Predicting insurance for sustainable development using gradient methods based neural networks

This paper compares the development of neural network based prediction models for sustainable insurance using gradient techniques for training of the network. Convergence of different gradient algorithms is compared on data sets taken from life insurance in the rural sector. For applying these gradient based algorithms, prediction models based on neural networks are simulated in MATLAB Neural Network Toolbox. Method of supervised learning is adopted for learning of multilayer feed forward perceptron networks along with these gradient based algorithms of error reduction under consideration.