Activity Report from the MLSP-TC

by Kenneth E. Hild II (MLSP Liaison to E-Newsletter)

We would like to take this opportunity to share with you some recent news and work related to our MLSP-TC.

MLSP Workshop:

This year's annual Machine Learning for Signal Processing (MLSP) Workshop will be held on September 18-21, 2011, at the Institute of Automation, Chinese Academy of Sciences, in Beijing, China. Beijing is home to the Bird's Nest, the 0.5 billion USD National Center for the Performing Arts (nicknamed The Egg), the Forbidden City, Tiananmen Square, Imperial Palace, the Great Hall of the People, and is a short drive from the Great Wall.

The paper submission deadline has been extended to May 22, 2011. We are accepting 6-page papers written in any area related to machine learning for signal processing. This includes, but is not limited to, machine learning and statistical signal processing for supervised, unsupervised, reinforcement, and semi-supervised learning, probabilistic modeling and kernel methods, artificial neural networks, and advanced (adaptive, nonlinear) signal processing. Potential applications include: speech processing, image processing (computer vision), pattern recognition, medical imaging, multimodal interactions, intelligent multimedia and webprocessing, robotics, sonar and radar, biomedical engineering, time series prediction, blind source separation, data fusion and mining.

MLSP Face Recognition Competition:

The annual MLSP Workshop is hosting their 7th annual competition. This year's competition will be on the application of face recognition. There has already been a tremendous response to this year's competition, but there is still time to be involved. The deadline for submissions is June 20, 2011.

The competition is open to both academia and industry. The competition will make use of two face image databases: one publicly-available database for training and one non-publicly-available database for testing. The training database (CASIA-FaceV5) contains 2,500 facial images of 500 subjects (i.e. 5 images/subject). The testing database contains 4,000 facial images of 1,000 subjects (i.e. 4 images/subject). These images were captured with varying illumination, pose, expression, accessories, and camera-to-subject distance. The face images in the training database are 640 x 480 pixel, 16-bit color bitmap (BMP) files. The goal is to write code that automatically recognizes the subject in each image of the testing database based on what was learned from the training database. Most years, the top three teams in the competition have received an award. Last year we awarded several publications, several Nokia N900 mobile computers, and several travel grants to the top three teams. The MLSP Competition Committee expects that we will be able to give out awards again this year.

Job Marketplace:

We have added a marketplace on the MLSP portion of the IEEE Signal Processing Society website. Be sure to check here for jobs related to machine learning and to post any jobs that you may know about.

New Members and Officers:

Starting January of this year, Tulay Adali is the new chairperson for the MLSP TC and Paris Smaragdis is the new vice-chair. We would like to thank the previous chairperson, Konstantinos I. Diamantaras, for all of his hard work. The MLSP TC would also like to welcome the following people, who have either just joined the MLSP TC or have recently had their tenure extended: Taylan Cemgil, Vince Calhoun, Francesco Palmieri, Raviv Raich, Andrew Singer, Gerard Dreyfus.

Awards and Recognitions:

Several members of the MLSP TC have recently received awards or recognitions. We wish to congratulate them on their impressive achievements.

Asoke Kumar Nandi (The University of Liverpool, UK) was elevated to IEEE Fellow status for contributions to signal processing and its applications. Asoke was also one of seven researchers recently funded by the Tekes program, which is part of the Finland Distinguished Professor Programme (FiDiPro), to perform an international research project. His project is entitled "Machine Learning for Future Music and Learning Technologies."

A recent IEEE Spectrum article, Teaching Machines About Madness, mentions the research of two of our members, Vince D. Calhoun (University of New Mexico) and Tülay Adali (University of Maryland, Baltimore County). This article discusses the development of a machine-learning system that can distinguish between people with schizophrenia and those with bipolar disorder using multiple functional magnetic resonance imaging brain scans. These conditions can be difficult for doctors to distinguish and require different treatments.

Tulay Adali (University of Maryland, Baltimore County) authored a paper with H. Li, M. Novey and J.-F. Cardoso, which received the IEEE Signal Processing Society Best Paper Award. The paper is entitled, "Complex ICA using nonlinear functions," and can be found in the September 2008 issue of IEEE Transactions on. Signal Processing. The award will be presented at ICASSP 2011 in Prague.

Weifeng (Aaron) Liu (Amazon) received the 2011 International Neural Network Society (INNS) Young Investigator Award, which will be awarded at the IJCNN 2011 conference in San Jose.

Gustavo Camps-Valls (University of València) recently received the IEEE Geoscience and Remote Sensing Society 2011 Prize Paper Award for a paper coauthored by J. Mooij and B. Schölkopf, which is entitled, "Remote Sensing Feature Selection by Kernel Dependence Measures."

We also have several members (quite possibly more than the two we list below) that are involved with the important task of reviewing grants for NIH.

Vince Calhoun (University of New Mexico) is a charter member for the BMIT study section. In addition, Yue Wang (Virginia Tech) has recently been invited and has accepted to serve as a member of the Biodata Management and Analysis Study Section, Center for Scientific Review.