Intelligent Artificial Neural Network Computing Techniques for Shelf Life Determination of Processed Cheese

In this study feedforward and competitive artificial neural network models were developed for predicting shelf life of processed cheese stored at 30 C. Processed cheese is a food product generally made from Cheddar cheese. Processed cheese has several advantages over unprocessed cheese, such as extended shelf-life, resistance to separation when cooked, and uniformity of product. Input parameters consisted of texture, aroma and flavour, moisture, free fatty acids, and sensory score was taken as output parameter. Backpropagation algorithm based on Bayesian regularization mechanism was selected for training the network. Neurons in each hidden layers varied from 1 to 50. The network was trained with 100 epochs with single as well as double hidden layers, and transfer function for hidden layer was tangent sigmoid while for the output layer, it was pure linear function. MSE, RMSE, R and E were used in order to compare the prediction potential of the developed ANN models. FANN with single hidden layer having twenty neurons exhibited the best results, hence regression equations based on these results were developed for predicting shelf life of the product, which came out as 28.32 days, which is comparable to experimental shelf life of 30 days.