Showing papers in "Chemometrics and Intelligent Laboratory Systems in 1993"

TL;DR: De Jong et al. as discussed by the authors proposed SIMPLS, an alternative approach to PLS regression, which calculates the PLS factors directly as linear combinations of the original variables, while obeying certain orthogonality and normalization restrictions.

TL;DR: It is shown that each mode of principal component analysis or ‘factor analysis’ is equivalent to solving a certain least squares problem where certain error estimators σ ij are assumed for the measured data matrix X ij and the best posssible scaling and a near-optimal scaling are introduced.

TL;DR: This tutorial consists of two parts which treat a variety of key issues concerning genetic algorithms, and elaborates on practical issues such as representation, configuration and hybridization with other techniques.

TL;DR: Wythoff et al. as discussed by the authors provided a short history of neural network research, and a review of chemical applications with a clear and detailed introduction to the theory behind backpropagation neural networks along with a discussion of practical issues facing developers.

TL;DR: The ability to search a large parameter space with no initial guesses per se, no derivatives of the objective function and to cope with local minima, make it a candidate method for several areas of chemistry.

TL;DR: In this article, a self-modeling curve resolution of individual spectroscopic titrations of a multiequilibria system based on evolving factor analysis does not always provide the right qualitative and quantitative solution, even when the constraints of non-negativity, unimodality and closure are applied.

TL;DR: In this article, the four-parameter logistic model is used to estimate the response-error relationship (RER) and the variance function is estimated via generalized least squares/variance function estimation (GLS/VFE).

TL;DR: In this article, Liang and Manne presented a classification of mixture problems and methods for their quantitative analysis, and discussed the advantages and limitations of available multivariate calibration and resolution methods with respect to the proposed classification of analytical mixture problems.

TL;DR: Toft et al. as discussed by the authors used eigenstructure tracking analysis for revealing noise pattern and local rank in instrumental profiles, which was applied to transmittance and absorbance IR spectroscopy.

TL;DR: In this paper, a "component stripping" technique is presented for unique resolution of multicomponent systems (with more than three overlapping components) by means of the heuristic evolving latent projections method.

TL;DR: It is concluded that the analysis of Kohonen maps yields valuable information which may be used for the practical design of a modular tree-like system of dedicated multi-layer feed-forward neural networks for the automated interpretation of infrared spectra.

TL;DR: In this study the interpretation of IR spectra using artificial neural networks is addressed by tackling specific sub-problems with small, dedicated neural networks intended to form the modules of a larger, structured system for spectrum interpretation.

TL;DR: Liang et al. as discussed by the authors proposed a two-way procedure for background correction of chromatographic/spectroscopic data by congruence analysis and least-squares fit of the zero-component regions.

TL;DR: In this article, a self-modeling chemometric method for obtaining the concentration profiles of individual components from evolutionary processes is applied to unresolved liquid chromatograms, making use of the fact that each component lies in a specific region along the evolutionary axis, called the window.

TL;DR: Validation and evaluation of neural networks in analytical chemistry should be very rigorous, as this is one way to create confidence in the new technique.

TL;DR: In this paper, Proton and carbon-13 NMR spectroscopic methods were applied in the differentiation of 53 German white wines from the regions Rheinhessen, Rheingau and Mosel-Saar-Ruwer.

TL;DR: In this article, the authors describe a hybrid GA in which a steepest descent, pseudo-Newton procedure is iterated with an incest-preventing GA, each providing a starting point for the other.

TL;DR: Hopke et al. as discussed by the authors combined chemical and meteorological data to infer source areas of airborne pollutants using potential source contribution function (PSCF) to find specifically where airborne contaminant emission sources can be found.

TL;DR: The criteria and procedures for the addition of new data sets to the collection are presented along with the mechanism for obtaining such data and restrictions on their use.

TL;DR: Machado et al. as discussed by the authors performed principal component analysis and evolving factor analysis on marine and soil fulvic acid spectra to identify the variations observed in synchronous fluorescence spectra of fulvic acids.

TL;DR: It is found that following an eigenanalysis resolution with PARAFAC frequently leads to significant improvement in the quality of the resolution, and this paper proposes synthesizing the two methods by using the resolution generated by eigen analysis as starting values for the iterative PARAFac algorithm.

TL;DR: A refinement of the quantitative structure-activity relationship presented in Part 1 of this series using the alternating conditional expectations (ACE) technique to account for nonlinearity is described in this paper.

TL;DR: This paper describes the performance of a GA that finds hypothetical physical structures of poly(ethylene terephthalate) yarns corresponding to a certain combination of mechanical and shrinkage properties and uses a validated ANN that has been trained for the complex relation between structure and properties of PET.

TL;DR: In this paper, neural networks were used to map the correlations between the IR spectral properties and the structure descriptors, and the correlations were then used to predict the spectral properties from the chemical structure.

TL;DR: Song et al. as discussed by the authors used three-layer artificial neural network models with back-propagation of error to investigate the quantitative structure-activity relationship of dihydropteridine reductase inhibitors.

TL;DR: CMAC was found to provide rapid, accurate deconvolutions for a wide range of peak-height ratios, peak widths, and resolutions and has significant advantages in ease of training and in detection of inadequate training that do not apply for a back-propagation neural network.

TL;DR: Hibbert as discussed by the authors described a genetic algorithm (GA) for generating isomeric structures of a molecule given a molecular formula and information about allowed bonding of atoms, where bonds are represented by tuples of the bonded atoms.

TL;DR: Brakstad et al. as mentioned in this paper proposed a method based on partial least squares regression (PLSR) on normalised spectral intensities as independent variables, and the carbon-carbon double bond position as response.

TL;DR: It is found that suitable designs do exist, and that the design depends on the system under research, while the proposed designs are compared with known designs.