A neural network which utilized data from the infrared spectra, carbon-13 NMR spectra, and molecular formulas of organic compounds was developed. The network, which had one layer of hidden units, was trained by backpropagation; network parameters were determined by a simplex optimization procedure. A database of 1560 compounds was used for training and testing. The trained network was able to identify with high accuracy the presence of a broad range of substructural features present in the compounds. The number of features identified and the accuracy were significantly greater as compared with networks using data from a single form of spectroscopy. The results have significance for the SESAMI computer-enhanced structure elucidation system.

A new method of structure generation called convergent structure generation has been developed to address limitations of earlier methods. The features of the program (HOUDINI) based on this method include the following: a single integrated representation of the collective substructural information; the use of parallel atom groups for efficient processing of families of alternative substructural inferences; and a managed structure generation procedure designed to build required structural features early in the process.

A neural network model having a layer of hidden units is described which can identify functional groups in organic compounds, based on their infrared spectra. This network shows substantially better performance than the simple linear model reported earlier. The effect of the training set size and composition, the number of hidden units used, and the training time were studied.

No abstract

No abstract