Estimation of the computational cost and analysis of application of fuzzy descriptors in QSAR problems

This paper demonstrates the analysis of the application of fuzzy descriptors – a new data description model in the quantitative structure-activity relationship problem. The computational complexity estimation demonstrates that the use of fuzzy descriptors increases the number of operations, however the evolutionary method allows us to decrease significantly the computational complexity of the algorithm. The results of fuzzy description application to prediction of the general toxicity class of chemical compounds demonstrate that the new model is useful.