Hybrid learning of fuzzy cognitive maps for sugarcane yield classification

Sugarcane is one of India's most important renewable commercial crops. The sugarcane cultivation and sugar industry plays a vital role towards socio-economic development in the rural areas by creating higher income and employment opportunities. Early detection and management of problems associated with sugarcane yield indicators enables the decision makers and planners to decide import or export policies. In this work, a hybrid approach using fuzzy cognitive map (FCM) learning algorithms for sugarcane yield classification is proposed, combining the key aspects of Data Driven Nonlinear Hebbian Learning (DDNHL) algorithm and Genetic Algorithm (GA) called FCM-DDNHL-GA. The FCM model developed for the proposed study includes various soil and climate parameters which influence the precision agriculture application of sugarcane yield prediction. The classification accuracies and inference capabilities of the hybrid learning algorithm of FCMs are analyzed and compared with some well-known machine learning algorithms for sugarcane yield monitoring application. Experimental results show the superiority of the hybrid learning approach by providing significantly higher classification accuracy.