| Personal credit is an important component of China's consumer credit market today,to stimulate domestic demand and promote economic play a key role,but the high non-performing loans of financial institutions has become a personal credit loan shackles,an effective personal credit evaluation model is deeply concerned about the community and look forward to.However,in view of the personal credit rating multivariate model,high noise,highly nonlinear problems,making identification difficult to accurately predict.Neural network as a nonlinear dynamical system with highly nonlinear global role,is robust and powerful adaptive self-learning ability.In theory,the neural network can approximate any nonlinear system accuracy has unique advantages in solving such problems on personal credit evaluation.Error back propagation neural network(Error Back-Propagation,BP)is based on the error gradient algorithm reverse propagation neural network model correction layers connected weights,is the most essential part of the artificial neural network applications and research.This article will use BP neural network to individual credit assessments in the field,and personal credit record data preprocessing,and then construct evaluation index system of personal credit lending process to classify the breach,to get a higher recognition accuracy.BP neural network for the slow convergence,easy to fall into local minima inherent defects,Fruit Fly Optimization Algorithm optimized the initial connection weights and thresholds of BP neural network.Optimized BP neural network proven its recognition rate network performance and breach of personal credit loans has increased,there are certain practical value.Meanwhile Fruit Fly Optimization Algorithm is improved,the introduction of speed of particle swarm optimization to update the way to improve the ability of global optimization,the introduction of mutation operator of genetic algorithm to prevent premature convergence,compared to the effect of particle swarm optimization algorithm,confirmed strengthen the ability of global optimization,make the algorithm more practical application. |
