Image processing news

Recent days brought to us two interesting articles about image processing. First one describes wavelets as a tool to analyze spectral and spatial informations inside images [1]. The second one presents immortal topic - a new idea of 2D electrophoretic images alignment [2]. Enjoy the lecture!


  1. P. Juneau, A. Garnier, and C. Duchesne, "The undecimated wavelet transform–multivariate image analysis (UWT-MIA) for simultaneous extraction of spectral and spatial information", Chemometrics and Intelligent Laboratory Systems, 2014.
  2. S. Nhek, E. Mosleth, M. Høy, M. Griessl, B. Tessema, U. Indahl, and H. Martens, "Nonlinear visualisation and pixel-based alignment of 2D electrophoresis images", Chemometrics and Intelligent Laboratory Systems, vol. 118, pp. 97-108, 2012.

Fuzzy and crisp data (with robust clustering...)

There are some robust methods of clustering: possibilistic, noise, metric, trimmed, semifuzzy, etc... How to transfer them to fuzzy data? A new Chemolab article [1] gives the discussion and some interesting proposal. I recommend this cited reference [2] for future reading.


  1. P. D'Urso, and L. De Giovanni, "Robust clustering of imprecise data", Chemometrics and Intelligent Laboratory Systems, vol. 136, pp. 58-80, 2014.
  2. R. Dave, and R. Krishnapuram, "Robust clustering methods: a unified view", IEEE Transactions on Fuzzy Systems, vol. 5, pp. 270-293, 1997.

New articles about variable selection

Interested in variable/feature selection? Last weeks brought to us three interesting papers: a study on PLS with autororrelated data [1], general multinomial logit models [2] and subsampling approach [3].


  1. N. Afanador, T. Tran, L. Blanchet, and L. Buydens, "Variable Importance in PLS in the Presence of Autocorrelated Data - Case studies in Manufacturing Processes", Chemometrics and Intelligent Laboratory Systems, 2014.
  2. G. Tutz, W. Pößnecker, and L. Uhlmann, "Variable selection in general multinomial logit models", Computational Statistics & Data Analysis, vol. 82, pp. 207-222, 2015.
  3. Z. Lin, X. Pan, B. Xu, J. Zhang, X. Shi, and Y. Qiao, "Evaluating the reliability of spectral variables selected by subsampling methods", Journal of Chemometrics, pp. n/a-n/a, 2014.

OCPLS in Matlab

One-class partial least squares (OCPLS) classifiers were implemented in new Matlab Toolbox [1]. What are one class classifiers? The answer could be find in this often cited article [2].


  1. L. Xu, M. Goodarzi, W. Shi, C. Cai, and J. Jiang, "A MATLAB toolbox for class modeling using one-class partial least squares (OCPLS) classifiers", Chemometrics and Intelligent Laboratory Systems, 2014.
  2. R.G. Brereton, "One-class classifiers", Journal of Chemometrics, vol. 25, pp. 225-246, 2011.

Baseline in chromatographic signals

The baseline problem is still important in processing of chromatograms. Many methods are proposed for baseline filtering [1]. In recent Chemolab an interesting algorithm for baseline and denoising, called BEADS [2] is proposed.


  1. . Komsta, "Comparison of Several Methods of Chromatographic Baseline Removal with a New Approach Based on Quantile Regression", Chromatographia, vol. 73, pp. 721-731, 2011.
  2. X. Ning, I.W. Selesnick, and L. Duval, "Chromatogram baseline estimation and denoising using sparsity (BEADS)", Chemometrics and Intelligent Laboratory Systems, 2014.

Biobjective sparse PCA

Sparse PCA is an idea to explain maximum variance with as few of original variables, as possible, increasing the interpretability. Spanish team presents in new JMVA [1] new heuristic approach of maximization of variance explained and sparseness simultaneously. If the reader knows PCA, but has no knowledge on sparse PCA, this article [2] is a "must read".


  1. E. Carrizosa, and V. Guerrero, "Biobjective sparse principal component analysis", Journal of Multivariate Analysis, vol. 132, pp. 151-159, 2014.
  2. H. Zou, T. Hastie, and R. Tibshirani, "Sparse Principal Component Analysis", Journal of Computational and Graphical Statistics, vol. 15, pp. 265-286, 2006.

Chemical rank of trilinear data

Trilinear data (such as fluorescence spectra) are very interesting subject of chemometric analysis. New paper by Li et al. [1] contains an interesting proposal for rank estimation of such data, called vector subspace projection with Monte Carlo simulation (VSPMCS). Beginners in tri- and multilinearity can start with a great review of A. Olivieri [2].


  1. Y. Li, H. Wu, X. Zhang, Y. Chen, H. Gu, Q. Zuo, Y. Zhang, S. Guo, X. Liu, and R. Yu, "Estimating the chemical rank of three-way fluorescence data by vector subspace projection with Monte Carlo simulation", Chemometrics and Intelligent Laboratory Systems, vol. 136, pp. 15-23, 2014.
  2. A.C. Olivieri, "Analytical Advantages of Multivariate Data Processing. One, Two, Three, Infinity?", Anal. Chem., vol. 80, pp. 5713-5720, 2008.

Chromatographic signals alignment

In a recent article in JSS [1], French authors discuss theoretically chromatographic alighment methods: COW, PTW, PARS, PAGA, PABS, Defferential Evolution, PAFFT, RAFFT, SPTW and compare their performance. This interesting article can be a good starting point for the reader interested in warping. And this chapter [2] can be a good companion.


  1. R. Korifi, Y.L. Dréau, and N. Dupuy, "Comparative study of the alignment method on experimental and simulated chromatographic data", Journal of Separation Science, pp. n/a-n/a, 2014.
  2. R. Jellema, "Variable Shift and Alignment", Comprehensive Chemometrics, pp. 85-108, 2009.