The use of ANN in plant pathology-related research has increased over the past few years and has been shown to be superior to conventional modeling approaches in several cases. Batchelor et al. (2) used ANN to predict soybean rust epidemics and compared their results with those obtained using regression and simulation models for the same dataset. They found that their best ANN model resulted in higher coefficient of multiple determination than the regression and imulation models, even when validated on an independent data set. De Wolf and Francl (16) reported superior performance of BPNN over logistic regression for the classification of incidence of tan spot of wheat, and over stepwise logistic regression and multivariate discriminant analysis for the detection of infection periods for the same disease (15). Similarly, Yang and Batchelor (68) reported that BPNN's performed better than conventional modeling approaches in predicting wheat scab epidemics, while Francl and Panigrahi (19) showed the superiority of the same class
of ANN's over discriminant analysis in predicting the wetness status of wheat leaves.