Spring 2005 Data and Vision Weekly Seminar

Department of Computer Science
University of Texas at San Antonio

Schedule:

02/18/05, Improving perception of intersecting 2D scalar fields
02/25/05, Wavelet decomposition of data streams
03/04/05, Joint presentation on GPUs for speeding up
03/11/05, Experimental Analysis of Wavelet Transforms for Estimating PSK Symbol Computation
03/25/05, Virtual grasping
04/08/05, Review of Generalized Belief Propagation
04/15/05, Low dimensional representations
04/29/05,
05/06/05



Schedule for Spring 2005

• 02/18/05 Improving perception of intersecting 2D scalar fields
  Speaker: Mark Robbins
  Presentation slides in [PDF]
  Abstract:
  
  The visualization of 2D scalar fields as 3D surfaces can provide useful insight into the nature of the data, as well as relationships between the fields. While 3D rendering attributes, such as color, transparency, lighting, and texturing, provide helpful techniques in conveying surface shape, perceptual accuracy can be difficult when these surfaces intersect. Our research proposes a method for decomposing intersecting surfaces at the lines where intersection occurs, thus allowing independent control of the rendering attributes for each decomposed component. We have devised a user study to investigate the impact this decomposition technique has on overall visualization perception.
  References:

  1. M. Robinson and K. Robbins, "Towards perceptual enhancement of multiple intersecting surfaces," Visualization and Data Analysis, San Jose CA, 2005. [PDF]
     
  2. V. Interrante, H. Fuchs, and S. M. Pizer, "Conveying the 3D shape of smoothly curving transparent surfaces via texture," IEEE Transactions on Visualization and Computer Graphics, 3(2), pp. 98-117, 1997. [PDF]
     
  3. D. House and C. Ware, "A method for the perceptual optimization of complex visualizations," Advanced Visual Interface, Trento Italy, pp. 148-155, 2002. [PDF]
     

  Back to top

  02/25/05, Wavelet decomposition of data streams
  Speaker: Dragana Veljkovic
  
  Presentation slides in [PDF]
  Abstract:
  

  Streaming data arises in a variety of applications from mining of web logs and anomaly detection in network to satellite imagery. One of the problems that arise in streaming scenarios is the massive scale of updates of the underlying signal over time. Another problem is the need to correctly process the data in one pass. The goal of this research is to examine to what extent data streams can be summarized in a small amount of space so that accurate estimates can be provided for basic queries of the underlying signal. The techniques for computing small space representations are inspired by traditional wavelet based approximations that consist of specific linear projections of the underlying data. A general sketch-based method for computing various linear projections is presented. We also show it is possible use to provide pointwise and rangesum estimation of data streams. These methods use small amounts of space and per-item time while streaming through data and provide accurate representations as the experiments with real data show.
  References:

  1. A. C. Gilbert, Y. Kotidis, S. Muthukrishnan and M. J. Strauss, "One-pass wavelet decomposition of data streams," IEEE transactions on knowledge and data engineering, Vol. 15, No. 3, May/June 2003. [PDF]
     
  2. A. C. Gilbert, Y. Kotidis, S. Muthukrishnan and M. J. Strauss, "Surfing wavelets on streams: one-pass summaries for approximate aggregate queries," Proceedings of the 27th VLDB Conference, Roma, Italy 2001. [PDF]
     
  3. A. C. Gilbert, S. Guha, P. Indyk, Y. Kotidis, S. Muthukrishnan and M. J. Strauss, "Fast, small-space algorithms for approximate histogram maintenance," STOC'02, May 19- 21, 2002, Montreal, Quebec, Canada. [PDF]
     

  Back to top

  03/04/05, General Purpose Computation Using Graphics Hardware
  Speaker: Egle Pilipaviciute and Mark Robbins
  Presentation slides [PDF]

  Abstract:

  Advances in GPU (graphics processing unit) technology have accelerated dramatically in the past few years and GPUs are now receiving much attention in general purpose computation. Some of the major hurdles still facing GPU technology are its nascent operation set, lack of abstract language support, and absence of automatic compiler optimization. The goal of our project is to better understand current GPU technology and investigate the use of the GPU as a general-purpose processing unit (GPGPU). We have performed several experiments of CPU-based computational problems ported to a GPU analog. We present an analysis comparing the performance of the CPU and GPU during these computational tasks.

  References:

  1. Randima Fernando and Mark J. Kilgard, The Cg Tutorial, Pearson Education, Inc., Boston, MA., 2003
     
  2. nVidia Corporation, GPU Gems, Pearson Education, Inc., Boston, MA., 2004
     
  3. SIGGRAPH 2004 GPGPU COURSE http://www.gpgpu.org/s2004/
     

  Back to top

  03/11/05, Experimental Analysis of Wavelet Transforms for Estimating PSK Symbol Rate
  Speaker: Kenneth Holladay
  Presentation slides
  Abstract:

  For automated surveillance applications, estimating the symbol rate of an unknown digital communication signal is an important step in the analysis process. Several papers have investigated using the wavelet transform in symbol rate estimation algorithms. Due to its complexity, closed form analysis of performance is often limited, and simulations may not include practical factors such as carrier frequency offset or symbol pulse shaping. This paper uses an automated statistically based test framework to investigate the performance of the wavelet transform against PSK signals with parameters that span a realistic portion of the High Frequency (HF) signal space. The analysis identifies signal and algorithm parameters that affect performance. We also demonstrate that accurate metrics for estimating the probability of failure/success under realistic operating conditions are available for the db6 wavelet.

  References:

  1. K. L. Holladay, and K. Robbins, "Experimental Analysis of Wavelet Transforms for Estimating PSK Symbol Rate". [PDF]
     

  Back to top

  03/18/05
  Spring Break

  03/25/05, Virtual grasping
  Speaker: Rachel Smith
  Presentation slides
  Abstract:
  

  Data gloves and other virtual reality devices are dropping in cost due to rapid developments in the gaming industry. This project investigates the feasibility of using a data glove for manipulating 3D objects in the user's normal desktop environment. Data gloves have sensors that measure finger and hand angles, as well as a transmitter for measuring hand position and orientation. Measurements are gathered while the user is wearing the data glove and performing basic operations such as opening and closing the hand, moving the hand and arm in different directions, and tracking objects on the screen. The collected data measurements are then used to test machine learning algorithms for detecting virtual grasping with data gloves. We analyze the effectiveness of an Adapter Resonance Theory (ART2) neural network for performing these recognition tasks.

  References:

  1. M. Zacksenhouse, P. Marcovici, "Interactive recognition of simultaneous manipulative hand movements," Mechatronics 11, 2001; 389-407. [PDF]
     
  2. M. Zacksenhouse, P. Marcovici, "Inherent structure of manipulative hand movements and its discriminative power," Intelligent Robots and Systems 1, 2000; 318-323. [PDF]
  3. J.A. Freeman, D. M. Skapura, Neural networks, algorithms, applications, and programming techniques, Reading: Addison-Wesley, 1992.
     
     
  4. Elliott J.M. and K.J. Connolly. A classification of manipulative hand movements. Developmental Medicine & Child Neurology 26 1984; 283-296. [PDF]

  Back to top

  04/01/05, No seminar.

  04/08/05, Review of Generalized Belief Propagation
  Speaker: Yufang Yin
  Presentation slides
  Abstract:

  Important inference problems in statistical physics, computer vision, error-correcting coding theory, and artificial intelligence can all be reformulated as the computation of marginal probabilities on factor graphs. The belief propagation (BP) algorithm is an efficient way to solve these problems that is exact when the factor graph is a tree, but only approximate when the factor graph has cycles. The generalized belief propagation algorithm is an extension of Loopy belief propagation that has been shown to provide near-optimal performance in many applications. For example, under non-trivial interference conditions, such as detection problem in two-dimensional channels with memory, the performance of this fully tractable GBP receiver is almost identical to the performance of the optimal maximum a-posteriori (MAP) receiver.
  

  References:

  1. O. Shental, A. J. Weiss, N. Shental and Y. Weiss, "Generalized Belief Propagation Receiver for Near-Optimal Detection of Two-Dimensional Channels with Memory," IEEE Information Theory Workshop (2004) [PDF]
     
  2. J. S. Yedidia, W. T. Freeman, and Y. Weiss, "Understanding Belief Propagation and its Generalizations," Mitsubishi Electric Laboratories, Tech. Rep. TR-2001-22 January 2002. [PDF]
     
  3. J. S. Yedidia, W. T. Freeman, and Y. Weiss, "Constructing Free Energy Approximations and Generalized Belief Propagation Algorithms," Mitsubishi Electric Laboratories, Tech. Rep. TR-2004-040, May 2004. [PDF]
  4. Jonathan Yedidia, Generalized Belief Propagation, Mitsubishi Electric Research Labs.
     

  Back to top

  04/15/05, Low dimensional representations
  Speaker: Egle Pilipaviciute

  Presentation slides
  Abstract
  

  Reference

  Back to top

  04/22/05
  No seminar.

  04/29/05
  Speaker: Tao Wei

  Back to top

  05/06/05
  Speaker: Douglas Pollock

  Back to top


Back to top

---

Questions and Comments?


Please send emails to qitian@cs.utsa.edu, or seminar co-organizers: Kay Robbins, Weining Zhang, Yufei Huang, and Qi Tian.

---