Assessment of applicability domain for multivariate counter-propagation artificial neural network predictive models by minimum Euclidean distance space analysis: A case study

• Published in: Analytica chimica acta Vol. 759; pp. 28 - 42
• Main Authors: Minovski, Nikola, Župerl, Špela, Drgan, Viktor, Novič, Marjana
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 08.01.2013; Elsevier
• Subjects: Analytical chemistry; Applicability domain; Artificial neural networks; Chemistry; Counter-propagation artificial neural network; Euclidean distance; Exact sciences and technology; Leverage approach; Quantitative structure–activity relationship; MEDS; AD; Quantitative structure–activity relationship; Leverage approach; QSAR; KANN; CP ANN; Artificial neural networks; Euclidean distance; Counter-propagation artificial neural network; Applicability domain; ED; Performance evaluation; Quantitative structure-activity relationship; Propagation; Chemical compound; Estimation; Prediction; Representation; Neural network; Multivariate analysis; Protein; Case study; Outlier; Structure activity relation; Quality; Test method; Reliability; Vector

[Display omitted] ► The concept of applicability domain (AD) in QSAR modeling is discussed. ► The AD assessment method for nonlinear neural network predictive models is proposed. ► The counter-propagation artificial neural network (CP-ANN) was applied for modeling. ► Minimal Euclidean distance space...