Catfish farming suffers off-flavor losses of $10M to $20M annually. KTAADN, INC., (KI) proposes investigating empirical correlations between "predisposing" environmental factors and off-flavors. In phase I, KI will analyze these factors using neural network (NN) predictors developed by KI under a US Air Force grant for predicting lightning strikes. Phase I will show the feasibility of a NN processor for rapid and accurate analyses of environmental effects on off-flavors. Review of off-flavor measurements from several research groups recommends the use in Phase I of weekly data from the Tishomingo National Fish Hatchery (TNFH). Phase II will focus on extending the Phase I results using those data found to be the best for predicting off-flavors. These data will train an improved off-flavors NN processor that will be quantitatively scored for off-flavor correlations. Phase III will produce a prototype commercial "Off-Flavors Predictor" that can be upgraded, as new data are collected, by fish farmers without special computer skills.
Anticipated Results:NN processors allow efficient research into correlations among environmental factors bearing on catfish off-flavors. When verified with data, it provides fish farmers, using "desk-top" computers, to optimize feeding and maintenance of fish ponds. It will improve off-flavors predictions based upon local situations. A successful "Off-Flavors Predictor" could lead to industrial savings up to $10M annually and more reliable product delivery schedules.
