Optimization of the ANNs Predictive Capability Using the Taguchi Approach: A Case Study

• Published in: Mathematics (Basel) 2021-04, Vol.9 (7), p.766
• Main Authors: Manni, Andrea, Saviano, Giovanna, Bonelli, Maria Grazia
• Format: Article
• Language: English
• Subjects: Artificial intelligence; artificial neural network; Artificial neural networks; Bias; Design of Experiment (DoE); Design of experiments; environmental pollution; Food science; forecasting; Function analysis; Genetic algorithms; Heavy metals; Inorganic contaminants; Learning; Network design; Neural networks; Neurons; Optimization; Optimization algorithms; Orthogonal arrays; Parameter identification; parametric design; Parametric statistics; Pollutants; Pollution sources; Reliability analysis; Taguchi methods; Topology; Variables
• ISSN: 2227-7390; 2227-7390
• DOI: 10.3390/math9070766

Artificial neural networks (ANNs) are a valid alternative predictive method to the traditional statistical techniques currently used in many research fields where a massive amount of data is challenging to manage. In environmental analysis, ANNs can analyze pollution sources in large areas, estimating difficult and expensive to detect contaminants from other easily measurable pollutants, especially for screening procedures. In this study, organic micropollutants have been predicted from heavy metals concentration using ANNs. Sampling was performed in an agricultural field where organic and inorganic contaminants concentrations are beyond the legal limits. A critical problem of a neural network design is to select its parametric topology, which can prejudice the reliability of the model. Therefore, it is very important to assess the performance of ANNs when applying different types of parameters of the net. In this work, based on Taguchi L12 orthogonal array, turning experiments were conducted to identify the best parametric set of an ANNs design, considering different combinations of sample number, scaling, training rate, activation functions, number of hidden layers, and epochs. The composite desirability value for the multi-response variables has been obtained through the desirability function analysis (DFA). The parameters’ optimum levels have been identified using this methodology.