Discrimination of Yellowfin Tuna Origin Based on VIS / NIR Spectroscopy Combined with DBN-BP Neural Networks Algorithm
Yellowfin tuna populations are mainly distributed in the three distinct regions, namely the Indian Ocean, Western Pacific Ocean, and the Atlantic Ocean regions. To determine the origination of yellowfin tuna from the three different zones is considered to be a challenging issue. In this report, tuna fish obtained from three different regions had different characteristic peaks on visible/near infrared (VIS/NIR) spectra. So the VIS/NIR spectroscopy was used as a quick and non-destructive method to identify the origin of yellowfin tuna.In order to precisely identify the tuna spectra acquired from the different areas, the neural network model of a deep belief network with back propagation (DBN-BP) was introduced to model the spectral variabilities, and its performance was evaluated by comparing with other efficient identification algorithms such as Back Propagation (BP) network, Support Vector Machine (SVM), and Partial Least Squares Discriminant Analysis (PLS-DA).The research results showed that the DBN-BP algorithm had the highest discrimination accuracy and the least prediction error compared to other algorithms. The combination of VIS/NIR spectroscopy and DBN-BP algorithm outperformed in identifying the origin of yellowfin tuna, with the prediction accuracy of 100% in calibration sets, 98% in cross-validation and 97.5% in external sets. The RMSE of calibration sets and external sets could reach 0.01 and 0.19, respectively. The established method will shed a new light on the traceability of the origins of aquatic products.