Screening GC-MS data for carbamate pesticides with temperature-constrained-cascade correlation neural networks

Aromatic carbamate pesticides are important agrochemicals. Mass spectral cl assification models were built for carbamates and their substructures using temperature-constrained-cascade correlation networks (TC-CCNs). The carbam ate classifier was applied to the mass spectral scans of a GC-MS run. The c lassification models were built from reference and experimental mass spectr a. Different network configurations were compared that used multiple networ k models with single outputs and single networks with multiple outputs. A m ajor source of variation caused by randomly partitioning the training and p rediction sets was reduced by an order of magnitude by using a method of La tin-partitions. This method also furnished a precision measure for comparin g classification methods. Multiple networks with single outputs generally p redicted better than single networks with multiple outputs. Hierarchical si ngle output networks achieved better than 98% classification accuracy in on e study. The TC-CCN models compared favorably to the K-nearest neighbors (K NN) and discriminant partial least squares (DPLS) reference methods. (C) 20 00 Elsevier Science B.V. All rights reserved.