Abstract
A neural network approach is investigated in this study for the evaluation of the quality of typical snack products. Although quality of food appears mostly subjective, some external features of the product are indicators of snack quality. External texture features together with the size and shape parameters of the snacks are used as input to a fully connected backpropagation network. Sensory data, which are indicators of product quality, are employed as the target or output of the network. The capability of the network in evaluating the quality of snack products is studied by employing different topology of backpropagation network. The analysis is validated through a comparison of the predicted sensory data from the network and those available from the taste panel. Methods of image processing for proper texture and shape and size analysis are presented. Different facets of backpropagation network are also investigated for better predictions of snack quality.
Sayeed, Mohammad Shaheen (1994). A neural network approach to snack quality evaluation. Master's thesis, Texas A&M University. Available electronically from
https : / /hdl .handle .net /1969 .1 /ETD -TAMU -1994 -THESIS -S2744.