Premium
Evaluation of Equilibria with Use of Artificial Neural Networks (ANN). II. ANN and Experimental Design as a Tool in Electrochemical Data Evaluation for Fully Dynamic (Labile) Metal Complexes
Author(s) -
Cukrowski Ignacy,
Farková Marta,
Havel Josef
Publication year - 2001
Publication title -
electroanalysis
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.574
H-Index - 128
eISSN - 1521-4109
pISSN - 1040-0397
DOI - 10.1002/1521-4109(200103)13:4<295::aid-elan295>3.0.co;2-r
Subject(s) - artificial neural network , stability (learning theory) , range (aeronautics) , standard deviation , approximation error , computer science , mean squared error , experimental data , mathematics , algorithm , biological system , statistics , artificial intelligence , materials science , machine learning , composite material , biology
A use of artificial neural networks (ANN) and various experimental designs (ED) for refinement of experimental data obtained in a polarographic metal‐ligand equilibrium study of fully dynamic (labile) metal complexes was thoroughly examined. ANN were tested on evenly and randomly distributed experimental error‐free and error‐corrupted data. It was found that randomly distributed experimental data did not influence the prediction power of ANN. Numerous tests demonstrated that ANN with appropriate ED can provide accurate prediction in the stability constants with the absolute errors in the range of ±0.05 log unit or smaller. ANNs were found exceptionally robust. Random experimental errors have not influenced estimates in stability constants much even when errors in pH up to the value of ±0.1 pH unit were introduced. A special procedure has been worked out that allows to minimize the influence of error‐corrupted data even further; no significant difference was observed between results obtained on error‐free and error‐corrupted data. This procedure makes it also possible to obtain a standard deviation in the calculated stability constants that is usually a difficult task when ANNs are used. The results obtained from ANN were compared with those obtained from a hard model based nonlinear regression techniques. No significant difference in evaluated data from these two, soft and hard model based approaches, was found. The use of ANN described here for polarographic data is of general nature and, in principal, can be adopted to other analytical techniques commonly used in metal‐ligand equilibrium studies.