Predication of Some Physiochemical Properties of Low Calorie Cake Containing Apple Fiber Using Artificial Neural Networks

In this research, in order to modelling and predicating some physiochemical properties of low calorie cake containing apple pulp fiber from different percentages, apple pulp fiber (0- 10%)and maintenance time (0-30days) were used and fat degree, water activity, dropping off, the number of mold & yeast, resinous, chewable, elasticity and brightness of cake samples were surveyed. For predication of changing processes from artificial neural networks, MATlab R2013a software was used .The survey showed that different networks, releasing feedback networks with typology of 8-5-3 with correlation coefficient more than 912/0 and the average square less than 0115/0 and applying sigmound activation hyperlogic ,mutant learning patterns and1000 learning cycle considered as the best neural modelling.The result showed that the selected optimal models also were surveyed and these models with high correlation coefficient(more than 689/0)were able to predict the changing processes .Among measurable properties, artificial neural networks were so accurate for predicting the level of fat.