Soft computing in the design of nontoxic chemicals

This research focuses on the use of soft computing to aid in the development of novel, state-of-the-art, nontoxic dyes which are of commercial importance to the U.S. textile industry. Where appropriate, modern molecular orbital (MO) and density functional (DF) techniques are employed to establish the necessary databases of molecular properties to be used in conjunction with the neural network approach. In this research, we focused on the following: (1) using molecular modeling to establish databases of various molecular properties of azo dyes required as input for our neural network approach; (2) designing and implementing a neural network architecture suitable to process these databases; and (3) investigating combinations of molecular descriptors needed to predict various properties of the azo dyes.