Jm. Sutter et al., AUTOMATED DESCRIPTOR SELECTION FOR QUANTITATIVE STRUCTURE-ACTIVITY-RELATIONSHIPS USING GENERALIZED SIMULATED ANNEALING, Journal of chemical information and computer sciences, 35(1), 1995, pp. 77-84
Citations number
36
Categorie Soggetti
Information Science & Library Science","Computer Application, Chemistry & Engineering","Computer Science Interdisciplinary Applications",Chemistry,"Computer Science Information Systems
The central steps in developing QSARs are generation and selection of
molecular structure descriptors and development of the model. Recently
, computational neural networks have been employed as nonlinear models
for QSARs. Neural networks can be trained efficiently with a quasi-Ne
wton method, but the results are dependent on the descriptors used and
: the initial parameters of the network. Thus, two potential opportuni
ties for optimization arise. The first optimization problem is the sel
ection of the descriptors for use by the neural network. In this study
, generalized simulated annealing (GSA) is employed to select an optim
al set of descriptors. The cost function used to evaluate the: effecti
veness of the descriptors is based on the performance of the neural ne
twork. The second optimization problem is selecting the starting weigh
ts and biases for the network. GSA is also used for this optimization.
The result is an automated descriptor selection algorithm that is an
optimization inside of an optimization. Application of the method to a
QSAR problem shows that effective descriptor subsets are found, and t
hey support models that are as good or better than those obtained usin
g traditional linear regression methods.