goals

schedule

references

previous years

Schedule

Abstracts

---

Evaluating a Script-Based Approach for Simulating Patient-Doctor Interaction
by Robert Dickerson

Our virtual patient system uses life-size, multimodal, interactive virtual humans to simulate the patient-doctor interview. We evaluate the performance of our script-based approach for natural language understanding in the limited domain of the medical history interview. Although medical students felt the system was not perfect, they felt the virtual patient performed well for learning diagnosis skills. We discuss the reasons for a script-based approach, what worked and what did not work, and future solutions for increasing recognition accuracy.

---

Experiences in Using Immersive Virtual Characters to Educate Medical Communication Skills
by Kyle Johnsen

This paper presents a system which allows medical students to experience the interaction between a patient and a medical doctor using natural methods of interaction with a high level of immersion. We also present our experiences with a pilot group of medical and physician assistant students at various levels of training. They interacted with projector-based life-sized virtual characters using gestures and speech. We believe that natural interaction and a high level of immersion facilitates the education of communication skills. We present the system details as well as the participants' performance and opinions. The study confirmed that the level of immersion contributed significantly to the experience, and participants reported that the system is a powerful tool for teaching and training. Applying the system to formal communication skills evaluation and further scenario development will be the focus of future research and refinement.

---

Distributed Bayesian Algorithms for Fault-Tolerant Event Region Detection in Wireless Sensor Networks
by S. S. Iyengar

Wireless sensor networks are envisioned to consist of thousands of devices, each capable of some limited computation, communication, and sensing, operating in an unattended mode. According to a recent National Research Council report, the use of such networks of embedded systems "could well dwarf previous revolutions in the information revolution." To our knowledge, this is the first paper to propose a solution to the fault-event disambiguation problem in sensor networks. Our proposed solution, in the form of Bayesian fault recognition algorithms, exploits the notion that measurement error due to faulty equipment are likely to be uncorrelated, while environmental conditions are spatially correlated. We show through theoretical and simulation results that the optimal threshold decision algorithm we present can reduce sensor measurement faults by as much as 85-95 percent for fault rates up to 10 percent.

---

Introduction to Quantum Information Processing
by Pawel Wocjan

A quantum computer is a (hypothetical) computational device that harnesses physical phenomena unique to quantum mechanics in order to realize a fundamentally new mode of information processing. The aim is to solve (some) computational problems more efficiently then any classical computer can do. In this talk, I give a short introduction to quantum computing. I start with some basic concepts of quantum mechanics and then introduce the so-called quantum circuit model, the most common model for quantum computers. I present the Deutsch-Jozsa algorithm, a simple algorithm that captures the basic ideas behind more complicated quantum algorithms. Then, if time permits, I will explain Shor's celebrated algorithm for factoring integers; and give a very brief sketch of my research in quantum information processing, on the so-called mutually unbiased bases problem.

---

A Time-Optimal Delaunay Refinement Algorithm in Two Dimensions
by Alper Ungor

We propose a new refinement algorithm to generate size-optimal quality-guaranteed Delaunay triangulations in the plane. The algorithm takes $O(n \log n + m)$ time, where $n$ is the input size and $m$ is the output size. This is the first time-optimal Delaunay refinement algorithm.
This is joint work with Sariel Har-Peled and accepted for ACM-SoCG05.

---

Evaluating the expected performance of sensing systems for interactive computer graphics
by Danette Allen

We introduce a general method for evaluating and comparing the expected performance of sensing systems for interactive computer graphics. Example applications include head tracking systems for virtual environments, motion capture systems for movies, and even multi-camera 3D vision systems for image-based visual hulls. Our approach is to estimate the asymptotic position and/or orientation uncertainty at many points throughout the desired working volume, and to visualize the results graphically. This global performance estimation can provide both a quantitative assessment of the expected performance, and intuition about the type and arrangement of sources and sensors, in the context of the desired working volume and expected scene dynamics.

---

Fast and Accurate Protein Structure Prediction
by Jinbo Xu

If we know the primary sequence of a protein, can we predict its three-dimensional structure by computational approaches? This is one of the most important and difficult problems in computational molecular biology and has tremendous implications for the field of proteomics. In this talk, I will present two efficient algorithms to solve two key problems of protein structure prediction: backbone prediction by protein threading technique and side-chain prediction. Both of these problems have proved to be NP-hard and hard to approximate. First, I will present a linear integer programming approach to the protein threading problem. With careful formulation, this approach will lead to an amazing result: almost all the relaxed linear program instances will produce integral solutions directly, without using branch-and-bound. This result indicates that the protein threading problem is tractable empirically although it is hard theoretically. Next, I will describe a novel approach to the protein side-chain packing problem, based on the tree-decomposition of a protein structure. This algorithm not only runs very efficiently in practice, but also yields a polynomial-time approximation scheme for some types of objective functions. I have implemented the proposed algorithms combined with several other components as two protein structure prediction programs RAPTOR (for backbone prediction) and SCATD (for side-chain prediction). RAPTOR ranks first among all the programs of its category in CASP5 and third in CASP6, the most recognized competition in the area of protein structure prediction. SCATD achieves an average speed five times faster than SCWRL 3.0, a widely-used side-chain packing program, and runs up to 90 times faster on large proteins.

References

References from the talks as well as the presentation materials will be available here (upon speakers approval).

Previous years

Fall04	Spring04	Fall03	Spring03	Fall02	Spring02
Fall01	Spring01	Fall00	Spring00	Fall99	Spring99

goals

schedule

references

previous years

Alper Üngör (ungor@cise.ufl.edu) August 2004