1. Title and Author
Title: 2011_01_08_Stabilizing the PARAFAC decomposition of fluorescence spectra by insertion of zeros ouside the data area
Authors: Lisbeth Garbecht Thygesena, ?smund Rinan*, Søren Barsbergb, Jens K.S. Møllerc
aRoyal Veterinary and Agricultural University, Department of Dairy and Food Science, Food Technology, Rolighedsvej 30, DK-1958 Frederiksberg, Denmark
bRoyal Veterinary and Agricultural University, Plant Fiber Laboratory, Agrovej 10, DK-2630 Taastrup, Denmark
cRoyal Veterinary and Agricultural University, Department of Dairy and Food Science, Food Chemistry, Rolighedsvej 30, DK-1958 Frederiksberg, Denmark
2. Summary of Paper
There has been an enormous increase in extracting information from analysis of organic products by fluorescence spectroscopy and PARAFAC modeling. The data obtained from analysis of organic products using fluorescence spectroscopy includes light scattering effects. There are different techniques to deal with these light scattering effects so that PARAFAC decomposes the signals without any difficulty. The best technique in some practical situations is the insertion of zeros in the region of light scattering.
Higher energy cannot be emitted by a molecule than its absorbed energy in the form of light. So, there is rarely any information either physical or chemical outside the data area. PARAFAC decomposes the signals using the least square technique.
There are four different ways of data pretreatment which are discussed in this paper and are given below:
1. Only missing values ? conventional method
2. Zeros below first order and missing values above second order ? mixed method
3. Only zeros ? all zeros method
4. Mostly zeros, but with a ribbon of missing values around the Rayleigh scatter lines- ribbon method
Out of these four methods, third method is best as it helps PARAFAC to converge faster and to give better estimates. Moreover, this technique give physical and chemical meanings of the model.
3. Contribution to ESEL
This paper deals with PARAFAC modeling by inserting zeros outside the data area to make the PARAFAC model a successful model by giving physical and chemical meanings of the results. But, at the same time, it is also said that insertion of zeros is useful in some cases. So, PARAFAC models must be checked with all the four techniques discussed above because any of the four techniques can give best results.
By: Aamir Alaud-din
1. Title and Author