Artificial Neural Networks (ANN) a Novel Data Analysis Technique for Modeling and Forewarning Diseases in Plant Pathology

摘要

Artificial Neural Networks model have been used in plant pathology to study dynamics, modeling and forecasting of several disease in recent years. Artificial neural networks are parallel computing systems made up of a large number of simple, highly interconnected processing elements called nodes or neurons that process information by their dynamic-state response to the external signals and can handle imprecise information. It provides a flexible way to connect disease outcome with environmental and other determinant variables. ANN can automatically approximate any nonlinear mathematical function. This aspect of neural networks is particularly useful when the relationship between the variables is not known or is complex and hence it is difficult to handle statistically. Due to the nature of linear relationship in the parameters, regression models may not provide accurate predictions in some complex situations such as non linear data and extreme values data. As regression models need to fulfill the regression assumptions and multiple co-linearity between independent and dependent variables, it causes regression models to be inefficient. The ANN model has non-linear pattern recognition capability which is valuable for modeling and forecasting complexnon-linear problems in practice. Neural networks are similar to regression models, in that both develop coefficients that model patterns by evaluation of the relationship between independent and dependent variables. However, neural networks don’t reqi^ire hypothetical information for modeling, unlike parametric statistical models. In general, neural networks consist of three or more groups of elements that represent sets of equations used by the model. These recently developed Artificial Neural Network (ANNs) techniques are interesting and have become the focus of much attention, largely because of their wide range of applicability and the ease with which they can treat complicated problems even if the data are imprecise and noisy. As disease can occur over the conditions provided by the host plants as well as when weather conditions are favourable. Therefore, there is a need to develop effective forewarning model, which can provide advance information for outbreak of the disease. In this paper an attempt is made to reveal the potential use of ANN in disease prediction.