One-Stop Source
Of Soybean Meal Information

Artificial neural network prediction of amino acid levels in feed ingredients.

Researchers at The Pennsylvania State University also investigated the use of artificial neural networks in predicting amino acid levels in feed ingredients. Artificial neural networks are biological-inspired tools that can serve as an alternative to regression analysis for complex data. Based on crude protein and proximate analysis of ingredients, two types of artificial neural networks and linear regression were evaluated for predicting amino acid levels in common feed ingredients. The research indicated that a general regression neural network (GRNN) program using proximate analysis data provided better prediction of amino acid levels in feed ingredients compared to linear regression. The research indicated that neural networks appear to be a promising method for modeling the relationship between proximate analysis of an ingredient and amino acid composition. The neural networks can be incorporated into a computer or spreadsheet program. The bottom line is that new computer programs are being studied for their application to feed formulation. It may be possible in the future to do a more precise job of formulating feeds to meeting nutrient specifications, reducing nutrient excesses and minimizing ration costs.

Roush, WB. and TL. Cravener. 1997. Artificial neural network prediction of amino acid levels in feed ingredients. Poultry Sci. 76.721-727.