“The algorithm is based on a linear latent variable model, whose parameters are estimated using independent component analysis involving statistical selection of independence support segment and posterior scale normalization,” said Yue Joseph Wang, associate professor of electrical and computer engineering at ARI. 

ARI Telecommunications Milestone

Late last month, ARI graduate student and research faculty met across international borders using video conferencing technology, marking a first for ARI.

Noor Maricar, ARI graduate student, successfully defended his Electrical Engineering PhD thesis from his home country of Malaysia, Aug. 31st.

Committee members present were Dr. Saifur Rahman, ARI director and committee chair; Dr. Scott Midkiff, professor of electrical engineering; Dr. Adil Godrej, assistant professor and associate director, civil and environmental engineering; and Fred Krimgold, co-director, Virginia Tech Center for Disaster Risk Management. Dr. Yilu Liu, professor of electrical engineering, joined from her office in Blacksburg. 

Three technologies made the three way visual and audio connection possible: Yahoo Messenger to broadcast participants’ images, conference calling to speak, and Microsoft Live Meeting to view Maricar’s dissertation slides online.

At the meeting’s conclusion, ARI’s newest PhD graduate said, “I invite you to submit a paper to our conference in Kuala Lumpur, and I owe you all a big durian (southeast Asian fruit) and some fried bananas.”

 

ARI Professor Invents Bioimaging Technique

Dr. Yue Joseph Wang, associate professor, Bradley Department of Electrical and Computer Engineering at ARI, has been granted a new US Patent, #6,728,396, on a Partial Independent Component Analysis (PICA) technique for independent component imaging.

“The algorithm is based on a linear latent variable model, whose parameters are estimated using independent component analysis involving statistical selection of independence support segment and posterior scale normalization,” Wang said. “When tested on various multivariate data sets with the ground truth, accurate yet robust source separation has been achieved remarkably. We anticipate this computational approach would have considerable applications in a wide variety of composite data visualization and imaging studies.”

The main application of PICA is blind source separation for data sources which until now could only be imaged as composite observations. Opportunities for PICA application exist in areas of information study where hidden sources can be recovered from heterogeneous data mixtures. Such opportunities can be found in biomedicine, bio-defense, intelligence analysis and market analysis. For example, “dissect gene expression profiles of mixed heterogeneous cell phenotypes, or simultaneous imaging of multiple functional molecular biomarkers or any other signatures,” Wang said.

Wang received his B.S. and M.S. from the Shanghai Jiaotong University, his Ph.D. in Electrical Engineering from University of Maryland. He completed a Postdoctoral Fellow at the Georgetown University School of Medicine. His most recent and current work has been supported (among others) by the National Institutes of Health.

Wang has recently been selected as a member to serve on the newly established IEEE Signal Processing Society Technical Committee on Bio-Imaging and Signal Processing (BISP); and is currently a member of IEEE-SPS Technical Committee on Machine Learning for Signal Processing (MLSP).

Contact Yue Wang at yuewang@vt.edu 

 

 

Read past issues:

September

August

July

June

      Virginia Tech's  

  Alexandria Research

            Institute

 206 N. Washington St. 

            Suite 400
   Alexandria, VA 22314
      P: (703) 518-8080 
      F: (703) 518-8085

           

   Copyright ARI 2004