2007 Computer Science Undergraduate Project Showcase

Wednesday, January 31st at 4pm in GB119

Presentation Schedule

• Richard Valenzano (CSC494, Professor Greg Wilson)
  • Visualization of Protein-Protein Interaction Networks
  • The Network Analysis, Visualization, and Graph Analysis Toronto (NAViGaTOR) software is a tool developed by the Jurisica group to analyze protein-protein interaction networks. This project involved researching graph drawing algorithms with the aim of improving upon the previously used force-directed placement method. This algorithm will be examined and compared with its replacement, the GRIP layout algorithm, which will be shown to offer significant improvement in terms of both running time and layout quality.
• Ian Chan (CSC490, Professor Ravin Balakrishnan)
  • Motion Tracking Objects For Video Game Interfaces
  • Exploring motion tracking as an interface to video games. In particular, mapping the orientation of a physical object in the real world to control an object in a virtual 3d environment. The concept of this research project came from the complicated task of flying an aircraft in a flight simulator. Reducing the inputs to a single object that a user can hold provides a more familiar and easy to use interface for novice users. This presentation will outline the research done and the results.
• Muhammad Ali (CSC494, Professor Greg Wilson)
  • Visualizing Genealogical Data
  • TreeViz is a Flash application capable of dynamically fetching data from a genealogical database and incrementally drawing out the corresponding family tree.
• Adrian Dalca (NSERC USRA, Professor Michael Brudno)
  • FRESCO: Flexible Alignment with Rectangle Scoring Schemes
  • Sequence alignment refers to a way of arranging the primary sequences of DNA to identify regions of similarity. These regions can signify structural or functional importance, or show evolutionary relationships. To give a quality score to an alignment, one builds a scoring scheme. While there has been significant work in developing methods to optimize alignments for particular scoring functions, there has been less research as to what are effective scoring scheme. In many simple scoring schemes the scoring function and the optimizing algorithm are tightly related. However for other schemas, finding the optimal alignment is non-trivial. We define a diagonal¢s bounding rectangle as the rectangle in the dynamic programming matrix delimited by the previous diagonal¢s last match and the next diagonal¢s first match. We call the scoring functions that make use of these bounding rectangles ¡rectangle scoring schemes¢. We present FRESCO, an algorithm that will align two sequences in O(n5) time under any scheme from this large class. By requiring some loose conditions on the scoring scheme the algorithm runs, without loss of accuracy, in O(n3(log^2n)) expected time. With further heuristics, it is possible for FRESCO to produce near-optimal alignments with linear running time. Most popular scoring schemes (Needleman- Wunsch, Karlin-Altschul (BLAST) E-value, DIALIGN) fall into the class of rectangular scoring schemes, and hence can be used with FRESCO. However, there is a large amount of important scoring schemes which cannot be computed with other algorithms but can be computed with FRESCO. While for many purposes, such as genome alignment, our algorithm is too slow for optimal solutions, FRESCO allows for effective testing of a large class of scoring models and investigation on how they perform on biological sequences.
• Shahan Khatchadourian (CSC495, Professor Mariano Consens)
  • VisTopK: Visualization of TopK Information Retrieval Algorithms
  • Developed as an Eclipse plug-in, VisTopK is an extensible visual tool to explore properties of and temporally navigate topk information retrieval algorithms applied against heterogeneous indexed collections. The application allows for easy integration of ongoing research as well as existing Eclipse-based tools.
• Derek Yau (CSC490, Professor Ravin Balakrishnan)
  • Next generation video game immersion motion capture
  • The concept of motion capture in video games has reached the mass consumer markets. However, today's consoles currently only use one "motion capture" input device. What happens when we add more and more input devices, allowing the players to express themselves more freely and naturally in games? My course project utilized two input "devices" to do what wouldn't be possible with a standard keyboard/mouse setup. I hope to discuss the future of interactivity in gaming as well.
• Raymond Lo (CSC494/495, Professor Steve Mann )
  • Real-Time Cyborglogs and Computational Frameworks for Cyborglogs
  • The goal of this research project is to develop a camera phone application integrated with a server architecture that enables the continuous archival of personal experience in real-time. Cyborglogger (or simply Glogger) is a camera phone application that allows camera phones to be used for creating cyborgLog ('glog). Glogger is equipped with features that are ideal for "sousveilliance" (the recording of an activity from the perspective of a participant in the activity). One of the many notable features of Glogger includes its ability to continuously capture and upload in real-time, without any conscious thought, the everyday experience of a person. The server architecture includes a custom-built community website that allows a sousveillance community to interact in real time. This leads to the emergence of a new form of communication, namely through the peer-to-peer sharing of visual narratives.
• Serge Gonchar (CSC490, Professor Ravin Balakrishnan)
  • Artist driven tools: Interactive flower modeling and simulation.
  • I explore ways of simulating the growth of plants, specifically flowers, with the hope of being able to draw, redraw and modify the drawn plant with just a stroke of a pen on a tablet. In essence, my research is on how to abstract sliders, input boxes and other various options of 3D modeling tools to just pen and pencil interface.