Abstrato

Support vector regression to predict survival of Lactobacillus acidophilus in concentrated yoghurt

Mozhgan Nasiri Shahri, Ali Mohamadi Sani, Maryam Tavakoli Fadiheh

In this paper we present a function to predict the survival of Lactobacillus acidophilus (LA) in concentrated yoghurt. For this purpose we used Artificial Intelligence tools based on Support Vector Machines for Regression (SVR). Various parameters including: pH, percentage of prebiotic compounds (inulin and oligo-fructose) and inoculum dosage of probiotic bacteria which are effective factors on LA survival were considered. Performance of developed model was evaluated by calculating the mean square error (MSE). The results showed that the mean square error on days 1, 7, 14 and 21 were 1.04x10-5, 1.08x10-5, 9.56x10-6, 7.73x10- 6 respectively and defined model had the capacity of estimation accuracy for predicting survival of LA during storage in the refrigerator. Low values of MSE indicate that SVR is able to predict LA count in concentrated yoghurt.