%0 Journal Article %T Mixture of Xylose and Glucose Affects Xylitol Production by Pichia guilliermondii: Model Prediction Using Artificial Neural Network %J Iranian Journal of Chemistry and Chemical Engineering %I Iranian Institute of Research and Development in Chemical Industries (IRDCI)-ACECR %Z 1021-9986 %A Magharei, Azadeh %A Vahabzadeh, Farzaneh %A Sohrabi,, Morteza %A Rahimi Kashkouli, Yousef %A Maleki, Mohammad %D 2012 %\ 03/01/2012 %V 31 %N 1 %P 119-131 %! Mixture of Xylose and Glucose Affects Xylitol Production by Pichia guilliermondii: Model Prediction Using Artificial Neural Network %K Artificial neural network %K Glucose and xylose mixture %K Pichia guilliermondii %K Response surface methodology %K Xylitol production %R 10.30492/ijcce.2012.10186 %X Production of several yeast products occur in presence of mixtures of monosaccharides. To study effect of xylose and glucose mixtures with system aeration and nitrogen source as the other two operative variables on xylitol production by Pichia guilliermondii, the present work was defined. Artificial Neural Network (ANN) strategy was used to athematically show interplay between these three controllable factors and the xylitol productivity response. In the first stage, model fitting was performed using Response Surface Methodology (RSM) and the appropriate fraction of this design then was applied for the ANN training step (Levenberg Marquardt ‘LM’ algorithm). The best ANN model configuration with the three test input variables composed of six neurons in the hidden layer and tangent sigmoid (TANSIG) and linear transfer function (PURELIN) were used as the activation functions for the data processing from inputs to the hidden layer and from the constructed neurons to the output nodes. The network performance was evaluated by Mean Squared Error (MSE) and the regression coefficient of determination (R2). These values respectively, for the RSM model fitting were 2.327× 10-4 and 0.9817, and for the ANN training data were 2.29 × 10-8 and 0.9999. While MSE and R2 values for the other two steps of ANN were 4.56 × 10-3 and 0.9741 (validating step) and1.52× 10-3 and 0.9325 (testing step), respectively. Positive synergism of ANN with RSM was confirmed. %U https://ijcce.ac.ir/article_10186_657bdc4e9f6e5d68239025bdacbf3416.pdf