RT Journal Article
ID 603bfc01337404da
A1 Ivakhnenko, Alexey G.
A1 Kovalishyn, V. V.
A1 Tetko, Igor V.
A1 Luik, A. I.
A1 Ivakhnenko, Grigoriy A.
A1 Ivakhnenko, N. A.
T1 Application of Self-Organizing Neural Networks with Active Neurons for Prediction of Bioactivity of Chemical Compounds by the Analogues Search Algorithm
JF Journal of Automation and Information Sciences
JO JAI(S)
YR 1999
FD 1999-10-01
VO 31
IS 7-9
SP 51
OP 58
K1 artificial neural networks
K1 self-organization
K1 criterion of error variation
K1 monosubstituted benzol
K1 prognosis accuracy
AB In the developed artificial neural network the analogues complexing algorithm is used as active neurons. Proposed procedure of structure self-organization and parameters choice is used for testing of two experimental subsamples of data. Two variants of sorting procedures of variables sets are considered. Analogues are found by distance calculation between the given output point and its closest neighbours. Received results show that self-organizing neural networks with active neurons, can be successfully applied in structure-activity relationship studies.
PB Begell House
LK https://www.dl.begellhouse.com/journals/2b6239406278e43e,69fb601f3b58afbf,603bfc01337404da.html