Automated selection of regression models using neural networks for carbon-13 NMR spectral predictions

Development of artificial neural networks for spectral interference correction in optical emission spectrometry

“…Artificial neural networks have been successfully applied to fields as diverse as calibration 17 , nonlinear system identification 18,19 , classification 20 , process control 21 , interpretation of IR-spectra 20, 21 and UV-spectra 22 , atomic emission spectrometry 23 , atomic absorption spectrometry 24 , nuclear magnetic resonance (NMR) [25][26][27] and ion mobility spectrometry (IMS) 28 .…”

Artificial Neural Networks and Their use in Chemistry

“…102 Nuclear magnetic resonance (NMR) has also been a fruitful area of ANN BP application. [103][104][105][106][107][108][109][110][111] Most studies have dealt with either the simulation of 13C spectra or the prediction of 1% shifts,103-105,107-109 although one study focused on prediction of phosphorus shifts.106 A BP network was used to predict secondary protein structure, which was then used to assist in NMR assignment. 110 In another study, 1H NMR spectra of binary mixtures of alditols were successfully classified.111 BP networks were used for multivariate calibrations of pyrolysis mass spectra.112,113 In one case the BP networks gave better concentration predictions than did both partial least-squares and principal components regression methods.112 Another study with similar results also noted that linear transfer functions gave better results than sigmoid functions.113 Structural features were successfully identified in a library search of mass spectra.114 Interlaboratory calibration of two mass spectrometers has been accomplished.~l5…”

“…That is, do you have several variables Practical Considerations in Solvina Problems with Neural Networks103 …”

Rapid optimization and minimal complexity in computational neural network multivariate calibration of chlorinated hydrocarbons using Raman spectroscopy

“…Unfortunately, analytical chemists have only infrequently made use of more advanced optimization methods for NN modeling. Some successful examples include both the research groups of Gemperline [37,38] and Jurs [39], who began using BP and switched to the quasiNewton Broyden-Fletcher-Goldfarb-Shanno method (BFGS), which was found to be considerably faster and, interestingly, gave better calibration fits and predictions [40][41][42][43]; Li et al [44] and Bos et al [45] have used conjugate gradient (CG) algorithms; Tetteh et al [46] utilized both radial basis and Levenberg-Marquardt (LM) algorithms in a QSAR study. Our own work has addressed the suitability of LM, CG and simulated annealing algorithms for computational neural network training (W.J.…”