Mahood Sadeghi; Masoud Yavarmanesh; Mostafa Shahidi Noghabi
Abstract
Nowadays, it has demonstrated that viruses can be transmitted by water and foods. Therefore, it causes the research to develop for detecting different viruses in water and foods. Among foods, milk can transfer potentially pathogenic viruses. On the other hand, to achieve every method for recovery and ...
Read More
Nowadays, it has demonstrated that viruses can be transmitted by water and foods. Therefore, it causes the research to develop for detecting different viruses in water and foods. Among foods, milk can transfer potentially pathogenic viruses. On the other hand, to achieve every method for recovery and extraction of viruses in raw milk it needs to know about impact of milk components on viruses. Artificial neural network (ANN) and Adaptive Nero Fuzzy Inference System (ANFIS) can help to estimate recovery efficiency of viruses in raw milk. The objective of this study was to evaluate the application of ANN and Adaptive Nero Fuzzy Inference System (ANFIS) to predict the impact of milk components on recovery and extraction of viral RNA in raw milk. Therefore, to run the model the amount of milk components (casein, whey protein, fat and lactose) and viral RNA extraction were as the input and the output of the network respectively. Also, to evaluate the efficiency of the network for the prediction, variables such as training, validating and test subsets as well as the hidden layers, transfer functions, learning rules and the hidden neurons were used. Based on the results, the best models in ANN were linear sigmoid transfer function, levenberg learning rule (r: 0.919) and linear sigmoid transfer function, levenberg learning rule (r: 0.956) for spiked model solution and (spiked – non spiked) model solution respectively and in Adaptive Nero Fuzzy Inference System (ANFIS) the best model were membership function Gaussian , Adaptive Nero Fuzzy Inference System (ANFIS) model TSK, linear tanh axon transfer functions and momentum learning rule (r: 0.879) and membership function Gaussian, Adaptive Nero Fuzzy Inference System (ANFIS) model TSK, linear axon transfer function, and step learning rule (r: 0.889) for spiked model solution and (spiked – non spiked) model solution respectively.
