General Information Network Training Decoding of an RDF Code

Structure Prediction

A CPG network can be operated in reverse mode, enabling the prediction of a structure code from an infrared spectrum. The input of a query infrared spectrum into a trained CPG network provides a structure code vector, which represents the radial distribution function with discrete values of g(r). This RDF code has then to be decoded to provide the Cartesian coordinates of a 3D structure. Only the infrared spectrum is used to determine the winning neuron during training, whereas the adjustment of the weights for both the infrared spectrum and structure code is performed.