Abstract
A computational paradigm is presented for making rapid and accurate estimations of physical properties for molecular systems. The method uses a set of descriptors as a numerical representation of the structure for a given system and relates these to a set of properties by utilizing a computational neural network. The neural network is capable of efficiently formulating all of the correlations necessary to make accurate predictions. Results have been obtained for up to 10 properties of 357 different polymers with an average prediction error of < 1%. The basic methodology has also proven efficient for making accurate predictions for properties of energetic materials, isomeric forms of saturated hydrocarbons, and fluoroorganic compounds.
Original language | English |
---|---|
Pages | 863-868 |
Number of pages | 6 |
State | Published - 1994 |
Event | Proceedings of the Artificial Neural Networks in Engineering Conference (ANNIE'94) - St. Louis, MO, USA Duration: Nov 13 1994 → Nov 16 1994 |
Conference
Conference | Proceedings of the Artificial Neural Networks in Engineering Conference (ANNIE'94) |
---|---|
City | St. Louis, MO, USA |
Period | 11/13/94 → 11/16/94 |