Neural Network Recognition of Chemical Class Information in Mobility Spectra Obtained at High Temperatures

Abstract

A minimal neural network was applied to a large library of high-temperature mobility spectra drawn from 16 chemical classes including 154 substances with 2000 spectra at various concentrations. A genetic algorithm was used to create a representative subset of points from the mobility spectrum as input to a cascade-type back-propagation network. This network demonstrated that significant information specific to chemical class was located in the spectral region near the reactant ions. This network failed to generalize the solution to unfamiliar compounds necessitating the use of complete spectra in network processing. An extended back-propagation network classified unfamiliar chemicals by functional group with a mean for average values of 0.83 without sulfides and 0.79 with sulfides. Further experiments confirmed that chemical class information was resident in the spectral region near the reactant ions. Deconvolution of spectra demonstrated the presence of ions, merged with the reactant ion peaks that originated from introduced samples. The ability of the neural network to generalize the solution to unfamiliar compounds suggests that these ions are distinct and class specific.

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 .…”

Discriminant Analysis of Fused Positive and Negative Ion Mobility Spectra Using Multivariate Self-Modeling Mixture Analysis and Neural Networks

“…Artificial neural networks (ANNs) are popular and successful methods for computer-based classification and quantitative analysis for IMS data. [7][8][9] Among all the available ANN algorithms, TCCCN is interesting in that it combines the advantages of cascade correlation and computational temperature constraints. 16 This combination makes it a nonlinear calibration method that is easier to use, stable, and faster than back-propagation networks.…”

“…Other pattern recognition algorithms, such as K-nearest neighbors (KNN), partial least squares (PLS), and artificial neural networks (ANNs), [7][8][9][10] have been applied to identify or classify chemicals. Instead of using the information in the drift time windows, pattern recognition algorithms use the entire IMS spectrum.…”

“…11 Unfortunately, significant information is located in the spectral region near the RIP. 8 The exclusion of the RIP spectral region would result in the loss of useful information for classification. A logarithmic transformation was used to attenuate the larger intensities of RIP and improve the performance of neural networks.…”

“…Previous studies have demonstrated the suitability of neural networks for IMS data analysis. [7][8][9] The temperature-constrained cascade-correlation neural network (TCCCN) was selected in this project because it trains rapidly, automatically configures its own architecture during training, and yields reproducible models. 15,16 Cascade-correlation neural networks have been used for screening gas chromatography-mass spectrometry (GC-MS) data, 10 quantitative and qualitative analysis of IMS measurements, 9 and bacteria identification.…”