Host: Division of Chemical Information and Computer Science, The Chemical Society of Japan
Co-host: The Pharmaceutical Society of Japan, Japan Society for Bioscience, Biotechnology, and Agrochemistry, The Japan Society for Analytical Chemistry, Japan Chemistry Program Exchange, Japanese Society for Information and Systems in Education (Approaval)
Pages JK03
We developed a neural network simulator for structure-activity correlation of molecules: Neco. A self-organized network model for high-speed learning was included in Neco, a perceptron type with the three layers. In the hidden layer the neurons is self-organized by using Mahalanobis generalized distance. Using this simulator, hydrophobic parameter, logP, of perillartine derivatives was predicted. We used for inputs the set of six parameters: five STERIMOL (L, Wl, Wu, Wr, and Wd) parameters and the sweet/bitter activity. The 22 sampled data is used for training. Our neural network accurately can predicted hydrophobic parameter, logP. Compared with a normal perceptron network, the learning ability of our network is quite higher and its convergence speed is greatly larger than it.