data Integration in Biomedicine GQ Zhang Case Western Reserve University Date: October 9, 2012 Time: 1430-1530 (2:30pm-3:30pm) Place: Swearingen 3A75 Abstract We present an ontology-driven data integration environment called PhysioMIMI (Multi-modality, Multi-resource Information Integration Environment for Physiological and Clinical Research) and illustrate a variety of application scenarios of this environment. PhysioMIMI uses a federated data management approach with a domain ontology as the semantic infrastructure driving data integration, query interface design, and data harmonization across clinical studies. The front-end of PhysioMIMI is a reusable and user-friendly query interface called VISAGE (Visual Aggregator and Explorer). The backend of PhysioMIMI uses an ontology-driven Map and Connect approach, in contrast to the traditional ETL (Extract, Transform and Load) process used in a data warehouse approach. The Map and Connect paradigm embodies flexibility for accommodating data quality improvements in source data by pushing data curation tasks upstream in a source-specific, decentralized way, so that updates can be managed distributively throughout the data reuse life-cycle. Dr. GQ Zhang is Professor of Computer Science and Division Chief of Medical Informatics at Case Western Reserve University's Engineering School and Medical School, respectively. He serves as a Director of Biomedical Informatics Core for CTSC, a member of the Consortium of Clinical Translational Science Award of the National Center for Advancing Translational Sciences, Associate Director of Case Comprehensive Cancer Center, Professor of Proteomics and Bioinformatics and Professor in the Center for Clinical Investigation. His research interests spans Data Management in Biomedicine, Biomedical Ontologies and Applications, Ontology Quality Assurance, Clinical Research Informatics, and Theoretical Computer Science. Dr. Zhang has served on numerous panels, editorial boards and programming committees. He is the author of over 120 publications ranging from automata theory, domain theory, ontology, imaging, to clinical research informatics.

Agenda:

• Introductions of Dean Ambler, new faculty, and our CSE Staff (this is for the newer students)
• State of the CSE Department
• One-Minute Madness (a brief presentation by each CSE faculty member about their research)
• Poster Session (Ph.D. students, MS students, Magellan Scholars, and select undergraduate students)

Refreshments: (drinks and hors d'oeuvres) to be served during the poster session, which will be held in the area in front of Amoco. This is your opportunity to learn about the world-class research underway in computing at the University of South Carolina. The research extends from the theory of computing to practical aspects, such as smart-phone apps. It includes computer vision, bioinformatics, multiagent systems, Bayesian reasoning, wireless networking, information security, quantum computing, and robotics. The symposium is also an opportunity to meet the students conducting this research. Awards: Best (and runner-up) Graduate Student Poster

The attachement below is a map of the activities.

Yiwei Zhang will defend his dissertation “Graph based optimization for scientific computing” Monday 6/18/2012 @ 13:30 in SWGN 3A75. "This work investigates transforming scientific computing problems such as phylogenetic analysis, system simulation into abstract graph problems and developing algorithms for optimization. First, we investigate an efficient algorithm for unequal genome phylogeny reconstruction. Currently few method can solve the median problem for unequal genomes which is essential in evolution tree and ancestral genome reconstruction. We present a new distance measurement model for unequal genomes and develop a branch-and-bound algorithm based on multiple breakpoint graph to calculate the median for unequal genomes. We also present a mixture method to infer the ancestral gene order based on maximum likelihood approach and adequate graph median solver. The median solver is extremely time consuming under high rearrangement rate compared with maximum likelihood approach but it gives better accuracy. So our mixture method aims to increase both the speed and accuracy for ancestral genome reconstruction. We also investigate the method to improve the simulation problem for large systems. The Modified Nodal Analysis (MNA) simulation method is widely used to solve electrical system networks. It has the disadvantage that the simulation time increases disproportionally with system size. To accelerate the simulation, latency insertion method has been introduced to partition the original system into a number of smaller sub-systems so that computation will be reduced. But so far there is no automatic method to optimally exploit the existing latency in the system. Our work focuses on an algorithm based on graph search to optimally activate the optional latency decouplers in the network. With the optimal activation, the system is partitioned in the way that maximum simulation speed up will be achieved."