New multi-way tensor decompositions for data analysis of highly complex data
Faculty of Life Sciences, Dept. Food Science
University of Copenhagen
Monday, February 12, 2007
Tensor-based models are becoming increasingly used these years. These include higher-order SVD, PARAFAC and Tucker models. Through the
last 20 years those methods have been successfully employed in chemistry and related areas. This talk will give an introduction to the methods as well as
highlight some applications of them. Applications will include early cancer detection using fluorescence spectroscopy, of metabolic response of
rats to toxins, etc. Some of the urgent mathematical and statistical problems will also be touched upon.
Professor Rasmus Bro works with chemometric development and applications at Copenhagen University in Denmark. Although a majority of the work is
concerned with food-related research, the group also collaborates with other branches of applied chemistry such as pharmaceutical production. Professor
Bro obtained his Ph.D. in 1998 at the University of Amsterdam, The Netherlands and has worked with multivariate calibration, multi-way
analysis, exploratory analysis, experimental design, numerical analysis,blind source separation, and curve resolution. He has more than 70
peer-reviewed papers on chemometrics. Rasmus Bro has developed several new algorithms within multi-way analysis and numerically constrained regression.
He currently has six Ph.D. students and much of the scientific work is extensively documented on the Web in the form of presentations, algorithms
and courses which can be found at www.models.life.ku.dk. He is an editor of Journal of Chemometrics.
Hosted by: Bulent Yener (x6907)